FilteredModel.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MF reloaded 1\n",
      "EMS reloaded\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from statsmodels.api import Logit\n",
    "import patsy\n",
    "from joblib import load, dump\n",
    "\n",
    "import model_functions as mf\n",
    "import eval_measures as ems\n",
    "\n",
    "from IPython.display import display"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "MODEL_VERSION = \"v4\"\n",
    "MODEL_SUFFIX = \"%s.first_author\" % MODEL_VERSION\n",
    "GENDERS = [\"-\", \"F\", \"M\"]\n",
    "TOP_15_ETHNICITIES = [\"ENGLISH\", \"GERMAN\", \"HISPANIC\", \"CHINESE\",\n",
    "                      \"JAPANESE\", \"SLAV\", \"FRENCH\", \"ITALIAN\", \"INDIAN\",\n",
    "                      \"NORDIC\", \"ARAB\", \"DUTCH\", \"KOREAN\", \"UNKNOWN\", \"OTHER\"]\n",
    "TOP_15_COUNTRIES = [\"USA\", \"UNKNOWN\", \"UK\", \"JAPAN\", \"GERMANY\", \"FRANCE\", \"ITALY\",\n",
    "                    \"CANADA\", \"CHINA\", \"AUSTRALIA\", \"SPAIN\", \"NETHERLANDS\",\n",
    "                    \"SWEDEN\", \"INDIA\", \"OTHER\"]\n",
    "\n",
    "sns.set_context(\"poster\")\n",
    "sns.set_style(\"ticks\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def filtered_data(df, max_source_ncites=100, version=1):\n",
    "    # Base version 1\n",
    "    df_filtered = df[(df.source_country == \"USA\")\n",
    "                & (df.source_is_eng)\n",
    "                & (df.source_is_journal)\n",
    "                & (df.source_ncites >= 10)\n",
    "                & (df.source_ncites <=max_source_ncites)]\n",
    "    if version == 2:\n",
    "        df_filtered = df_filtered[(df.eth1 == \"ENGLISH\")\n",
    "                & (df.eth2 == \"UNKNOWN\")]\n",
    "    print df_filtered.shape, df.shape\n",
    "    return df_filtered\n",
    "\n",
    "\n",
    "def fit_model(df, base_formula, include_prc=True, MODEL_VERSION=\"v4\", MODEL_SUFFIX=\"\", version=1):\n",
    "    results_full = []\n",
    "    nrows, ncols = 1,1\n",
    "    figsize = 5\n",
    "    if include_prc:\n",
    "        fig, ax = plt.subplots(nrows, ncols, sharex=True, sharey=True,\n",
    "                               squeeze=False, figsize=(ncols*figsize, nrows*figsize))\n",
    "        fig1, ax1 = plt.subplots(nrows, ncols, sharex=True, sharey=True,\n",
    "                                 squeeze=False, figsize=(ncols*figsize, nrows*figsize))\n",
    "        print \"Created figure with %s rows and %s columns\" % (nrows, ncols)\n",
    "        ax = ax.flatten()\n",
    "        ax1 = ax1.flatten()\n",
    "\n",
    "    formula = base_formula\n",
    "    i = 0\n",
    "    measures, summary = model_fit(df, formula, include_prc=include_prc)\n",
    "\n",
    "    prc = measures.pop(\"prc\", None)\n",
    "    roc = measures.pop(\"roc\", None)\n",
    "    if include_prc and prc is not None and roc is not None:\n",
    "        ## PRC\n",
    "        plot_prc(prc, ax=ax[i], color=\"k\")\n",
    "        ax[i].plot([measures[\"recall\"]], [measures[\"precision\"]], marker=\"o\",\n",
    "                   color=\"k\", linestyle=\"none\", ms=15)\n",
    "        ax[i].plot([0,1], [1,0], \"--k\", linewidth=1)\n",
    "        ax[i].set_xlabel(\"Recall\")\n",
    "        ax[i].set_ylabel(\"Precision\")\n",
    "        ## ROC\n",
    "        plot_prc(roc, ax=ax1[i], color=\"k\")\n",
    "        ax1[i].plot([measures[\"fpr\"]], [measures[\"tpr\"]], marker=\"o\",\n",
    "                   color=\"k\", linestyle=\"none\", ms=15)\n",
    "        ax1[i].plot([0,1], [0,1], \"--k\", linewidth=1)\n",
    "        ax1[i].set_xlabel(\"False Positive Rate\")\n",
    "        ax1[i].set_ylabel(\"True Positive Rate\")\n",
    "    print measures\n",
    "    print display(summary)\n",
    "    results_full.append((\"all\", (measures, summary)))\n",
    "\n",
    "    if include_prc:\n",
    "        fig.tight_layout()\n",
    "        fig.savefig(\"PRC_model.%s.filtered_%s.pdf\" % (MODEL_SUFFIX, version), bbox_inches=\"tight\")\n",
    "        fig1.tight_layout()\n",
    "        fig1.savefig(\"ROC_model.%s.filtered_%s.pdf\" % (MODEL_SUFFIX, version), bbox_inches=\"tight\")\n",
    "        \n",
    "    dump([results_full],\n",
    "         \"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" % (MODEL_VERSION, MODEL_SUFFIX, version))\n",
    "    return [results_full]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "with pd.HDFStore('out/ModelData.20160529.h5') as cstore:\n",
    "    df = cstore['first_author']\n",
    "    \n",
    "with pd.HDFStore('out/ModelData.20160529.h5') as cstore:\n",
    "    df_la = cstore['last_author']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>source_id</th>\n",
       "      <th>source_year</th>\n",
       "      <th>source_n_mesh</th>\n",
       "      <th>source_n_mesh_ex</th>\n",
       "      <th>source_is_eng</th>\n",
       "      <th>source_country</th>\n",
       "      <th>source_is_journal</th>\n",
       "      <th>source_is_review</th>\n",
       "      <th>source_is_case_rep</th>\n",
       "      <th>source_is_let_ed_com</th>\n",
       "      <th>...</th>\n",
       "      <th>eth2</th>\n",
       "      <th>pos</th>\n",
       "      <th>pos_nice</th>\n",
       "      <th>sink_last_ncites</th>\n",
       "      <th>sink_prev_ncites</th>\n",
       "      <th>auth_last_npapers</th>\n",
       "      <th>auth_prev_papers</th>\n",
       "      <th>jj_sim</th>\n",
       "      <th>is_self_cite</th>\n",
       "      <th>eth_weight</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>16526454</td>\n",
       "      <td>2005</td>\n",
       "      <td>11</td>\n",
       "      <td>53</td>\n",
       "      <td>True</td>\n",
       "      <td>JAPAN</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>UNKNOWN</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>20</td>\n",
       "      <td>14.6053</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>16526454</td>\n",
       "      <td>2005</td>\n",
       "      <td>11</td>\n",
       "      <td>53</td>\n",
       "      <td>True</td>\n",
       "      <td>JAPAN</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>UNKNOWN</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>20</td>\n",
       "      <td>14.6053</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>14595765</td>\n",
       "      <td>2003</td>\n",
       "      <td>17</td>\n",
       "      <td>71</td>\n",
       "      <td>True</td>\n",
       "      <td>JAPAN</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>UNKNOWN</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>14595765</td>\n",
       "      <td>2003</td>\n",
       "      <td>17</td>\n",
       "      <td>71</td>\n",
       "      <td>True</td>\n",
       "      <td>JAPAN</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>UNKNOWN</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>5.5691</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>14595765</td>\n",
       "      <td>2003</td>\n",
       "      <td>17</td>\n",
       "      <td>71</td>\n",
       "      <td>True</td>\n",
       "      <td>JAPAN</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>UNKNOWN</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>22</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>5.5691</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 54 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   source_id  source_year  source_n_mesh  source_n_mesh_ex source_is_eng  \\\n",
       "0   16526454         2005             11                53          True   \n",
       "1   16526454         2005             11                53          True   \n",
       "2   14595765         2003             17                71          True   \n",
       "3   14595765         2003             17                71          True   \n",
       "4   14595765         2003             17                71          True   \n",
       "\n",
       "  source_country source_is_journal source_is_review source_is_case_rep  \\\n",
       "0          JAPAN              True            False              False   \n",
       "1          JAPAN              True            False              False   \n",
       "2          JAPAN              True            False              False   \n",
       "3          JAPAN              True            False              False   \n",
       "4          JAPAN              True            False              False   \n",
       "\n",
       "  source_is_let_ed_com     ...         eth2  pos  pos_nice  sink_last_ncites  \\\n",
       "0                False     ...      UNKNOWN    1         1                 2   \n",
       "1                False     ...      UNKNOWN    1         1                 1   \n",
       "2                False     ...      UNKNOWN    1         1                 5   \n",
       "3                False     ...      UNKNOWN    1         1                 5   \n",
       "4                False     ...      UNKNOWN    1         1                 9   \n",
       "\n",
       "   sink_prev_ncites  auth_last_npapers  auth_prev_papers   jj_sim  \\\n",
       "0                 1                  1                20  14.6053   \n",
       "1                 2                  1                20  14.6053   \n",
       "2                 8                  1                 7   0.0000   \n",
       "3                17                  1                 7   5.5691   \n",
       "4                22                  1                 7   5.5691   \n",
       "\n",
       "   is_self_cite  eth_weight  \n",
       "0             0         1.0  \n",
       "1             1         1.0  \n",
       "2             0         1.0  \n",
       "3             0         1.0  \n",
       "4             0         1.0  \n",
       "\n",
       "[5 rows x 54 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index([u'source_id', u'source_year', u'source_n_mesh', u'source_n_mesh_ex',\n",
       "       u'source_is_eng', u'source_country', u'source_is_journal',\n",
       "       u'source_is_review', u'source_is_case_rep', u'source_is_let_ed_com',\n",
       "       u'source_T_novelty', u'source_V_novelty', u'source_PT_novelty',\n",
       "       u'source_PV_novelty', u'source_ncites', u'source_n_authors', u'sink_id',\n",
       "       u'sink_year', u'sink_n_mesh', u'sink_n_mesh_ex', u'sink_is_eng',\n",
       "       u'sink_is_journal', u'sink_is_review', u'sink_is_case_rep',\n",
       "       u'sink_is_let_ed_com', u'sink_T_novelty', u'sink_V_novelty',\n",
       "       u'sink_PT_novelty', u'sink_PV_novelty', u'sink_n_authors', u'year_span',\n",
       "       u'journal_same', u'mesh_sim', u'title_sim', u'lang_sim',\n",
       "       u'affiliation_sim', u'pubtype_sim', u'cite_sim', u'author_sim',\n",
       "       u'gender_sim', u'eth_sim', u'n_common_authors', u'gender', u'eth1',\n",
       "       u'eth2', u'pos', u'pos_nice', u'sink_last_ncites', u'sink_prev_ncites',\n",
       "       u'auth_last_npapers', u'auth_prev_papers', u'jj_sim', u'is_self_cite',\n",
       "       u'eth_weight'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_prc(prc, ax, color=\"k\", label=\"PRC\"):\n",
    "    precision, recall = prc\n",
    "    ax.plot(recall, precision, \n",
    "            marker=\"None\", linestyle=\"-\",\n",
    "            color=color, label=label)\n",
    "\n",
    "def get_all_eval_measures(predict, endog, include_prc=False):\n",
    "    measures = {}\n",
    "    pred_table = ems.cm(predict, endog)\n",
    "    measures[\"precision\"] = ems.precision(pred_table)\n",
    "    measures[\"recall\"] = ems.recall(pred_table)\n",
    "    measures[\"accuracy\"] = ems.accuracy(pred_table)\n",
    "    measures[\"f_score\"] = ems.fscore_measure(pred_table)\n",
    "    measures[\"rmse\"] = ems.rmse(predict, endog)\n",
    "    measures[\"mae\"] = ems.mae(predict, endog)\n",
    "    measures[\"auc\"] = ems.auc(predict, endog)\n",
    "    tn, fp, fn, tp = map(float, pred_table.flatten()) # WRT to 1 as positive label\n",
    "    measures[\"tn\"] = tn\n",
    "    measures[\"fn\"] = fn\n",
    "    measures[\"fp\"] = fp\n",
    "    measures[\"tp\"] = tp\n",
    "    measures[\"tpr\"] = tp * 1. / (tp + fn) \n",
    "    measures[\"fpr\"] = fp * 1. / (fp + tn) \n",
    "    print \"In eval measures function.\"\n",
    "    if include_prc:\n",
    "        ## Include the precision recall values\n",
    "        prc = ems.prc(predict, endog, float_precision=3)\n",
    "        measures[\"prc\"] = prc\n",
    "        roc = ems.roc(predict, endog, float_precision=3)\n",
    "        measures[\"roc\"] = roc\n",
    "    return measures\n",
    "\n",
    "\n",
    "def model_fit(df, formula, testing=False, include_prc=False):\n",
    "    if testing:\n",
    "        # If testing the just print X and y columns\n",
    "        print df.shape, formula, include_prc\n",
    "        return ({\"llf\": 0.1}, \"TEMP SUMMARY\")\n",
    "    ## Not testing. Fit the models and return the measures\n",
    "    print df.shape, formula, include_prc\n",
    "    y,X = patsy.dmatrices(formula, data=df, return_type=\"dataframe\")\n",
    "    print \"Created dataframes\"\n",
    "    print \"X.shape = %s, y.shape = %s\" % (X.shape, y.shape)\n",
    "    model = Logit(y,X)\n",
    "    res = model.fit()\n",
    "    predict = res.predict()\n",
    "    measures = get_all_eval_measures(predict, model.endog, include_prc=include_prc)\n",
    "    measures[\"llf\"] = res.llf\n",
    "    measures[\"aic\"] = res.aic\n",
    "    measures[\"bic\"] = res.bic\n",
    "    measures[\"prsquared\"] = res.prsquared\n",
    "    measures[\"df_model\"] = res.df_model\n",
    "    return (measures, res.summary2())\n",
    "\n",
    "\n",
    "def print_model(results_filtered, fp):\n",
    "    columns = [\"llf\", \"aic\", \"bic\", \"prsquared\", \"df_model\",\n",
    "               \"precision\", \"recall\", \"f_score\", \"auc\", \"mae\",\n",
    "               \"rmse\", \"tp\", \"fp\", \"tn\", \"fn\", \"tpr\", \"fpr\"]\n",
    "    measures = pd.DataFrame(columns=columns)\n",
    "    for j,(m,s) in results_filtered[0]:\n",
    "        print >> fp, \"%s\\t%s\\t%s\" % (\"*\"*50, j, \"*\"*50)\n",
    "        measures.ix[j,columns] = [m[k] for k in columns]\n",
    "        print >> fp, s\n",
    "    print >> fp, \"%s\\t%s\\t%s\" % (\"*\"*50, \"All Measures\", \"*\"*50)\n",
    "    print >> fp, measures.to_string()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Fit models version 1\n",
    "\n",
    "**Filters applied**\n",
    "* pub type = journal article\n",
    "* affiliation = USA\n",
    "* language = English\n",
    "* \\# source references = [10,x], where x represents the second kink in the fit vs. empirical plots (perhaps 100 for first authors and ~ 40 or 50 for last authors)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "base_formula = (\"is_self_cite ~ I(auth_prev_papers == 0) + I(auth_prev_papers == 1) \"\n",
    "           \"+ np.log10(auth_prev_papers + 1) + I(np.log10(auth_prev_papers + 1)**2)\"\n",
    "           \"+ C(gender, levels=GENDERS)\"\n",
    "           \"+ mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)\"\n",
    "           \"+ np.log10(source_ncites) \"\n",
    "           \"+ I(np.log10(source_ncites)**2) + I(np.log10(source_ncites)**3)\"\n",
    "           \"+ I(source_n_authors > 20) + np.log10(np.clip(source_n_authors, 0, 20))\"\n",
    "           \"+ I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)\"\n",
    "           \"+ np.log10(source_n_mesh_ex + 1) + I(source_n_mesh_ex == 0)\" \n",
    "           \"+ np.log10(sink_n_mesh_ex + 1) + I(sink_n_mesh_ex == 0)\"\n",
    "           \"+ I(year_span < 0) + I(year_span == 0) + mf.score_log_1(year_span) + I(mf.score_log_1(year_span)**2)\"\n",
    "           \"+ I(sink_prev_ncites == 0) + np.log10(sink_prev_ncites + 1) + I(np.log10(sink_prev_ncites + 1)**2)\"\n",
    "           \"+ I(jj_sim == 0) + np.log10(jj_sim + 1) + I(np.log10(jj_sim + 1)**2) + journal_same\"\n",
    "           \"+ sink_is_eng + sink_is_journal + sink_is_review + sink_is_case_rep + sink_is_let_ed_com\"\n",
    "           \"+ np.log10(np.nan_to_num(source_V_novelty) + 1)\"\n",
    "           \"+ np.log10(np.nan_to_num(sink_V_novelty) + 1) + I(np.log10(np.nan_to_num(sink_V_novelty) + 1)**2)\"\n",
    "          )\n",
    "\n",
    "version = 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### First author"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(13046025, 54) (41618369, 54)\n"
     ]
    }
   ],
   "source": [
    "df_filtered = filtered_data(df, max_source_ncites=100, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Created figure with 1 rows and 1 columns\n",
      "(13046025, 54) is_self_cite ~ I(auth_prev_papers == 0) + I(auth_prev_papers == 1) + np.log10(auth_prev_papers + 1) + I(np.log10(auth_prev_papers + 1)**2)+ C(gender, levels=GENDERS)+ mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)+ np.log10(source_ncites) + I(np.log10(source_ncites)**2) + I(np.log10(source_ncites)**3)+ I(source_n_authors > 20) + np.log10(np.clip(source_n_authors, 0, 20))+ I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)+ np.log10(source_n_mesh_ex + 1) + I(source_n_mesh_ex == 0)+ np.log10(sink_n_mesh_ex + 1) + I(sink_n_mesh_ex == 0)+ I(year_span < 0) + I(year_span == 0) + mf.score_log_1(year_span) + I(mf.score_log_1(year_span)**2)+ I(sink_prev_ncites == 0) + np.log10(sink_prev_ncites + 1) + I(np.log10(sink_prev_ncites + 1)**2)+ I(jj_sim == 0) + np.log10(jj_sim + 1) + I(np.log10(jj_sim + 1)**2) + journal_same+ sink_is_eng + sink_is_journal + sink_is_review + sink_is_case_rep + sink_is_let_ed_com+ np.log10(np.nan_to_num(source_V_novelty) + 1)+ np.log10(np.nan_to_num(sink_V_novelty) + 1) + I(np.log10(np.nan_to_num(sink_V_novelty) + 1)**2) True\n",
      "Using class based MultiVal\n",
      "using complimentary weights for 2 columns. w and 1-w\n",
      "Using class based MultiVal\n",
      "using complimentary weights for 2 columns. w and 1-w\n",
      "Created dataframes\n",
      "X.shape = (13046025, 50), y.shape = (13046025, 1)\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.156567\n",
      "         Iterations 10\n",
      "In eval measures function.\n",
      "{'fp': 23221.0, 'auc': 0.84903504374953287, 'rmse': 0.20644421066231908, 'prsquared': 0.20208540994182556, 'recall': 0.023606192829846676, 'precision': 0.3949713392391871, 'f_score': 0.04454978531817287, 'bic': 4085977.6349808867, 'tn': 12380642.0, 'df_model': 49.0, 'aic': 4085258.4352785051, 'mae': 0.083943202870148309, 'tp': 15159.0, 'fpr': 0.001872078077611789, 'tpr': 0.023606192829846676, 'llf': -2042579.2176392525, 'fn': 627003.0, 'accuracy': 0.9501592247447019}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "        <td>Model:</td>              <td>Logit</td>      <td>Pseudo R-squared:</td>     <td>0.202</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <td>Dependent Variable:</td>   <td>is_self_cite</td>         <td>AIC:</td>        <td>4085258.4353</td>\n",
       "</tr>\n",
       "<tr>\n",
       "         <td>Date:</td>        <td>2016-07-26 11:51</td>       <td>BIC:</td>        <td>4085977.6350</td>\n",
       "</tr>\n",
       "<tr>\n",
       "   <td>No. Observations:</td>      <td>13046025</td>      <td>Log-Likelihood:</td>   <td>-2.0426e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "       <td>Df Model:</td>             <td>49</td>            <td>LL-Null:</td>       <td>-2.5599e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "     <td>Df Residuals:</td>        <td>13045975</td>       <td>LLR p-value:</td>       <td>0.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "      <td>Converged:</td>           <td>1.0000</td>           <td>Scale:</td>          <td>1.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "    <td>No. Iterations:</td>        <td>10.0000</td>             <td></td>                <td></td>      \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "                                    <td></td>                                    <th>Coef.</th>  <th>Std.Err.</th>     <th>z</th>      <th>P>|z|</th> <th>[0.025</th>  <th>0.975]</th> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>                                                            <td>-4.0473</td>  <td>0.2895</td>  <td>-13.9810</td>  <td>0.0000</td> <td>-4.6147</td> <td>-3.4799</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 0)[T.True]</th>                                     <td>-1.2318</td>  <td>0.0179</td>  <td>-68.9285</td>  <td>0.0000</td> <td>-1.2669</td> <td>-1.1968</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 1)[T.True]</th>                                     <td>-0.1594</td>  <td>0.0099</td>  <td>-16.1068</td>  <td>0.0000</td> <td>-0.1788</td> <td>-0.1400</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.F]</th>                                       <td>-0.0518</td>  <td>0.0057</td>   <td>-9.0221</td>  <td>0.0000</td> <td>-0.0630</td> <td>-0.0405</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.M]</th>                                       <td>-0.0297</td>  <td>0.0053</td>   <td>-5.6235</td>  <td>0.0000</td> <td>-0.0401</td> <td>-0.0194</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_authors > 20)[T.True]</th>                                     <td>-0.0369</td>  <td>0.0246</td>   <td>-1.4975</td>  <td>0.1343</td> <td>-0.0851</td> <td>0.0114</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_mesh_ex == 0)[T.True]</th>                                     <td>-0.5111</td>  <td>0.0241</td>  <td>-21.2304</td>  <td>0.0000</td> <td>-0.5583</td> <td>-0.4639</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_n_mesh_ex == 0)[T.True]</th>                                       <td>0.0193</td>   <td>0.0285</td>   <td>0.6760</td>   <td>0.4991</td> <td>-0.0366</td> <td>0.0751</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span < 0)[T.True]</th>                                             <td>-0.4453</td>  <td>0.0284</td>  <td>-15.6950</td>  <td>0.0000</td> <td>-0.5009</td> <td>-0.3897</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span == 0)[T.True]</th>                                            <td>0.4459</td>   <td>0.0099</td>   <td>44.8231</td>  <td>0.0000</td> <td>0.4264</td>  <td>0.4654</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_prev_ncites == 0)[T.True]</th>                                     <td>0.1662</td>   <td>0.0071</td>   <td>23.3037</td>  <td>0.0000</td> <td>0.1522</td>  <td>0.1802</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(jj_sim == 0)[T.True]</th>                                               <td>-0.1717</td>  <td>0.0148</td>  <td>-11.5941</td>  <td>0.0000</td> <td>-0.2008</td> <td>-0.1427</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>journal_same[T.True]</th>                                                 <td>0.5091</td>   <td>0.0049</td>  <td>104.3436</td>  <td>0.0000</td> <td>0.4995</td>  <td>0.5187</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_eng[T.True]</th>                                                  <td>2.0747</td>   <td>0.0844</td>   <td>24.5715</td>  <td>0.0000</td> <td>1.9092</td>  <td>2.2402</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_journal[T.True]</th>                                              <td>0.5478</td>   <td>0.0217</td>   <td>25.1857</td>  <td>0.0000</td> <td>0.5051</td>  <td>0.5904</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_review[T.True]</th>                                               <td>-0.7406</td>  <td>0.0051</td>  <td>-145.1799</td> <td>0.0000</td> <td>-0.7506</td> <td>-0.7306</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_case_rep[T.True]</th>                                             <td>-0.8818</td>  <td>0.0119</td>  <td>-73.8284</td>  <td>0.0000</td> <td>-0.9052</td> <td>-0.8584</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_let_ed_com[T.True]</th>                                           <td>-0.4206</td>  <td>0.0213</td>  <td>-19.7131</td>  <td>0.0000</td> <td>-0.4624</td> <td>-0.3788</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(auth_prev_papers + 1)</th>                                       <td>1.8621</td>   <td>0.0132</td>  <td>141.5131</td>  <td>0.0000</td> <td>1.8363</td>  <td>1.8879</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(auth_prev_papers + 1) ** 2)</th>                               <td>-0.1271</td>  <td>0.0044</td>  <td>-29.1441</td>  <td>0.0000</td> <td>-0.1356</td> <td>-0.1185</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[0]</th>  <td>-0.0334</td>  <td>0.0060</td>   <td>-5.5429</td>  <td>0.0000</td> <td>-0.0453</td> <td>-0.0216</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[1]</th>  <td>-0.0379</td>  <td>0.0071</td>   <td>-5.3229</td>  <td>0.0000</td> <td>-0.0518</td> <td>-0.0239</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[2]</th>  <td>0.1802</td>   <td>0.0060</td>   <td>30.1552</td>  <td>0.0000</td> <td>0.1685</td>  <td>0.1919</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[3]</th>  <td>-0.1468</td>  <td>0.0089</td>  <td>-16.4600</td>  <td>0.0000</td> <td>-0.1642</td> <td>-0.1293</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[4]</th>  <td>0.1163</td>   <td>0.0078</td>   <td>14.9781</td>  <td>0.0000</td> <td>0.1011</td>  <td>0.1316</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[5]</th>  <td>-0.0663</td>  <td>0.0082</td>   <td>-8.1183</td>  <td>0.0000</td> <td>-0.0823</td> <td>-0.0503</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[6]</th>  <td>-0.0922</td>  <td>0.0101</td>   <td>-9.1132</td>  <td>0.0000</td> <td>-0.1120</td> <td>-0.0724</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[7]</th>  <td>0.0135</td>   <td>0.0067</td>   <td>2.0229</td>   <td>0.0431</td> <td>0.0004</td>  <td>0.0265</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[8]</th>  <td>0.0302</td>   <td>0.0116</td>   <td>2.6159</td>   <td>0.0089</td> <td>0.0076</td>  <td>0.0529</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[9]</th>  <td>-0.0276</td>  <td>0.0093</td>   <td>-2.9673</td>  <td>0.0030</td> <td>-0.0458</td> <td>-0.0094</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[10]</th> <td>0.0062</td>   <td>0.0140</td>   <td>0.4434</td>   <td>0.6575</td> <td>-0.0212</td> <td>0.0336</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[11]</th> <td>-0.0054</td>  <td>0.0122</td>   <td>-0.4458</td>  <td>0.6557</td> <td>-0.0293</td> <td>0.0185</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[12]</th> <td>-0.0620</td>  <td>0.0257</td>   <td>-2.4117</td>  <td>0.0159</td> <td>-0.1125</td> <td>-0.0116</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[13]</th> <td>-0.1099</td>  <td>0.0077</td>  <td>-14.2527</td>  <td>0.0000</td> <td>-0.1250</td> <td>-0.0948</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_ncites)</th>                                              <td>-3.2230</td>  <td>0.5624</td>   <td>-5.7303</td>  <td>0.0000</td> <td>-4.3253</td> <td>-2.1206</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 2)</th>                                      <td>2.5789</td>   <td>0.3788</td>   <td>6.8072</td>   <td>0.0000</td> <td>1.8363</td>  <td>3.3214</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 3)</th>                                      <td>-0.7712</td>  <td>0.0838</td>   <td>-9.2033</td>  <td>0.0000</td> <td>-0.9355</td> <td>-0.6070</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.clip(source_n_authors, 0, 20))</th>                           <td>0.2416</td>   <td>0.0302</td>   <td>7.9893</td>   <td>0.0000</td> <td>0.1824</td>  <td>0.3009</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)</th>                   <td>-0.3357</td>  <td>0.0225</td>  <td>-14.9013</td>  <td>0.0000</td> <td>-0.3798</td> <td>-0.2915</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_n_mesh_ex + 1)</th>                                       <td>-0.1069</td>  <td>0.0093</td>  <td>-11.4371</td>  <td>0.0000</td> <td>-0.1253</td> <td>-0.0886</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_n_mesh_ex + 1)</th>                                         <td>-0.1431</td>  <td>0.0098</td>  <td>-14.6666</td>  <td>0.0000</td> <td>-0.1622</td> <td>-0.1239</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.score_log_1(year_span)</th>                                            <td>-0.2041</td>  <td>0.0262</td>   <td>-7.7946</td>  <td>0.0000</td> <td>-0.2554</td> <td>-0.1528</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(mf.score_log_1(year_span) ** 2)</th>                                    <td>-0.7667</td>  <td>0.0162</td>  <td>-47.2011</td>  <td>0.0000</td> <td>-0.7985</td> <td>-0.7348</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_prev_ncites + 1)</th>                                       <td>-0.1470</td>  <td>0.0113</td>  <td>-13.0397</td>  <td>0.0000</td> <td>-0.1691</td> <td>-0.1249</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(sink_prev_ncites + 1) ** 2)</th>                               <td>-0.2262</td>  <td>0.0042</td>  <td>-53.9181</td>  <td>0.0000</td> <td>-0.2345</td> <td>-0.2180</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(jj_sim + 1)</th>                                                 <td>0.2049</td>   <td>0.0186</td>   <td>11.0076</td>  <td>0.0000</td> <td>0.1684</td>  <td>0.2414</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(jj_sim + 1) ** 2)</th>                                         <td>-0.0441</td>  <td>0.0056</td>   <td>-7.8858</td>  <td>0.0000</td> <td>-0.0551</td> <td>-0.0332</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(source_V_novelty) + 1)</th>                        <td>-0.0962</td>  <td>0.0024</td>  <td>-40.5615</td>  <td>0.0000</td> <td>-0.1008</td> <td>-0.0915</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(sink_V_novelty) + 1)</th>                          <td>0.1995</td>   <td>0.0135</td>   <td>14.7545</td>  <td>0.0000</td> <td>0.1730</td>  <td>0.2261</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)</th>                  <td>-0.0444</td>  <td>0.0023</td>  <td>-19.1661</td>  <td>0.0000</td> <td>-0.0489</td> <td>-0.0399</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary2.Summary'>\n",
       "\"\"\"\n",
       "                                                    Results: Logit\n",
       "======================================================================================================================\n",
       "Model:                               Logit                             Pseudo R-squared:                  0.202       \n",
       "Dependent Variable:                  is_self_cite                      AIC:                               4085258.4353\n",
       "Date:                                2016-07-26 11:51                  BIC:                               4085977.6350\n",
       "No. Observations:                    13046025                          Log-Likelihood:                    -2.0426e+06 \n",
       "Df Model:                            49                                LL-Null:                           -2.5599e+06 \n",
       "Df Residuals:                        13045975                          LLR p-value:                       0.0000      \n",
       "Converged:                           1.0000                            Scale:                             1.0000      \n",
       "No. Iterations:                      10.0000                                                                          \n",
       "----------------------------------------------------------------------------------------------------------------------\n",
       "                                                                      Coef.  Std.Err.     z     P>|z|   [0.025  0.975]\n",
       "----------------------------------------------------------------------------------------------------------------------\n",
       "Intercept                                                            -4.0473   0.2895  -13.9810 0.0000 -4.6147 -3.4799\n",
       "I(auth_prev_papers == 0)[T.True]                                     -1.2318   0.0179  -68.9285 0.0000 -1.2669 -1.1968\n",
       "I(auth_prev_papers == 1)[T.True]                                     -0.1594   0.0099  -16.1068 0.0000 -0.1788 -0.1400\n",
       "C(gender, levels=GENDERS)[T.F]                                       -0.0518   0.0057   -9.0221 0.0000 -0.0630 -0.0405\n",
       "C(gender, levels=GENDERS)[T.M]                                       -0.0297   0.0053   -5.6235 0.0000 -0.0401 -0.0194\n",
       "I(source_n_authors > 20)[T.True]                                     -0.0369   0.0246   -1.4975 0.1343 -0.0851  0.0114\n",
       "I(source_n_mesh_ex == 0)[T.True]                                     -0.5111   0.0241  -21.2304 0.0000 -0.5583 -0.4639\n",
       "I(sink_n_mesh_ex == 0)[T.True]                                        0.0193   0.0285    0.6760 0.4991 -0.0366  0.0751\n",
       "I(year_span < 0)[T.True]                                             -0.4453   0.0284  -15.6950 0.0000 -0.5009 -0.3897\n",
       "I(year_span == 0)[T.True]                                             0.4459   0.0099   44.8231 0.0000  0.4264  0.4654\n",
       "I(sink_prev_ncites == 0)[T.True]                                      0.1662   0.0071   23.3037 0.0000  0.1522  0.1802\n",
       "I(jj_sim == 0)[T.True]                                               -0.1717   0.0148  -11.5941 0.0000 -0.2008 -0.1427\n",
       "journal_same[T.True]                                                  0.5091   0.0049  104.3436 0.0000  0.4995  0.5187\n",
       "sink_is_eng[T.True]                                                   2.0747   0.0844   24.5715 0.0000  1.9092  2.2402\n",
       "sink_is_journal[T.True]                                               0.5478   0.0217   25.1857 0.0000  0.5051  0.5904\n",
       "sink_is_review[T.True]                                               -0.7406   0.0051 -145.1799 0.0000 -0.7506 -0.7306\n",
       "sink_is_case_rep[T.True]                                             -0.8818   0.0119  -73.8284 0.0000 -0.9052 -0.8584\n",
       "sink_is_let_ed_com[T.True]                                           -0.4206   0.0213  -19.7131 0.0000 -0.4624 -0.3788\n",
       "np.log10(auth_prev_papers + 1)                                        1.8621   0.0132  141.5131 0.0000  1.8363  1.8879\n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                               -0.1271   0.0044  -29.1441 0.0000 -0.1356 -0.1185\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[0]  -0.0334   0.0060   -5.5429 0.0000 -0.0453 -0.0216\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[1]  -0.0379   0.0071   -5.3229 0.0000 -0.0518 -0.0239\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[2]   0.1802   0.0060   30.1552 0.0000  0.1685  0.1919\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[3]  -0.1468   0.0089  -16.4600 0.0000 -0.1642 -0.1293\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[4]   0.1163   0.0078   14.9781 0.0000  0.1011  0.1316\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[5]  -0.0663   0.0082   -8.1183 0.0000 -0.0823 -0.0503\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[6]  -0.0922   0.0101   -9.1132 0.0000 -0.1120 -0.0724\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[7]   0.0135   0.0067    2.0229 0.0431  0.0004  0.0265\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[8]   0.0302   0.0116    2.6159 0.0089  0.0076  0.0529\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[9]  -0.0276   0.0093   -2.9673 0.0030 -0.0458 -0.0094\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[10]  0.0062   0.0140    0.4434 0.6575 -0.0212  0.0336\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[11] -0.0054   0.0122   -0.4458 0.6557 -0.0293  0.0185\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[12] -0.0620   0.0257   -2.4117 0.0159 -0.1125 -0.0116\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[13] -0.1099   0.0077  -14.2527 0.0000 -0.1250 -0.0948\n",
       "np.log10(source_ncites)                                              -3.2230   0.5624   -5.7303 0.0000 -4.3253 -2.1206\n",
       "I(np.log10(source_ncites) ** 2)                                       2.5789   0.3788    6.8072 0.0000  1.8363  3.3214\n",
       "I(np.log10(source_ncites) ** 3)                                      -0.7712   0.0838   -9.2033 0.0000 -0.9355 -0.6070\n",
       "np.log10(np.clip(source_n_authors, 0, 20))                            0.2416   0.0302    7.9893 0.0000  0.1824  0.3009\n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)                   -0.3357   0.0225  -14.9013 0.0000 -0.3798 -0.2915\n",
       "np.log10(source_n_mesh_ex + 1)                                       -0.1069   0.0093  -11.4371 0.0000 -0.1253 -0.0886\n",
       "np.log10(sink_n_mesh_ex + 1)                                         -0.1431   0.0098  -14.6666 0.0000 -0.1622 -0.1239\n",
       "mf.score_log_1(year_span)                                            -0.2041   0.0262   -7.7946 0.0000 -0.2554 -0.1528\n",
       "I(mf.score_log_1(year_span) ** 2)                                    -0.7667   0.0162  -47.2011 0.0000 -0.7985 -0.7348\n",
       "np.log10(sink_prev_ncites + 1)                                       -0.1470   0.0113  -13.0397 0.0000 -0.1691 -0.1249\n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                               -0.2262   0.0042  -53.9181 0.0000 -0.2345 -0.2180\n",
       "np.log10(jj_sim + 1)                                                  0.2049   0.0186   11.0076 0.0000  0.1684  0.2414\n",
       "I(np.log10(jj_sim + 1) ** 2)                                         -0.0441   0.0056   -7.8858 0.0000 -0.0551 -0.0332\n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)                        -0.0962   0.0024  -40.5615 0.0000 -0.1008 -0.0915\n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)                           0.1995   0.0135   14.7545 0.0000  0.1730  0.2261\n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)                  -0.0444   0.0023  -19.1661 0.0000 -0.0489 -0.0399\n",
       "======================================================================================================================\n",
       "\n",
       "\"\"\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmIQQIiEKTEIIkdD/A511p11Q+mW4AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9aa0db52d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f976e3ca850>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "MODEL_VERSION = \"v4\"\n",
    "MODEL_SUFFIX = \"%s.first_author\" % MODEL_VERSION\n",
    "\n",
    "results_full = fit_model(df_filtered, base_formula, include_prc=True,\n",
    "                            MODEL_VERSION=MODEL_VERSION, MODEL_SUFFIX=MODEL_SUFFIX, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "results_filtered = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" \n",
    "                        % (MODEL_VERSION, MODEL_SUFFIX, version))\n",
    "with open(\"Filtered_models.%s.filtered_%s.txt\" % (MODEL_SUFFIX, version), \"wb+\") as fp: \n",
    "    print_model(results_filtered, fp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Last author"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(8988820, 54) (41618369, 54)\n"
     ]
    }
   ],
   "source": [
    "df_filtered_la = filtered_data(df_la, max_source_ncites=50, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Created figure with 1 rows and 1 columns\n",
      "(8988820, 54) is_self_cite ~ I(auth_prev_papers == 0) + I(auth_prev_papers == 1) + np.log10(auth_prev_papers + 1) + I(np.log10(auth_prev_papers + 1)**2)+ C(gender, levels=GENDERS)+ mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)+ np.log10(source_ncites) + I(np.log10(source_ncites)**2) + I(np.log10(source_ncites)**3)+ I(source_n_authors > 20) + np.log10(np.clip(source_n_authors, 0, 20))+ I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)+ np.log10(source_n_mesh_ex + 1) + I(source_n_mesh_ex == 0)+ np.log10(sink_n_mesh_ex + 1) + I(sink_n_mesh_ex == 0)+ I(year_span < 0) + I(year_span == 0) + mf.score_log_1(year_span) + I(mf.score_log_1(year_span)**2)+ I(sink_prev_ncites == 0) + np.log10(sink_prev_ncites + 1) + I(np.log10(sink_prev_ncites + 1)**2)+ I(jj_sim == 0) + np.log10(jj_sim + 1) + I(np.log10(jj_sim + 1)**2) + journal_same+ sink_is_eng + sink_is_journal + sink_is_review + sink_is_case_rep + sink_is_let_ed_com+ np.log10(np.nan_to_num(source_V_novelty) + 1)+ np.log10(np.nan_to_num(sink_V_novelty) + 1) + I(np.log10(np.nan_to_num(sink_V_novelty) + 1)**2) True\n",
      "Using class based MultiVal\n",
      "using complimentary weights for 2 columns. w and 1-w\n",
      "Using class based MultiVal\n",
      "using complimentary weights for 2 columns. w and 1-w\n",
      "Created dataframes\n",
      "X.shape = (8988820, 50), y.shape = (8988820, 1)\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.303245\n",
      "         Iterations 10\n",
      "In eval measures function.\n",
      "{'fp': 9565.0, 'auc': 0.74200244838358387, 'rmse': 0.29850695794030124, 'prsquared': 0.11235991803906931, 'recall': 0.008948330905371478, 'precision': 0.4752002633600351, 'f_score': 0.017565884743782678, 'bic': 5452436.775788228, 'tn': 8011365.0, 'df_model': 49.0, 'aic': 5451736.2011811836, 'mae': 0.17755997544964644, 'tp': 8661.0, 'fpr': 0.001192505108509861, 'tpr': 0.008948330905371478, 'llf': -2725818.1005905918, 'fn': 959229.0, 'accuracy': 0.8922223384159433}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "        <td>Model:</td>              <td>Logit</td>      <td>Pseudo R-squared:</td>     <td>0.112</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <td>Dependent Variable:</td>   <td>is_self_cite</td>         <td>AIC:</td>        <td>5451736.2012</td>\n",
       "</tr>\n",
       "<tr>\n",
       "         <td>Date:</td>        <td>2016-07-26 12:01</td>       <td>BIC:</td>        <td>5452436.7758</td>\n",
       "</tr>\n",
       "<tr>\n",
       "   <td>No. Observations:</td>       <td>8988820</td>      <td>Log-Likelihood:</td>   <td>-2.7258e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "       <td>Df Model:</td>             <td>49</td>            <td>LL-Null:</td>       <td>-3.0709e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "     <td>Df Residuals:</td>         <td>8988770</td>       <td>LLR p-value:</td>       <td>0.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "      <td>Converged:</td>           <td>1.0000</td>           <td>Scale:</td>          <td>1.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "    <td>No. Iterations:</td>        <td>10.0000</td>             <td></td>                <td></td>      \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "                                    <td></td>                                     <th>Coef.</th>  <th>Std.Err.</th>     <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>   <th>0.975]</th> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>                                                             <td>0.8550</td>   <td>0.4878</td>   <td>1.7528</td>   <td>0.0796</td>  <td>-0.1011</td>  <td>1.8112</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 0)[T.True]</th>                                      <td>-1.2593</td>  <td>0.0388</td>  <td>-32.4420</td>  <td>0.0000</td>  <td>-1.3354</td>  <td>-1.1832</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 1)[T.True]</th>                                      <td>-0.2089</td>  <td>0.0242</td>   <td>-8.6456</td>  <td>0.0000</td>  <td>-0.2562</td>  <td>-0.1615</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.F]</th>                                        <td>0.0004</td>   <td>0.0059</td>   <td>0.0643</td>   <td>0.9488</td>  <td>-0.0112</td>  <td>0.0120</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.M]</th>                                        <td>-0.0114</td>  <td>0.0054</td>   <td>-2.1289</td>  <td>0.0333</td>  <td>-0.0219</td>  <td>-0.0009</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_authors > 20)[T.True]</th>                                      <td>0.0232</td>   <td>0.0236</td>   <td>0.9827</td>   <td>0.3258</td>  <td>-0.0231</td>  <td>0.0696</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_mesh_ex == 0)[T.True]</th>                                      <td>-0.7059</td>  <td>0.0201</td>  <td>-35.0348</td>  <td>0.0000</td>  <td>-0.7453</td>  <td>-0.6664</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_n_mesh_ex == 0)[T.True]</th>                                        <td>-0.0930</td>  <td>0.0222</td>   <td>-4.1898</td>  <td>0.0000</td>  <td>-0.1364</td>  <td>-0.0495</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span < 0)[T.True]</th>                                              <td>-0.3821</td>  <td>0.0321</td>  <td>-11.9150</td>  <td>0.0000</td>  <td>-0.4449</td>  <td>-0.3192</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span == 0)[T.True]</th>                                             <td>0.7146</td>   <td>0.0092</td>   <td>77.9346</td>  <td>0.0000</td>  <td>0.6966</td>   <td>0.7326</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_prev_ncites == 0)[T.True]</th>                                      <td>0.2968</td>   <td>0.0064</td>   <td>46.6158</td>  <td>0.0000</td>  <td>0.2843</td>   <td>0.3092</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(jj_sim == 0)[T.True]</th>                                                <td>-0.1677</td>  <td>0.0126</td>  <td>-13.2876</td>  <td>0.0000</td>  <td>-0.1924</td>  <td>-0.1430</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>journal_same[T.True]</th>                                                  <td>0.6124</td>   <td>0.0040</td>  <td>152.3333</td>  <td>0.0000</td>  <td>0.6045</td>   <td>0.6203</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_eng[T.True]</th>                                                   <td>2.6562</td>   <td>0.1022</td>   <td>26.0004</td>  <td>0.0000</td>  <td>2.4560</td>   <td>2.8565</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_journal[T.True]</th>                                               <td>0.5370</td>   <td>0.0197</td>   <td>27.2198</td>  <td>0.0000</td>  <td>0.4983</td>   <td>0.5757</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_review[T.True]</th>                                                <td>-0.6616</td>  <td>0.0042</td>  <td>-158.7592</td> <td>0.0000</td>  <td>-0.6697</td>  <td>-0.6534</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_case_rep[T.True]</th>                                              <td>-1.0378</td>  <td>0.0107</td>  <td>-96.6286</td>  <td>0.0000</td>  <td>-1.0589</td>  <td>-1.0168</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_let_ed_com[T.True]</th>                                            <td>-0.3867</td>  <td>0.0193</td>  <td>-20.0859</td>  <td>0.0000</td>  <td>-0.4244</td>  <td>-0.3489</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(auth_prev_papers + 1)</th>                                        <td>2.4576</td>   <td>0.0161</td>  <td>152.6372</td>  <td>0.0000</td>  <td>2.4261</td>   <td>2.4892</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(auth_prev_papers + 1) ** 2)</th>                                <td>-0.4035</td>  <td>0.0045</td>  <td>-90.3873</td>  <td>0.0000</td>  <td>-0.4123</td>  <td>-0.3948</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[0]</th>   <td>0.0584</td>   <td>0.0047</td>   <td>12.5306</td>  <td>0.0000</td>  <td>0.0492</td>   <td>0.0675</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[1]</th>   <td>-0.0203</td>  <td>0.0067</td>   <td>-3.0183</td>  <td>0.0025</td>  <td>-0.0335</td>  <td>-0.0071</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[2]</th>   <td>0.1815</td>   <td>0.0062</td>   <td>29.2588</td>  <td>0.0000</td>  <td>0.1694</td>   <td>0.1937</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[3]</th>   <td>0.1517</td>   <td>0.0110</td>   <td>13.8463</td>  <td>0.0000</td>  <td>0.1302</td>   <td>0.1732</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[4]</th>   <td>0.1617</td>   <td>0.0084</td>   <td>19.3556</td>  <td>0.0000</td>  <td>0.1453</td>   <td>0.1781</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[5]</th>   <td>-0.0203</td>  <td>0.0068</td>   <td>-3.0101</td>  <td>0.0026</td>  <td>-0.0336</td>  <td>-0.0071</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[6]</th>   <td>0.0554</td>   <td>0.0084</td>   <td>6.6307</td>   <td>0.0000</td>  <td>0.0390</td>   <td>0.0718</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[7]</th>   <td>0.0745</td>   <td>0.0065</td>   <td>11.4532</td>  <td>0.0000</td>  <td>0.0617</td>   <td>0.0872</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[8]</th>   <td>0.0938</td>   <td>0.0092</td>   <td>10.1998</td>  <td>0.0000</td>  <td>0.0758</td>   <td>0.1118</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[9]</th>   <td>-0.0356</td>  <td>0.0092</td>   <td>-3.8793</td>  <td>0.0001</td>  <td>-0.0536</td>  <td>-0.0176</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[10]</th>  <td>-0.0296</td>  <td>0.0108</td>   <td>-2.7285</td>  <td>0.0064</td>  <td>-0.0508</td>  <td>-0.0083</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[11]</th>  <td>0.0974</td>   <td>0.0164</td>   <td>5.9443</td>   <td>0.0000</td>  <td>0.0653</td>   <td>0.1296</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[12]</th>  <td>-0.1005</td>  <td>0.0225</td>   <td>-4.4661</td>  <td>0.0000</td>  <td>-0.1446</td>  <td>-0.0564</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[13]</th>  <td>0.0638</td>   <td>0.0067</td>   <td>9.5826</td>   <td>0.0000</td>  <td>0.0508</td>   <td>0.0769</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_ncites)</th>                                              <td>-17.7558</td>  <td>1.0554</td>  <td>-16.8238</td>  <td>0.0000</td> <td>-19.8243</td> <td>-15.6872</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 2)</th>                                       <td>13.5447</td>  <td>0.7718</td>   <td>17.5491</td>  <td>0.0000</td>  <td>12.0320</td>  <td>15.0574</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 3)</th>                                       <td>-3.4377</td>  <td>0.1862</td>  <td>-18.4636</td>  <td>0.0000</td>  <td>-3.8026</td>  <td>-3.0728</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.clip(source_n_authors, 0, 20))</th>                            <td>-0.1366</td>  <td>0.0259</td>   <td>-5.2808</td>  <td>0.0000</td>  <td>-0.1872</td>  <td>-0.0859</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)</th>                    <td>-0.1128</td>  <td>0.0192</td>   <td>-5.8795</td>  <td>0.0000</td>  <td>-0.1504</td>  <td>-0.0752</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_n_mesh_ex + 1)</th>                                        <td>-0.1537</td>  <td>0.0079</td>  <td>-19.4048</td>  <td>0.0000</td>  <td>-0.1692</td>  <td>-0.1382</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_n_mesh_ex + 1)</th>                                          <td>-0.1789</td>  <td>0.0079</td>  <td>-22.5491</td>  <td>0.0000</td>  <td>-0.1945</td>  <td>-0.1634</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.score_log_1(year_span)</th>                                             <td>0.8259</td>   <td>0.0217</td>   <td>38.0900</td>  <td>0.0000</td>  <td>0.7834</td>   <td>0.8684</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(mf.score_log_1(year_span) ** 2)</th>                                     <td>-1.1601</td>  <td>0.0128</td>  <td>-90.3568</td>  <td>0.0000</td>  <td>-1.1852</td>  <td>-1.1349</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_prev_ncites + 1)</th>                                        <td>0.1991</td>   <td>0.0086</td>   <td>23.1865</td>  <td>0.0000</td>  <td>0.1822</td>   <td>0.2159</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(sink_prev_ncites + 1) ** 2)</th>                                <td>-0.2088</td>  <td>0.0029</td>  <td>-72.8039</td>  <td>0.0000</td>  <td>-0.2144</td>  <td>-0.2032</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(jj_sim + 1)</th>                                                  <td>0.3594</td>   <td>0.0163</td>   <td>22.0233</td>  <td>0.0000</td>  <td>0.3274</td>   <td>0.3914</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(jj_sim + 1) ** 2)</th>                                          <td>-0.1456</td>  <td>0.0050</td>  <td>-29.0227</td>  <td>0.0000</td>  <td>-0.1555</td>  <td>-0.1358</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(source_V_novelty) + 1)</th>                         <td>-0.1345</td>  <td>0.0019</td>  <td>-69.5541</td>  <td>0.0000</td>  <td>-0.1383</td>  <td>-0.1308</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(sink_V_novelty) + 1)</th>                           <td>0.1570</td>   <td>0.0103</td>   <td>15.2612</td>  <td>0.0000</td>  <td>0.1368</td>   <td>0.1771</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)</th>                   <td>-0.0415</td>  <td>0.0018</td>  <td>-23.1135</td>  <td>0.0000</td>  <td>-0.0450</td>  <td>-0.0380</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary2.Summary'>\n",
       "\"\"\"\n",
       "                                                     Results: Logit\n",
       "=========================================================================================================================\n",
       "Model:                                Logit                              Pseudo R-squared:                   0.112       \n",
       "Dependent Variable:                   is_self_cite                       AIC:                                5451736.2012\n",
       "Date:                                 2016-07-26 12:01                   BIC:                                5452436.7758\n",
       "No. Observations:                     8988820                            Log-Likelihood:                     -2.7258e+06 \n",
       "Df Model:                             49                                 LL-Null:                            -3.0709e+06 \n",
       "Df Residuals:                         8988770                            LLR p-value:                        0.0000      \n",
       "Converged:                            1.0000                             Scale:                              1.0000      \n",
       "No. Iterations:                       10.0000                                                                            \n",
       "-------------------------------------------------------------------------------------------------------------------------\n",
       "                                                                      Coef.   Std.Err.     z     P>|z|   [0.025   0.975] \n",
       "-------------------------------------------------------------------------------------------------------------------------\n",
       "Intercept                                                              0.8550   0.4878    1.7528 0.0796  -0.1011   1.8112\n",
       "I(auth_prev_papers == 0)[T.True]                                      -1.2593   0.0388  -32.4420 0.0000  -1.3354  -1.1832\n",
       "I(auth_prev_papers == 1)[T.True]                                      -0.2089   0.0242   -8.6456 0.0000  -0.2562  -0.1615\n",
       "C(gender, levels=GENDERS)[T.F]                                         0.0004   0.0059    0.0643 0.9488  -0.0112   0.0120\n",
       "C(gender, levels=GENDERS)[T.M]                                        -0.0114   0.0054   -2.1289 0.0333  -0.0219  -0.0009\n",
       "I(source_n_authors > 20)[T.True]                                       0.0232   0.0236    0.9827 0.3258  -0.0231   0.0696\n",
       "I(source_n_mesh_ex == 0)[T.True]                                      -0.7059   0.0201  -35.0348 0.0000  -0.7453  -0.6664\n",
       "I(sink_n_mesh_ex == 0)[T.True]                                        -0.0930   0.0222   -4.1898 0.0000  -0.1364  -0.0495\n",
       "I(year_span < 0)[T.True]                                              -0.3821   0.0321  -11.9150 0.0000  -0.4449  -0.3192\n",
       "I(year_span == 0)[T.True]                                              0.7146   0.0092   77.9346 0.0000   0.6966   0.7326\n",
       "I(sink_prev_ncites == 0)[T.True]                                       0.2968   0.0064   46.6158 0.0000   0.2843   0.3092\n",
       "I(jj_sim == 0)[T.True]                                                -0.1677   0.0126  -13.2876 0.0000  -0.1924  -0.1430\n",
       "journal_same[T.True]                                                   0.6124   0.0040  152.3333 0.0000   0.6045   0.6203\n",
       "sink_is_eng[T.True]                                                    2.6562   0.1022   26.0004 0.0000   2.4560   2.8565\n",
       "sink_is_journal[T.True]                                                0.5370   0.0197   27.2198 0.0000   0.4983   0.5757\n",
       "sink_is_review[T.True]                                                -0.6616   0.0042 -158.7592 0.0000  -0.6697  -0.6534\n",
       "sink_is_case_rep[T.True]                                              -1.0378   0.0107  -96.6286 0.0000  -1.0589  -1.0168\n",
       "sink_is_let_ed_com[T.True]                                            -0.3867   0.0193  -20.0859 0.0000  -0.4244  -0.3489\n",
       "np.log10(auth_prev_papers + 1)                                         2.4576   0.0161  152.6372 0.0000   2.4261   2.4892\n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                                -0.4035   0.0045  -90.3873 0.0000  -0.4123  -0.3948\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[0]    0.0584   0.0047   12.5306 0.0000   0.0492   0.0675\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[1]   -0.0203   0.0067   -3.0183 0.0025  -0.0335  -0.0071\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[2]    0.1815   0.0062   29.2588 0.0000   0.1694   0.1937\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[3]    0.1517   0.0110   13.8463 0.0000   0.1302   0.1732\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[4]    0.1617   0.0084   19.3556 0.0000   0.1453   0.1781\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[5]   -0.0203   0.0068   -3.0101 0.0026  -0.0336  -0.0071\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[6]    0.0554   0.0084    6.6307 0.0000   0.0390   0.0718\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[7]    0.0745   0.0065   11.4532 0.0000   0.0617   0.0872\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[8]    0.0938   0.0092   10.1998 0.0000   0.0758   0.1118\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[9]   -0.0356   0.0092   -3.8793 0.0001  -0.0536  -0.0176\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[10]  -0.0296   0.0108   -2.7285 0.0064  -0.0508  -0.0083\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[11]   0.0974   0.0164    5.9443 0.0000   0.0653   0.1296\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[12]  -0.1005   0.0225   -4.4661 0.0000  -0.1446  -0.0564\n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[13]   0.0638   0.0067    9.5826 0.0000   0.0508   0.0769\n",
       "np.log10(source_ncites)                                              -17.7558   1.0554  -16.8238 0.0000 -19.8243 -15.6872\n",
       "I(np.log10(source_ncites) ** 2)                                       13.5447   0.7718   17.5491 0.0000  12.0320  15.0574\n",
       "I(np.log10(source_ncites) ** 3)                                       -3.4377   0.1862  -18.4636 0.0000  -3.8026  -3.0728\n",
       "np.log10(np.clip(source_n_authors, 0, 20))                            -0.1366   0.0259   -5.2808 0.0000  -0.1872  -0.0859\n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)                    -0.1128   0.0192   -5.8795 0.0000  -0.1504  -0.0752\n",
       "np.log10(source_n_mesh_ex + 1)                                        -0.1537   0.0079  -19.4048 0.0000  -0.1692  -0.1382\n",
       "np.log10(sink_n_mesh_ex + 1)                                          -0.1789   0.0079  -22.5491 0.0000  -0.1945  -0.1634\n",
       "mf.score_log_1(year_span)                                              0.8259   0.0217   38.0900 0.0000   0.7834   0.8684\n",
       "I(mf.score_log_1(year_span) ** 2)                                     -1.1601   0.0128  -90.3568 0.0000  -1.1852  -1.1349\n",
       "np.log10(sink_prev_ncites + 1)                                         0.1991   0.0086   23.1865 0.0000   0.1822   0.2159\n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                                -0.2088   0.0029  -72.8039 0.0000  -0.2144  -0.2032\n",
       "np.log10(jj_sim + 1)                                                   0.3594   0.0163   22.0233 0.0000   0.3274   0.3914\n",
       "I(np.log10(jj_sim + 1) ** 2)                                          -0.1456   0.0050  -29.0227 0.0000  -0.1555  -0.1358\n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)                         -0.1345   0.0019  -69.5541 0.0000  -0.1383  -0.1308\n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)                            0.1570   0.0103   15.2612 0.0000   0.1368   0.1771\n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)                   -0.0415   0.0018  -23.1135 0.0000  -0.0450  -0.0380\n",
       "=========================================================================================================================\n",
       "\n",
       "\"\"\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noxE9eFfeqc0AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f976a398190>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3XlYjfn/P/DnaaFIn8hOaTBkSRFtY20TWZJlhGKylSLL\nGGuGIdnKUtGEJilkREYZosggyjQTozCYVKQiSyWt9++Pvt0/R4tTnXPuc06vx3W5rs59zn3fz3um\n8+pe3guPYRgGhBBCvkiO6wCEECItqGASQoiAqGASQoiAqGASQoiAqGASQoiAqGASQoiAOC2YDMNg\n37596NOnD3x9fb/4+Xv37sHe3h66urowMjLCxo0b8fHjRzEkJYQQDgvmmzdvMG/ePERFRUFeXv6L\nn8/NzYWjoyO6du2K8PBw7NmzBzdv3oS7u7tQc9na2sLW1lao2xQ3OgbJQMcgOYR1HJwVzN9++w2K\niooIDw+HnNyXYxw9ehTNmjXD5s2b0bNnTxgZGWHVqlWIjIxEZmamGBITQpo6zgqmubk5/P39oaKi\nItDnb926BQMDAygoKLDLTExMAADx8fEiyUgIIZ/irGB26dKlXp9/9uxZtXWUlZWhrq6OtLQ0ISYj\nhJCaSc1T8sLCQigrK1db3qJFCxQUFHCQiBDS1Ch8+SOyq6abwP/++y86d+6MZ8+eCfQwShKVlJQA\ngFTf26VjkAyycAwAUFxcjGfPnsHGxqbaM5PTp08LvB2pKZitWrWq8UwyPz8frVq1Etp+OnfujLS0\nNFhaWgptm1wxMzPjOkKj0TFIBlk4BgAoKipCy5YtG7y+1BTMbt26ISMjg2/Z+/fv8ebNG/To0aNB\n26zpL8uzZ89gaWmJ0NBQdOzYsUHbJURWlJaW4vHjx/j333/x+PFj9l9aWhpKS0uFvr9mzZpBWVkZ\nysrKUFRUhKKiIuTl5aGgoABFRUUoKCiw/3g8HrtebT+/f/8ejx8/xqBBg/D8+XMEBASgW7duDc4n\nNQVz2LBhCA4ORklJCZo1awYAuHr1KuTl5TF06FCh7afqMrxjx47o2rWr0LZLiKQrKytDSkoK7ty5\nw/5LTk5mL8sFpaSkBE1NTXTu3Blt27aFurp6tX+tW7eGiooKWrZsyf5r0aIFXyuYxkpISMC4ceNw\n6tQpaGtrw8zMrNG32TgrmO/evUNpaSmqxi/+8OEDXr16BQBo06YNdu/ejZSUFBw+fBgAMHPmTISG\nhmLt2rVYvHgxXr58CS8vL0yfPh3t2rXj6jAIkVqlpaVISEjAlStXEBsbi1u3bqGoqOiL68nLy6NH\njx7o06cPunfvDk1NTWhqaqJbt27Q1NRE27Zt+c7yuJCcnIzx48cjMDAQ5ubmQrsHy1nBdHV1xZ07\nd9jXv/zyCwIDA8Hj8RATE4NXr17xHaSamhqCgoKwZcsWTJw4ESoqKpgwYQKWL1/ORXxCpA7DMEhO\nTkZ0dDRiY2Nx/fp1FBYW1rnOV199BX19fejo6KBv377o06cPevbsiebNm4spdf2lpqbCysoKvr6+\nGDdunFC3zaMpKvhlZmbCzMwMMTExdElOpN7Hjx9x5coVnDt3DpGRkdWeA3yqU6dOMDIywuDBgzF4\n8GDo6+tDXV1djGmF45dffoGCggLs7e3ZZcL6XkvNPUxCiGBKSkpw8eJFHD9+HL/99lutZ5Ft27bF\nqFGjYGpqilGjRqFXr16cX0oLw3fffSeybVPBJEQGlJeXIy4uDsePH0d4eDjevHlT7TMKCgoYNmwY\nxo0bBwsLC/Tr10+gcRzI/0cFkxAplpqail9++QUhISHIysqq9r6qqirGjx+P8ePHY/To0VBTU+Mg\npeyggkmIlHn//j1OnjyJwMDAGgeeUVJSwrhx42BnZ4exY8dCSUmJg5Ti8ebNGzx//hz9+/cXy/6o\nYBIiJZ4+fYp9+/bh8OHD1Xq9ycvLw8LCAjNmzMDEiROhqqrKUUrxyc/Px5gxY2Bubo4tW7aIZZ9U\nMAmRYAzD4MaNG/D29sbZs2dRUVHB937v3r3h6OgIe3t7dOrUiaOU4vfhwweMGzcOenp62Lx5s9j2\nSwWTEAlUXl6OU6dOwcvLC4mJiXzvNWvWDHZ2dliwYAGMjY1l4sl2fRQXF2PSpEno1q0b9u/fL9bj\np4JJiAQpKyvD8ePHsWXLFjx69IjvvXbt2mHRokVwdnZGhw4dOErILYZhMH36dPzvf/9DYGCg2J/y\nU8EkRAKUlpYiNDQUHh4eePz4Md97/fr1w7JlyzBz5kyZfoAjCB6PB2dnZ4wcOVKo/c4FRQWTEA6V\nlZUhODgYHh4eePr0Kd97RkZG2LBhA6ysrJrcZXdduBx6kQomIRxgGAbnzp3DmjVrkJKSwvfe0KFD\n8eOPP8LMzIwKpYShgkmImN28eROrVq3C9evX+ZaPGDECP/74I0aOHEmFUkJRvyhCxCQzMxPTp0/H\nN998w1cs9fX1cfnyZVy9ehWjRo2iYvkJT09P/Pnnn1zHYFHBJETEiouLsXXrVvTu3RthYWHs8u7d\nu+PEiRNISEiQmSkghMnT0xNHjx6FhoYG11FYdElOiAhFRUXBzc0NT548YZepqalh06ZNcHJyYmcP\nIPz27t2Lw4cP49q1a2jfvj3XcVhUMAkRgcePH2Pp0qWIiopil/F4PMybNw8eHh40S0AdAgIC4O3t\njWvXrqFz585cx+FDBZMQISopKYGnpye2bt3KNxeOkZERfHx8MHjwYA7TSb60tDR4eHggJiamUZOV\niQoVTEKEJCkpCd999x3u3r3LLuvQoQO2b98Oe3t7GntSAFpaWkhJSWnUVLiiRP8HCWmk4uJirF+/\nHgYGBmyxlJOTw7Jly/Dw4UPMnj2bimU9SGqxBOgMk5BGSUhIgKOjI+7fv88u69u3LwIDA2FoaMhh\nMiIK9GePkAb4+PEjVq1aBWNjY7ZYysvLY+3atUhKSqJiKaCPHz9yHaFeqGASUk/x8fHQ09PDjh07\n2PEpdXR0cPv2bXh4eEj0FLSSJDExEf369as2GLIko4JJiIDKy8vx008/YejQoXj48CGAyonFfvzx\nR9y5cwf6+vocJ5Qed+/exbhx47Bnzx6oqKhwHUdgdA+TEAFkZGRg1qxZuHbtGrts4MCBCAwMhJ6e\nHofJpM+DBw9gZWUFHx8fjB8/nus49UJnmIR8wenTp6Grq8sWSx6Ph7Vr1+L27dtULOvpyZMnsLCw\ngKenJ6ZNm8Z1nHqjM0xCalFUVIRly5bh559/Zpd17twZISEhGDVqFIfJpNf9+/exbt06zJ49m+so\nDUIFk5AapKWlwdbWFn/99Re7bMKECTh8+DDatm3LYTLpNmHCBK4jNApdkhPymejoaOjr67PFsnnz\n5vDz80NERAQVyyaOCiYh/6eiogJbt26FlZUV8vLyAFR21YuPj8eiRYtonEpCBZMQAHj//j0mT56M\ndevWgWEYAMDo0aNx584dDBw4kON00ik/Px8JCQlcxxAqKpikyUtJScGQIUMQERHBLlu/fj2ioqKg\nrq7OYTLp9eHDB4wfPx4hISFcRxEqeuhDmrRTp05hzpw5KCwsBACoqqoiODgYEydO5DiZ9CouLsak\nSZOgqamJPXv2cB1HqOgMkzRJDMNg06ZNmDp1Klss+/Xrh8TERCqWjVBaWopvv/0WqqqqCAwMlLlR\nmugMkzQ5xcXFmDdvHt/l4rRp03D48GGp6qYniebOnYvy8nKcPHkSCgqyV15k74gIqUNeXh4mTZrE\n18Vxy5YtWLt2LT0FF4IFCxZg8ODBMjtXERVM0mQ8fvwY1tbWePToEYDK9pVHjhzBt99+y3Ey2TF0\n6FCuI4gUFUzSJNy4cQMTJ07E69evAQBt27bF2bNnYWJiwnEyIk1k644sITU4fvw4TE1N2WLZu3dv\n3Lp1i4olqTfOC2ZQUBDMzc2ho6MDa2trvmlJaxIfH4+ZM2diyJAh0NfXh4uLC549eyamtESaMAwD\nT09PzJgxg53BccSIEbh58yZ69OjBcTrp5+XlhdjYWK5jiBWnBTM0NBTe3t5wdXXFuXPnMG3aNKxc\nuRI3btyo8fP//PMP5s+fjwEDBuDXX39FSEgICgoK8N1336GoqEjM6Ykkq6iowIoVK7B27Vp2mYOD\nA6Kjo9GmTRsOk8mGffv24cCBA9DW1uY6ingxHBo+fDjj6enJt8zFxYWZNWtWjZ/fvn07Y2hoyLcs\nNTWV6d27N3Pt2jWhZMrIyGB69erFZGRkCGV7RPxKSkoYe3t7BgD7b+PGjUxFRQXX0WRCQEAAo6mp\nyaSlpXEdRWDC+l5zdob55MkTZGdnV7uPZGJigqSkJPYS6lM8Hq9aQ1hFRUX2PUKKiopga2uLo0eP\nAqj8vdi/fz9+/PFH+h0RgpCQEGzatAkxMTHo1q0b13HEjrOCmZ6eDh6Phy5duvAt19DQQEVFBTIy\nMqqtY2tri6KiIgQGBqK4uBhFRUXYv38/vvrqKxgZGYkrOpFQhYWFsLa2RmRkJIDKP6YnTpyAs7Mz\nx8lkQ25uLjZs2IDo6Gj07NmT6zic4KxZUVV3tBYtWvAtr3pd00xyPXr0gJ+fH5YsWYJdu3YBAL76\n6iscOnRIJnsVEMG9f/8e1tbWuH79OoDK36PTp09j9OjRHCeTHe3atUNKSgqUlJS4jsIZzp+S18ej\nR4+wfPly2Nra4uTJkzhy5Ag6d+6MhQsXsgWYND1v376FpaUlWyxbtWqFixcvUrEUgaZcLAEOzzBb\ntWoFoPqZZNXrmvr0+vr6QkNDg+/JZ58+fTB06FCEh4fDwcGhXhlsbW2rLavp3imRXK9fv4alpSWS\nkpIAAP/73/9w8eJFGBoacpyMSCInJ6dq3TZPnz4t8PqcnWF269YNDMMgPT2db3laWhoUFBSgqalZ\nbZ2nT5+ie/fufMtUVFTQpk0baovZBOXk5GDUqFFssWzTpg1iY2OpWArJhw8fuI4gcTg7w9TS0oKG\nhgb++OMPmJmZscvj4uJgbGzMPv3+VIcOHZCWlsa3LD8/H7m5uejQoUO9M9T0lyUzM5MvD5FML1++\nhKmpKVJTUwFU3l+LiYmBjo4Ox8lkw927dzFu3Dj8+eefaNeuHddxhMbf3x9du3Zt8Pqc3sN0cXFB\neHg4IiIi8OLFCwQEBCAhIQEuLi4AKnsSzJ07l/28vb097t69iz179uDp06dITU3F6tWroaCggDFj\nxnB1GETMXr9+DQsLC7ZYdurUCXFxcVQsheTBgwewsrLCzp07ZapYCgOnj5ZtbGxQVFQEX19f5OTk\nQEtLC35+ftDV1QUAvHr1CpmZmeznR44cCR8fH/j5+SEwMBCKiooYMGAAjhw5Ag0NDa4Og4jRu3fv\nMHr0aPzzzz8AKucJv3r1Kr7++muOk8mGJ0+ewMLCAp6enjSKU02E045edlBPH8mVn5/PmJiYsL13\n2rVrx6SmpnIdS2akp6czWlpazP79+7mOInSc9vR59+4dysrKhFu5CalDUVERJk6ciJs3bwIAWrdu\njcuXLze9vswilJaWhqVLl1JD/zoIXDATExPh4OAAXV1dGBsb4/HjxwCAiIgIhIWFiSwgISUlJZgy\nZQo7Mk5VO8sBAwZwnEy2DBs2DG5ublzHkGgCFcybN29izpw5yM7OxtSpU9l5m4HKph0bN27E+fPn\nRRaSNF1lZWWYMWMG+/ulrKyMqKgoDBkyhONkpCkSqGD6+vpixIgR+P3337F+/Xq+grlgwQLMnDkT\nhw4dEllI0jRVVFTA0dER4eHhAIBmzZrh7NmzGDZsGMfJSFMlUMFMTU3FjBkzap0y09LSkr1EJ0QY\nGIbBokWL2FGHFBQUcOrUKVhYWHCcTDYUFRXhypUrXMeQOgI3K/r0rPJzRUVFNTY0J6QhGIbBihUr\n8PPPPwMA5OTkEBoaivHjx3OcTDYUFxfD1tYWHTt2xKhRo7iOI1UEOsPU1dVFYGBgjf2sCwoKsG/f\nPrbtJCGN5enpid27d7OvAwMDMW3aNA4TyY7S0lJMnz4dLVu2xMGDB7mOI3UEOsN0dXXFnDlzMHbs\nWIwYMQI8Hg9HjhxBWVkZrl69iuLiYgQHB4s6K2kCgoODsW7dOva1n58fZs+ezWEi2VFeXg4HBweU\nlJTgzJkzNCRiAwh0hjl48GAcO3YMWlpa+PXXX8EwDM6cOYPo6Gjo6ekhJCQEenp6os5KZFx0dDRf\nV9iNGzdi0aJFHCaSLW5ubsjNzUV4eHi1EXuIYAT+EzNgwAAcOnQIZWVlePv2LYDKxsPy8vJgGAZl\nZWX0F4s02F9//YXJkyezHSLmzZuHDRs2cJxKtjg6OqJ3795NfkzLxhDoDNPMzAz//vsvgMqnlW3b\ntkXbtm0hLy8PALhw4QKN8EMaLD09HdbW1uxYqNbW1jhw4ADNwSNkgwYNQsuWLbmOIdXqPCV88eIF\nAOD58+d48eJFjf+xy8vL8eeffyIvL080CYlMe/v2LcaOHYusrCwAlbd/wsLC6GqFSKQ6fyurzhp5\nPB6cnJxq/RzDMNTzgtRbSUkJJk+ejPv37wOoHCM1MjKSzoKIxKqzYN68eRN37tzB4sWLMW3aNLRv\n377Gz7Vv3x5jx44VSUAimxiGwfz589n+4a1bt8bvv//eoIGgSXV+fn7o2LEjJk+ezHUUmVJnwWzd\nujUsLCzg6uqKb7/9ttbBRPPz85Geno6+ffuKJCSRPZs2bWKbojVr1gwRERE08pCQHD58GDt27EBc\nXBzXUWSOQA99XF1d6xx5OTExkdrKEYEdOXIEmzZtYl8HBQVh+PDhHCaSHceOHcOGDRtw+fJlaGlp\ncR1H5gh8Z/3EiROIiorCy5cvUVFRwS6vqKhATk4ODWVPBHL9+nXMnz+ffb1161bY2dlxmEh2nDlz\nBitWrMDly5dpBHoREegM8+jRo9i4cSPS09PRvn17PH/+HGpqalBQUEBWVhasrKzg5+cn6qxEyqWl\npcHW1halpaUAgLlz52L16tUcp5INBQUFWL16Nc6fP49+/fpxHUdmCXSGGRYWhhkzZsDd3R08Hg/a\n2trw8PCAtrY24uLi4O3tjbZt24o6K5FiBQUFmDBhAnJzcwEAI0aMwP79+6mtpZCoqKjg3r171INH\nxAQ6w8zMzMTYsWP5frmrLstHjBiB8ePHw8PDQzQJidSrqKjArFmzcO/ePQDAV199hVOnTtGXW8jo\nv6foCVQwFRUVUVRUxL5WVVVlzxSAyh4Et2/fFn46IhPWr1+Ps2fPAqg8Ezp37hxdkRCpJPDgG/v2\n7WMbGHfv3h0nT55k3797965o0hGpd+zYMXh6egKo7ABx/PhxuscmBPn5+VxHaJIEKphubm5IT0+H\nj48PAGDKlCmIiYnB8OHDYWtrix07dtC0AaSau3fvYt68eezrHTt2YNy4cRwmkg0PHz5Enz598N9/\n/3EdpckR6KGPtrY2oqKikJGRAaCyYBYUFOC3335jJ6mi2ebIp969e4fJkyezt3JmzZqFFStWcJxK\n+j19+hTm5ubYsmULvvrqK67jNDkCt8OsGqGoypw5czBnzhxRZCJSjmEYzJkzh53nSUdHBz///DM9\nEW+kjIwMmJmZYc2aNfTd48gXL8nLy8vx+vXrL27o8uXLQglEpN+uXbsQEREBoPIBYXh4OFq0aMFx\nKun28uVLmJmZwdXVlQZV5lCdBXP//v0wMDDA0KFDoa+vj3379lX7zKNHjzBnzhwsXrxYZCGJ9IiL\ni+NrjH7kyBHqdSIEOTk5WLhwId3W4Fitl+RnzpzBvn37MHDgQOjp6eHhw4c4cOAAlJWVMX/+fLx5\n8wZ79+7FqVOnIC8vD0dHR3HmJhIoKysL3377LdtG94cffoCNjQ3HqWTDgAEDMGDAAK5jNHm1Fsyw\nsDBYWlrynVXu3bsXx44dg6KiIvbv34+CggKMHz8eS5cuRadOncQSmEim0tJSTJs2DdnZ2QAqOzRQ\nZwYia2q9JH/48CEmTZrEt2z69OnIysrC9u3boaOjg9OnT2P79u1ULAnWrFmD69evAwA6duyIEydO\n0KjpRObUWjCLioqqjUBUNbjr9u3bcfjwYRq/kAAATp06BS8vLwCAvLw8Tp48iY4dO3KcSnoVFxcj\nMjKS6xikBnWeAtTWDKRXr14iCUOkz8OHD/nuX1MnhsYpLS3F9OnTIS8vD2tra2qKJWHomok0WGFh\nISZPnsx205s8eTKWLVvGcSrpVV5ejtmzZ6OkpARnzpyhYimBGnSGSf8jCcMwWLhwITu+QK9evRAY\nGEi/Gw1UUVGBBQsWIDs7G5GRkTTykISqs2C6u7tXm8GPx+Nh3bp11Roi83g8HDlyRPgJiUQKCgpC\naGgoAKBFixYIDw+Hqqoqx6mkl7u7Ox48eICLFy9CWVmZ6zikFrUWzM6dO+PNmzd48+YN3/JOnToh\nLy+P5iFvwh49esTXUcHf3x/9+/fnMJH0c3BwwA8//AAVFRWuo5A61Fowq6Y/JeRTxcXFsLOzQ2Fh\nIQDA3t4e9vb2HKeSfr179+Y6AhGAQMO7EVJl3bp1SEpKAgD06NGD5nIiTQoVTCKwCxcusO0tFRQU\ncPz4cbRq1YrjVISID+cFMygoCObm5tDR0YG1tTWioqLq/HxBQQHc3d1haGiIQYMGYd68eew4nUR0\nsrOz+eae37JlC4YMGcJhIukVGBiIgIAArmOQBuC0YIaGhsLb2xuurq44d+4cpk2bhpUrV+LGjRu1\nruPs7Ixnz54hODgYx44dQ2FhIZydncWYuulhGAaOjo7IyckBAJiZmWHlypUcp5JOx48fh7u7O0aO\nHMl1FNIAnDZcDwgIwIwZM9gRbbS0tJCYmAh/f39888031T7/xx9/4J9//sGVK1egpqYGoHLsxfv3\n76OkpITaronIgQMHcP78eQCAuro6goODISfH+cWJ1Dlz5gyWLVuGy5cvU285KcXZb/2TJ0+QnZ0N\nExMTvuUmJiZISkpCSUlJtXWuXLkCQ0NDtlgCQJcuXWBpaUnFUkRSU1P5xmA8dOgQOnfuzGEi6XTh\nwgUsXLgQ58+fpyZYUkzggvnff/9hzZo1GDt2LAYPHowHDx4AqCxicXFx9d5xeno6eDweunTpwrdc\nQ0MDFRUVNd6XfPToEbS0tHDw4EGMHj0axsbGWL58ObUJFZGSkhLMnDkTHz9+BADMmzePxrdsgNLS\nUqxevRpnz57FoEGDuI5DGkGggnn//n1MnjwZsbGx0NLSYtvgAcCdO3ewaNEi3Lx5s147rtrG5z2G\nql4XFBRUWycvLw+///47Hj16BG9vb3h6eiI5ORkODg7soLVEeH788Uf89ddfAICePXti9+7dHCeS\nToqKirhz5w6MjY25jkIaSaB7mN7e3tDW1kZAQABUVFT4hnVbuXIlXr58if3791e7vBa2srIyKCkp\nYceOHWyf5ebNm+O7777D9evXMXz48Hptz9bWttqymm4FNEV37tzBjh07AFQO2RYSEkK9UBqBxgaV\nDE5OTtVu350+fVrg9QU6w0xOTsb8+fNr/cJMmTKFHYRBUFXt9z4/k6x6XdO+WrZsid69e/MN8KCv\nrw8ej4eHDx/Wa/+kdiUlJXB0dGTP2tesWQNDQ0OOUxHCPYH+7JWWlgp9QIBu3bqBYRikp6fzTZKV\nlpYGBQUFaGpq1rjO533bKyoqwDBMg85+avrLkpmZCTMzs3pvS5Zs374d9+7dAwD07dsX69ev5ziR\ndHn79i3fg0kiOfz9/dG1a9cGry/QGWafPn1w8uTJGt+rqKjAoUOH6t0XVktLCxoaGvjjjz/4lsfF\nxcHY2BiKiorV1hk2bBiSk5Px9u1bdllSUhJ4PB71xRWS+/fvY/PmzQAqR6A6fPgwmjdvznEq6fHf\nf/9hwIABSE5O5joKEQGBCub8+fPx+++/w87Ojh3z8MKFC9izZw/GjBmD+Ph4ODk51XvnLi4uCA8P\nR0REBF68eIGAgAAkJCTAxcUFAODl5YW5c+eynx8/fjw6deqEJUuW4PHjx7h9+zY2bdoEfX19evoo\nBOXl5Zg7dy5KS0sBAEuXLoWRkRHHqaRH1dXJ6tWroaury3UcIgqMgC5dusSMHTuW6d27N9+/MWPG\nMJcuXRJ0M9UcO3aMMTMzY3R0dJjx48czV69eZd9bvXo1Y2lpyff5ly9fMq6ursygQYOYwYMHM6tX\nr2by8/MbvP/PZWRkML169WIyMjKEtk1p4e3tzQBgADDdu3dnCgoKuI4kNV6+fMn06tWL2blzJ9dR\nSA2E9b3mMQzD1KfAvnz5kp1KtWPHjuzEaLKi6iwhJiamUfc6pM2TJ0+go6ODoqIiAEBMTAxMTU05\nTiUdXr9+jZEjR2Lq1KnYsGED13FIDYT1vRbooY+zszPGjx8PU1NTdOzYkWYElDEMw2D+/PlssZw/\nfz4Vy3p4//497O3tqX99EyBQwfz7779x5coVKCsrw9TUFNbW1hg2bFiND2aI9Dl06BCuXLkCoHKk\n/Z07d3KcSLp89dVX+OGHH7iOQcRAoIJ548YNJCYmIjo6GpcvX0ZUVBRUVVVhaWkJa2trGBkZ0eRX\nUur58+f4/vvv2df+/v743//+x2EiQiSXQAVTTk4OhoaGMDQ0hLu7O/7++29cvHgRly9fxqlTp9C2\nbVtYWVlRez0pwzAMnJ2d8f79ewCAnZ0dxo8fz3EqQiRXg0Yr0tPTw6pVq3Dp0iV4e3tDWVmZnUGQ\nSI8TJ07g3LlzAIC2bdti7969HCeSfGVlZQgLC0M9n5USGVHvDq4MwyAxMRGXLl1CTEwMsrKyoKqq\niilTpogiHxGR3NxcLFmyhH29b98+tGvXjsNEkq+8vByzZ89GXl4eJk+eTP3DmyCB/o8XFxfj+vXr\nuHz5Mq5cuYJ3796hVatWMDMzw9ixY2FiYgJ5eXlRZyVC5ObmhlevXgGo7BAwffp0jhNJtoqKCixc\nuBBZWVmIioqiYtlECfR/3dDQEMXFxWjZsiVfkaSn5NLp3LlzOH78OABAVVUVBw4coId2dWAYBm5u\nbkhJSUF0dLTQx1Ug0kOggllVJIcNG0Yjm0u5d+/e8c2BtGvXrmqDOBN+O3bsQHx8PGJiYmiIuyZO\noIJZNbV1/nIkAAAgAElEQVQqkX4rV67E8+fPAQCjRo3CvHnzOE4k+WbMmIF58+ZRcytSe8E0MzOD\nv78/vv76a4GGO+PxeLh8+bJQwxHhio2NxcGDBwEAysrKOHjwIF2KC0BDQ4PrCERC1FowDQwM0LJl\nSwDAkCFD6Isl5QoLCzF//nz2tYeHB3r06MFhIkKkT60F09PTk/1527ZtdW4kPz8f+fn5wktFhG7T\npk14+vQpgMqHeJ82KSKECEbgAYRTUlJqfT8+Pp7aYUqw1NRUdgIzRUVFHD58mJqB1eLEiRPYunUr\n1zGIhKrzoU9iYiKAymYV9+/f55stskp5eTkuXrxY43uEewzDYMmSJSgrKwMAfP/99+jXrx/HqSRT\nREQEli5dSvfiSa3qLJiLFi1CQUEBeDxeneP8MQwDc3NzoYcjjRceHs4WgK5du2LdunUcJ5JMFy5c\nwIIFC/D777+jf//+XMchEqrOgpmQkIDU1FTY2trC1dW1xvZ6PB4P7dq1o6kMJFBhYSGWL1/Ovvby\n8mIf5JH/7+rVq3BwcEBERAT09fW5jkMkWJ0Fk8fjoW/fvvD09ISpqSm1Q5Mynp6eyMjIAFDZ5nLq\n1KkcJ5I8FRUVcHd3R1hYGExMTLiOQyRcrQXzxYsXaN++PRQUFGBoaIjCwsIv3qfs3Lmz0AOShnn8\n+DE7ELCCggJ8fHyoaVgN5OTkEBcXBzm5Bg3cRZqYOhuunzp1Cv369YOpqalAX7bU1FShhiMNt3Tp\nUpSUlAAAlixZQg966kDFkgiq1oK5detWdrKgrVu30tmJFImMjERUVBQAoEOHDvjxxx85TkSIbKi1\nYE6aNIn92dbWVixhSON9/PgRbm5u7OudO3dCVVWVw0SS5dWrV1BXV6cTANIgAl+LREZG4tatW+zr\nGzduYNasWRg/fjzbP5lwb9euXWyPnm+++QazZs3iOJHkyMzMhIGBAeLi4riOQqSUQAUzLCwMP/zw\nAzIzMwEAGRkZcHJyQmZmJrp27QofHx+EhISINCj5smfPnrG9VOTk5ODr60tnUv8nOzsb5ubmcHZ2\nxsiRI7mOQ6SUQMO7hYaGYt68eWz3x7CwMPB4PISFhaFDhw4IDAzEyZMn6WyGYytWrGDnFnd2doae\nnh7HiSTD69evYWFhATs7O5o7nDSKQGeYGRkZGD58OPs6NjYWJiYm6NChAwBg4MCB7Nkn4ca1a9cQ\nHh4OAFBXV8dPP/3EcSLJ8O7dO1hZWWHMmDF19lYjRBACFczmzZujtLQUQGX7zKdPn2LUqFHs+x8/\nfqRLPw5VVFTw9ejZvHkz2rRpw2EiyVFcXIypU6di27Zt9DtKGk2gS3JtbW2EhoZCXV0dBw4cQPPm\nzWFhYcG+f+nSJXTv3l1kIUndjh07hj///BNA5chSn4572dS1b98eP/zwA9cxiIwQqGAuXrwY8+fP\nR0xMDBiGwbJly9gzmN27d+PYsWPYtWuXSIOSmn348AFr1qxhX3t5edGMhoSIiEDfLH19fVy4cAHJ\nyclQV1fHoEGD2Pd69+6Nffv2wdLSUmQhSe28vb3Z+8cWFhawsrLiOBEhskvgU5H27dvzXYZXGTt2\nrFADEcFlZWWxo+HLycnBy8urSd+nKy8vx9GjR+Hg4EDdHYlICFww//33Xxw+fBh37txBTk4O5OTk\n0KFDBxgbG2PevHlsN0oiPu7u7uyAKHPnzoWOjg7HibhTUVEBJycnPHnyBNOnT4eSkhLXkYgMEqhg\nJicnw8HBATweD7q6utDT0wPDMMjJycHp06dx/vx5HD9+nCbVEqPk5GQEBgYCAFRUVJp0MyKGYbB0\n6VLcv38f0dHRVCyJyAhUMPft2wcdHR34+flVGxMzLy8Pzs7O2L17N3x9fUUSkvBjGAYrVqwAwzAA\ngNWrV6Njx44cp+IGwzBYs2YNbty4gZiYGKioqHAdicgwgW70JCcnw9HRscYBhNu0aQNHR0ckJCQI\nPRyp2fnz5xETEwOgcs7sT9tgNjUBAQGIiopCdHQ01NTUuI5DZJxABbOkpATKysq1vq+mpoaPHz8K\nLRSpXXl5OV+7wq1bt9b5/0bWTZ06FZcuXYK6ujrXUUgTIFDB7NatG65evVrr+7GxsejWrZuwMpE6\nhISEsFMeDxo0CDNmzOA4EbfatGnTZG9HEPET6B7mjBkzsGnTJmRlZcHU1JTtQ/7y5UtcunQJV65c\nadIPHcSluLiYrz/0tm3bqPkMIWIkUMG0s7PDu3fvEBAQgOjoaLatH8MwaNWqFVatWtXgCbaCgoIQ\nEhKC7OxsaGpqYtGiRbC2thZo3c2bNyM0NBRHjx7FkCFDGrR/aXLgwAGkp6cDAExNTWlqY0LETOB2\nmE5OTpgzZw7u3buHnJwc8Hg8dOjQATo6OmjWrFmDdh4aGgpvb2/89NNP0NPTQ1xcHFauXAk1NTV8\n8803da579+5d/Prrr02mofb79+/h4eHBvm6Kg0mcPXsWcXFx8Pb25joKaaLq1elYSUlJqGdyAQEB\nmDFjBmxsbAAAWlpaSExMhL+/f50Fs6KiAhs3boSNjQ1+/fVXoeWRZN7e3nj16hUAYPLkyU3ijPpT\nFy9exPz583H+/Hmuo5AmrM4bYHfu3MGcOXMwaNAgDBw4ELNmzRLa8P5PnjxBdnZ2tbmgTUxMkJSU\nxM54WJPg4GAUFRXB0dGRbYsoy3JycuDl5QWgsgvkli1bOE4kXlevXsWsWbNw5swZDB48mOs4pAmr\ntWD+9ddfmDNnDlJTU2FgYAATExN2aorIyMhG7zg9PR08Hg9dunThW66hoYGKigpkZGTUuN7Lly/h\n4+ODjRs3QlFRsdE5pIGHhwcKCgoAAN999x20tbU5TiQ+t27dwtSpUxEWFvbF2zSEiFqtl+T+/v7o\n2bMnjhw5wjZYLy0txapVq7Br1y6MGzeuUTuu6gPdokULvuVVr6sKxOe2bNkCCwsLGBoa4vnz543K\nIA3S0tJw4MABAJUDOTe1KXM9PDxw5MgRmJqach2FkNoL5l9//YUNGzbw9e5RVFSEm5sbrKys8Pz5\n82pnh6IWGxuLxMREXLhwQSjbq2n64LpuBXDhxx9/ZEe7d3V1hYaGBseJxOu3335rcg+3iOg4OTlV\ne0h9+vRpgdev9ZL8/fv30NTUrLZcU1MTDMPUegYoqFatWgGofiZZ9frzPsFFRUXYsmULVq1ahdat\nWwOAzN+/fPjwIY4ePQoAUFVV5RsouKmgYkkkSZ1PyeXl5ast+7QNZmN069YNDMMgPT0dX3/9Nbs8\nLS0NCgoK1Yr1P//8g6ysLGzYsAHu7u58eWbPng0NDQ1cvHixXhlq+suSmZkJMzOzeh6NaGzdupX9\n77x8+XLq/kdII/n7+zdqKErO5jLQ0tKChoYG/vjjD74CFRcXB2Nj42oPdHR0dHDu3Dm+ZTk5OXB0\ndISnpycGDhwoltzi8vTpU4SGhgKoPLt0c3PjOJHo5ebmonXr1jTFBpFYdf5mpqSk4MOHDzW+d//+\nfeTn5/Mtq2/bQBcXF7i7u0NPTw8GBgaIjIxEQkICWyi8vLyQkpKCw4cPQ0lJCT179uRbv2rQiS5d\nushcX/Zt27ahvLwcQOWcSrI+Ek9OTg5GjBgBT09Ptl0uIZKmzoJZ1zzO69evZ39mGAY8Hg+pqan1\n2rmNjQ2Kiorg6+uLnJwcaGlpwc/PD7q6ugCAV69efXG+c1m8x5Weno6goCAAQMuWLbF06VJuA4lY\nXl4eLCws8O2331KxJBKt1oLp6ekplgB2dnaws7NrUIYuXbrUu0hLgx07drBPxp2dndG2bVuOE4nO\nu3fvMHr0aFhaWja5JlNE+tRaMCdNmiTOHOT/ZGVl4dChQwAqu6KuWLGC40SiU1hYCGtraxgYGGDH\njh0yebVAZAuNDSZhdu7cieLiYgDAggULZHqsx/LyckycOBE+Pj5ULIlUoMeREiQnJwf+/v4AgGbN\nmvGNrC6LVFVVsXLlSq5jECIwOsOUILt370ZRUREAwNHRUew9qQghdaOCKSHy8vLYWTcVFBSwevVq\njhMRQj5HBVNC7N27l+0W6uDgIHPtShmGQUBAgMT11SekPupVMJ8/f47z588jKCgIr1+/BlD7qEJE\ncO/evcPevXsBVI53KWt9xhmGwdKlSxEUFEQFk0g1gR76lJWVYdOmTQgPD0dFRQV4PB6MjIygrq4O\nHx8f3Lt3DwEBAdUGzCCC8fX1xbt37wBUtkv9vEeTNGMYBmvXrsX169cRExNDvyNEqgl0hunv74/f\nfvsNixYtwunTp/kG3rCyskJ6ejr2798vspCyrKCgALt37wZQ2Wtp3bp1HCcSLg8PD5w7dw4XL16U\n+e6dRPYJVDDPnj0LFxcXuLq6om/fvnzvDRw4EEuWLEFUVJRIAsq6AwcOsLc3pkyZgj59+nCcSHhO\nnDiB4OBgXL58WaZ7K5GmQ6BL8qysLAwaNKjW93v27Ml+6YngioqKsGvXLvb1p/3zZcGECRMwfPhw\nmW58T5oWgc4wW7dujf/++6/W91NTU9GmTRuhhWoqDh48iJycHADAxIkTMWDAAI4TCVeLFi3QuXNn\nrmMQIjQCFUxTU1Ps2bMH169fZ5fxeDyUlJQgIiICu3btgrm5uchCyqLi4mLs2LGDff3poMiEEMkk\n0CX5ihUrcP/+fcyfP5+dpMzBwQEFBQUoLy9H//79sWzZMpEGlTW//PILO4nbmDFjoK+vz3EiQsiX\nCFQwVVVVERYWhosXL+LGjRvIzs4GAHTq1AnGxsYYPXp0jdNZkJqVlpZi27Zt7GtZOLuMjo7GsWPH\n2HE8CZFFAg++IS8vj7Fjx2Ls2LGizNMkHDt2DM+ePQNQebvD2NiY40SNExcXh5kzZ+LMmTNcRyFE\npAQqmImJiQJtrL5TVDRFDMOw7S4B6X8yfuvWLUyZMgUnTpzA0KFDuY5DiEgJVDDt7e0FGq9QFkc/\nF7Zr164hOTkZAKCrq4uRI0dyG6gRkpKSMGHCBBw5ckRiZtokRJQEKpg1TRXBMAxyc3Nx5coVNGvW\nDAsXLhR6OFlU1WccANzc3KR64Ny9e/fiwIEDdJuGNBkCFcy6pqtYuHAhfvjhByQnJ+Obb74RWjBZ\nlJaWhrNnzwIA2rVrV+tcRtIiKChIqgs+IfUllOHdJk2ahBMnTghjUzLN19cXFRUVACr/0CgpKXGc\nqHGoWJKmRigFMy8vr9oc5YRfQUEBO7mZgoICnJ2dOU5ECKkvgS7JIyIialxeWlqKFy9eICwsDL17\n9xZqMFkTHBzMDuE2bdo0qesy+OrVK7Rs2RLKyspcRyGEMwIVzNWrV4PH4/EN6/aprl27Sn3zGFGq\nqKjAvn372Ndubm4cpqm/vLw8mJubw83NDd999x3XcQjhjEAFMzg4uMblcnJyUFVVRY8ePainTx2i\no6Px8OFDAICxsTEMDAw4TiS49+/fw8rKCubm5pgzZw7XcQjhlEAFs6ioCAMGDEDr1q1FnUcmfd6U\nSFoUFhbC2toagwcPxs6dO+khD2nyBHros3z5crYrH6mfBw8e4MKFCwCALl26wNbWluNEgvn48SNs\nbGzQs2dP+Pr6UrEkBAIWzEmTJtEEVg3k4+PD/uzi4gJFRUUO0wiOx+PB2toahw4dgpwcTS5KCCDg\nJXmLFi2QmZkJExMT6Orqok2bNlBQ4F+Vx+Nh69atIgkprd6+fYsjR44AAJSUlLBgwQKOEwmuefPm\nWLp0KdcxCJEoAhXMgIAA9ucbN27U+BkqmNUdPnwYhYWFAIBZs2ZBXV2d40SEkMYQqGA+ePBA1Dlk\nTnl5OXx9fdnXS5Ys4TANIUQYar05FRERwTa0JvX322+/IS0tDUDlmJc6OjrcBqoDwzA4cOAA9dYi\n5AtqLZhr1qxBZmamOLPIFGlqSrR+/XocPHgQ5eXlXEchRKLVekleW68e8mXJycmIi4sDAHTv3h3W\n1tYcJ6qdh4cHIiIiEBcXBzU1Na7jECLRBJ6iggju559/Zn9evHixxPaC2r17N44cOYK4uDi0bduW\n6ziESLw6C+arV6/w4sULgTcmbQNKiEJRURGOHTsGoLJpzuzZszlOVLMLFy5g3759iIuLQ6dOnbiO\nQ4hUqLNgOjk51WtjNEUFEB4ezj4smzx5ssR2JzU3N8f169fRpUsXrqMQIjXqLJhTp05F+/btxZVF\nJlSNeQkAc+fO5TBJ3RQUFKhYElJPdRbMb7/9Fv369RNpgKCgIISEhCA7OxuamppYtGhRnQ9Jbt68\nCR8fHzx69AgqKiowMTHB999/LxGNwh8/fsz3sEeaJzgjhFTHaSfh0NBQeHt7w9XVFefOncO0adOw\ncuXKWnsTJSUlYcGCBdDT00N4eDh27tyJpKQkienCFxgYyP7s6OgoUX2wqdUDIY3H6Tc6ICAAM2bM\ngI2NDbS0tDB79myYmprC39+/xs8fOXIEvXr1wqpVq6ClpQUDAwMsWbIEd+7cwcuXL8Wcnl9ZWRmC\ngoIAVI4TKkljR167dg0TJ06koklII9V6ST5p0iSRPrB48uQJsrOzYWJiwrfcxMQEHh4eKCkpQbNm\nzfje2759O4qKiviWVV2Kv3nzBh07dhRZ3i+5cOECsrKyAABWVlYSc3/w9u3bmDJlCo4fP05DtBHS\nSLWeYXp6eoq0mVB6ejp4PF61wqKhoYGKigpkZGRUW0dJSalaEY+NjYWKigp69OghsqyCOHz4MPuz\npDzs+fvvvzFhwgT88ssvMDMz4zoOIVKPs0vyqlF8WrRowbe86nVBQcEXtxEfH4+QkBA4OTlVOxsV\np5cvX+LcuXMAKucbHzduHGdZqqSkpGDMmDHYv3+/RPc0IkSaSG1Pn5s3b8LFxQWjR49u8BldTaOf\nN2SQ5ODgYLYftoODA6fFu8qhQ4ewc+dOTJ48mesohEiMmk6uTp8+LfD6nBXMVq1aAah+Jln1WkVF\npdZ1Y2NjsXTpUlhbW8PDw0N0IQXAMIxEXo57eXnRPUtChIyzgtmtWzcwDIP09HR8/fXX7PK0tDQo\nKChAU1OzxvUSExPh5uaGmTNnYvXq1Y3KUNNflszMzHrd77tx4wYePXoEoPKBVZ8+fRqVSVioWBJS\nnb+/P7p27drg9Tm7h6mlpQUNDQ388ccffMvj4uJgbGxc49w3ubm5cHV1xZQpUxpdLIVFEs8uCSGi\nwek9TBcXF7i7u0NPTw8GBgaIjIxEQkICQkNDAVReVqakpLBFae/evWjevDkWLlyIV69e8W2rVatW\naN68uVjz5+fn4+TJkwAqbyFMmzZNrPuvkpeXBx6PJ7H91gmRFZwWTBsbGxQVFcHX1xc5OTnQ0tKC\nn58fdHV1AVSOlvTpIMbx8fHIzc3FqFGjqm3L09MTNjY2YssOVI5K/+HDBwDAtGnT6rzvKirv37+H\nlZUVpk+fjuXLl4t9/4Q0JZw/Jbezs4OdnV2N73l6evK9jomJEUckgR0/fpz9eebMmWLff2FhIcaN\nGwd9fX0sW7ZM7PsnpKmRnM7OUubVq1e4dOkSAKBTp04YMWKEWPf/8eNH2NjYoHv37vDz86OHPISI\nARXMBjp16hTKysoAVF6Oi3NU9dLSUkydOhVt2rTBoUOHJGqQD0JkGX3TGujTy/HabimIipycHMaO\nHYuQkBAoKHB+V4WQJoO+bQ2QmZnJNofq3r07DAwMxLp/eXl5ODs7i3WfhBA6w2yQU6dOsUOlTZ8+\nne4fEtJEUMFsgE97CE2dOpXDJIQQcaKCWU/Z2dm4fv06gMrL8ao2o6K0f/9+5OTkiHw/hJC6UcGs\np4iICPZy3NbWVuSX41u3bqVmQ4RICHroU0+fXo7XNDycMO3ZswdBQUGIi4tDu3btRLovQsiXUcGs\nhzdv3iA2NhYA0LlzZxgaGopsXz///DP27NmDa9euoVOnTiLbDyFEcFQw6yEyMpJtrD5p0iSRNRiP\nj4/Hli1bcPXq1VqHuSOEiB8VzHoQ1+W4kZEREhIS6MySEAlDD30EVFhYiAsXLgAA2rRpg+HDh4ts\nXzwej4olIRKICqaALly4gI8fPwIAJk6cSF0SCWmCqGAKKDw8nP1Z2JfjVc2UCCGSjQqmAMrKyvD7\n778DAFq2bAlzc3Ohbfv27dsYOXIkO+skIURy0XWlAG7fvo23b98CAMzNzaGkpCSU7f7999+YMGEC\nAgMDxTo8HCGkYegMUwBVZ5cAMGbMGKFsMyUlBWPGjIGfnx+sra2Fsk1CiGhRwRRA1dNxALCysmr0\n9h4/fgxLS0vs3LkTU6ZMafT2CCHiQQXzC7Kzs/Hnn38CAPr06YNu3bo1epu//vorNmzYgFmzZjV6\nW4QQ8aF7mF9w8eJF9mdhXY6vWbNGKNshhIgXnWF+waeX48IqmIQQ6UQFsw7l5eXsGWaLFi0wbNgw\njhMRQrhEBbMOiYmJyMvLAwCYmpqiefPm9d7G+/fv8eLFC2FHI4RwgApmHRp7//LDhw8YN24c/P39\nhRmLEMIRKph1uHr1KvuzhYVFvdb9+PEjbGxsoKWlhY0bNwo3GCGEE1Qwa1FcXIxbt24BqBwsuGfP\nngKvW1paimnTpkFNTQ2BgYEiGzeTECJe9E2uRXJyMjs60YgRIwSeU6e8vJxtXxkSEkKjGhEiQ6hg\n1uL27dvszyNHjhR4PTk5OYwdOxYnT55Es2bNRJCMEMIVOv2pRdXlOFB5hikoHo+H2bNniyISIYRj\ndIZZi6SkJABAhw4d0KtXL47TEEIkARXMWlTdv/zmm29oTnBCCAAqmF/0pal0Dxw4gP/++09MaQgh\nXKKC+QV1Fcy9e/fCy8urQT2ACCHShx761EFOTg76+vo1vhcQEIDdu3cjLi4OnTt3FnMyQggXqGDW\noX///lBRUam2/OjRo/jpp59w9epVoYyPSQiRDnRJXod3794hOjqab1bHlJQU/PDDD4iOjq5X7x9C\niPSjglmHZ8+eYfTo0dDR0cG///4LAOjbty+Sk5PRt29fjtMRQsSN84IZFBQEc3Nz6OjowNraGlFR\nUXV+/t69e7C3t4euri6MjIywceNGtgmQqNy/fx8mJiZs0Wzfvr1I90cIkUycFszQ0FB4e3vD1dUV\n586dw7Rp07By5UrcuHGjxs/n5ubC0dERXbt2RXh4OPbs2YObN2/C3d1d5FlfvXqFSZMm8V2eE0Ka\nFk4LZkBAAGbMmMEOgzZ79myYmprWOn7k0aNH0axZM2zevBk9e/aEkZERVq1ahcjISGRmZoo87/37\n93H58mWR74cQIpk4K5hPnjxBdnY2TExM+JabmJggKSkJJSUl1da5desWDAwM+EYAqlo/Pj5etIH/\nz/79+8WyH0KI5OGsYKanp4PH46FLly58yzU0NFBRUYGMjIxq6zx79qza55WVlaGuro60tDRRxmXF\nxcWJZT+EEMnDWcEsLCwEUDm52KeqXhcUFNS4jrKycrXlLVq0qPHzopCfny+W/RBCJA/nT8mlTatW\nrbiOQAjhCGc9faoKz+dnhlWva+ph06pVqxrPJPPz8xtUyGxtbastKy4uBoBaR0o3NjYWywMmQojw\nvHz5EgCwcOHCamM/nD59WuDtcFYwu3XrBoZhkJ6ejq+//ppdnpaWBgUFBWhqata4zuf3Nt+/f483\nb96gR48eQsn17NkzAKhx/wDw+PFjmJmZCWVfhBDxKisra9RgOZwVTC0tLWhoaOCPP/7gK0BxcXEw\nNjaGoqJitXWGDRuG4OBglJSUsNM/XL16FfLy8hg6dGi9M9T0l8XGxgZFRUUICAiAvLx8vbcpCZyc\nnABAqqf3pWOQDLJwDEDlmWVZWRnOnj0LJSWlBm+H08E3XFxc4O7uDj09PRgYGCAyMhIJCQkIDQ0F\nAHh5eSElJQWHDx8GAMycOROhoaFYu3YtFi9ejJcvX8LLywvTp09Hu3bthJJJTk4OLVu2lOpBNar+\nmHTt2pXjJA1HxyAZZOEYAKB58+Zo3rx5o4olwHHBrDqb8/X1RU5ODrS0tODn5wddXV0Alb1rPr1f\nqKamhqCgIGzZsgUTJ06EiooKJkyYgOXLl3N1CISQJoTz4d3s7OxgZ2dX43uenp7VlvXq1QvBwcGi\njkUIIdVQsyJCCBEQj6HRJAghRCB0hkkIIQKigkkIIQKigkkIIQKigkkIIQKigkkIIQKigkkIIQKi\ngkkIIQKigkkIIQKigkkIIQKigkkIIQJqcgUzKCgI5ubm0NHRgbW1NaKiour8/L1792Bvbw9dXV0Y\nGRlh48aN+Pjxo5jS1qy+x3Dz5k3Y2dlBX18fI0aMwJo1a/D69Wsxpa1dfY/jU5s3b4a2tjYSExNF\nmPDL6nsMBQUFcHd3h6GhIQYNGoR58+bVOOGfONX3GOLj4zFz5kwMGTIE+vr6cHFxYQfe5grDMNi3\nbx/69OkDX1/fL36+wd9rpgkJCQlhdHR0mDNnzjD//fcfExQUxPTp04e5fv16jZ/PyclhBg8ezKxe\nvZr5999/mfj4eMbCwoL5/vvvxZz8/6vvMfz5559Mv379mG3btjH//fcfc/v2bcbS0pKZNWuWmJPz\nq+9xfCo5OZnR0dFhtLW1mYSEBDGkrVlDjmHWrFmMvb098+DBAyY1NZWZPn06Y21tLcbU/Op7DPfu\n3eP7fUpJSWEcHByYUaNGMR8+fBBz+kp5eXmMo6MjY2lpyfTr14/x8fGp8/ON+V43qYI5fPhwxtPT\nk2+Zi4tLrcXDy8uLMTExYUpLS9llly9fZrS1tZmMjAyRZq1NfY9hyZIlzKRJk/iWRUZGMtra2kxW\nVpbIcn5JfY+jSnl5OTNp0iTG3d2d84JZ32O4du0ao6enx7x584ZdlpmZyVy8eJEpLi4Wadba1PcY\ntm/fzhgaGvItS01NZXr37s1cu3ZNZDnrEhQUxCxcuJDJz89ndHR0vlgwG/O9bjKX5E+ePEF2djZM\nTEz4lpuYmCApKQklJSXV1rl16xYMDAz4JkSrWj8+Pl60gWvQkGPYvn07O2J9FXV1dQDAmzdvRBe2\nDiTfX9YAABEISURBVA05jirBwcEoKiqCo6MjGA4H2mrIMVy5cgWGhoZQU1Njl3Xp0gWWlpbsyObi\n1JBj4PF4kJPjLxtV08nweDzRha2Dubk5/P39a5w4sSaN+V43mYKZnp4OHo+HLl268C3X0NBARUVF\njfeRnj17Vu3zysrKUFdXR1pamijj1qghx6CkpITWrVvzLYuNjYWKiorQJo6rr4YcB1A585+Pjw82\nbtxY45xP4tSQY3j06BG0tLRw8OBBjB49GsbGxli+fDny8vLEFZtPQ47B1tYWRUVFCAwMRHFxMYqK\nirB//3589dVXMDIyEld0Pp/n/5LGfK+bTMEsLCwEALRo0YJvedXrmqbvLSwshLKycrXlLVq0qPHz\notaQY/hcfHw8QkJC4OTkxMlZDdDw49iyZQssLCxgaGgo2oACaMgx5OXl4ffff8ejR4/g7e0NT09P\nJCcnw8HBARUVFaIP/ZmGHEOPHj3g5+eH/fv3Y+DAgdDX18eDBw9w6NChWqemljSN+V5LxxESobh5\n8yZcXFwwevRozJ07l+s49RIbG4vExERcuHCB6ygNVlZWBiUlJezYsYO9fG3evDm+++47XL9+HcOH\nD+c44Zc9evQIy5cvh62tLSZMmMDOsLpw4UKEhYWhZcuWXEcUqSZzhtmqVSsA1f9qVr2u6f5Hq1at\navyLk5+fz25PnBpyDFViY2Ph5OQEKysreHl5iS6kAOp7HEVFRdiyZQtWrVrF3l7g8v4l0LD/Fy1b\ntkTv3r357vXp6+uDx+Ph4cOHIkxbs4Ycg6+vLzQ0NLB27Vr0798fQ4YMwe7du5GZmYnw8HDRhxaC\nxnyvm0zB7NatGxiGQXp6Ot/ytLQ0KCgoQFNTs8Z1Pr+P8/79e7x584aT+38NOQYASExMhJubG2bM\nmAFPT89qN+3Frb7H8c8//yArKwsbNmxAv3790K9fP4wePRo8Hg+zZ8/G6NGjxRkfQMN/n969e8e3\nrKKiAgzDCPzAQpgacgxPnz5F9+7d+ZapqKigTZs2nLfFFFRjvtdNpmBqaWlBQ0MDf/zxB9/yuLg4\nGBsb1/gQYdiwYUhMTOR7Wnj16lXIy8tj6NChIs/8uYYcQ25uLlxdXTFlyhSsXr1aXFHrVN/j0NHR\nwblz5xAREYGzZ8/i7NmzOHjwIBiGwdatWxEQECDO+AAa/vuUnJyMt2/fssuSkpLA4/HQu3dvkWf+\nXEOOoUOHDtUejOTn5yM3NxcdOnQQZVyhadT3ujHtn6TNmTNnmP79+zNnzpxhnj9/zvz8889Mv379\nmL///pthGIbZtWsX4+joyH7+zZs3jLGxMbNixQomLS2NuXXrFjN8+HDmp59+4uoQ6n0M69atY4YN\nG8ZkZWUxubm5fP8+fvzI1WHU+zg+l5mZyfTu3ZvTdpj1PYbi4mLGysqKsbe3Z/7991/m1q1bjKWl\nJTNz5kyuDqHex3DlyhVGW1ub2b17N/PkyRMmJSWFWbRoEaOnp8ekp6dzcgxv375lcnNzmZycHEZH\nR4fZvn07+zteXl4u1O91k3roY2Njg6KiIvj6+iInJwdaWlrw8/ODrq4uAODVq1fIzMxkP6+mpoag\noCBs2bIFEydOhIqKCiZMmIDly5dzdQj1Pob4+Hjk5uZi1KhR1bbl6ekJGxsbsWX/VH2PoyZctfur\nUt9jaNasGfv79O2330JOTg7m5uZYt24dV4dQ72MYOXIkfHx84Ofnh8DAQCgqKmLAgAE4cuQINDQ0\nODkGV1dX3Llzh339yy+/IDAwEDweDzExMUL9XtM0u4QQIqAmcw+TEEIaiwomIYQIiAomIYQIiAom\nIYQIiAomIYQIiAomIYQIiAomIYQIqEk1XG8qfH1965zXhMfj4fr16+xAwl+SkJAABwcHHD16FEOG\nDBFWzFrZ29tXm6tHQUEB7du3h5mZGRYvXgxVVVWR7JfH4yE4OLjWz2hra8PV1RWurq5C3//nqv67\nf05OTg5t2rTBkCFD4OzsjF69eok8C6lEBVNG8Xg8+Pv7o23btjW+//mgwoJsT5zat28Pf39/dlSi\nwsJC/PXXX/j555+RkJCAU6dOCX0Q4c2bN/O9jo6OxrZt2xAbG8suCw8PR/v27YW63y9ZvHgxRo4c\nyb4uKSnB48ePERAQgOnTp+PXX3+t92Awv/zyC2JjY3H06FEhp5VtVDBl2Ndff43OnTtzHaNBFBUV\n0bdvX75lBgYGaN++PdasWYPz589j4sSJQt2nlpYW3+uEhIRqfyj69esn1H0KokuXLtX2O3DgQAwZ\nMgRWVlYIDg7Gpk2b6rXNxMREzruWSiO6h9nEffjwAV5eXhg1ahT69++PkSNHYvXq1cjNza1zvUOH\nDmHs2LHQ09ODkZERFi5ciAcPHvB95s6dO5gzZw4GDhyIgQMHYsaMGbh582aj8ladad27d49dlpeX\nh/Xr12Po0KHo378/RowYgc2bN1cb8/BLme3t7dlLYHt7e4SEhOD58+fQ1tbGmjVrAFRekvv6+qKw\nsBC6urrs8k/FxsZCW1sbly5dAlB5Rrhr1y6Ympqy+bZu3cqOeN5QWlpaUFNTw4sXL/iWh4WFYcKE\nCdDR0YGRkREcHR35/nuZmpoiNjYWCQkJfNPSMgyDgwcPwsrKCv3794eJiYnETMksKahgNnHu7u44\nevQonJ2dERISgiVLliA2NhaLFy/m+9ynQw4cPXoUe/bswbRp0xAYGIht27ahsLAQc+bMwYcPHwAA\nd+/exZw5c8Dj8bBv3z74+PhATU0N8+fPb9Rc4lXTIFRN6VBSUoLZs2fj6tWrcHNzQ1BQEJydnfHb\nb7/BycmpXpk/9f/aO/eQJr8/jr8fL9NaYowmikaFJTPnkjayJmZespYGjqKkRjdNo0gUukdaRhZo\ngtrFC8HKFEdkw0hInNqFylJKsOxKRUbN7GKUMtv6fP+QPT/XTJff4PdFzwv84znnPM95n+eBj+dz\nztnnk52djdmzZ8PLywvV1dV270MoFCI8PBxNTU12wYzr6uowadIkREREAAAyMjJQWVmJdevWQavV\nIjU1FXq9HmlpaaN+DwBgNBrR09ODadOm8WU6nQ5ZWVkICwvD2bNncfToUXR3dyM5ORnd3d0AgJKS\nEojFYgQFBeHixYtYvXo1gIFgLAUFBYiPj4dWq8WuXbtw+/ZtbNiwAT9+/PhXWscKzCUfxxARnJyc\nsHXrVqxatQoAEBISgtevX6O0tBRv374dMsHUrVu3EBAQgA0bNvBlMpkM1dXV+PbtGyZOnIjCwkJ4\ne3ujpKSEzx2kVCqhUqlQVFQ07MbKcLS2tgL4n2tcW1uL58+f49SpU3xEJoVCASJCdnY2mpubERoa\n6pDmwcyYMQNCoRA9PT12SwNWli1bBoPBgJaWFn4zzGw2o7GxEYsXL4ZAIEBbWxsMBgMOHTrEGyaF\nQoGJEydi7969uHfv3h9vpPX39+PJkyfIycmBQCBAYmIiX/f582fExcVh9+7dfJlAIEBSUhJu3LgB\ntVqNWbNmwdXVFUKhkB+b0WhEZWUlNm3axG9oKRQK+Pn5QaPR4MqVK/+3yFb/JdgMcwwzUiAqjuOQ\nm5uLlJQUm3I/Pz8AwLt374a8b8qUKXj27BkuXLgAk8kEABCJREhOToaXlxfMZjPu3r2LiIgIm0Rr\nTk5OfBDdP6Wvrw+3bt3C4cOH4evri7i4OABAc3MzXFxc7PLhLFq0CETEh/0aSfNoiIyMhJubG+96\nAwMpXHt6ehAfHw8AuHnzJjiOQ2xs7JD6Hjx4MGI/e/fuhUQi4f9kMhlWrVoFFxcXaLVazJw5k2+7\nZcsWuxQk1rBrv/ueVt0Wi8VOp1wuh4eHh0M6xwNshjlGISJER0cPWeft7Y2mpiYAwOPHj1FWVoa7\nd+/i06dPsFgsAAaM6WCDO3iDYMeOHXj16hUyMzNx+PBhyOVyLF26FGq1GgKBAF++fEF/fz8qKytR\nUVFh0zfHceA4Dj09PfD09Pytfuva4WCcnJwQGhqK7OxsuLu7AwC6urogEong7Oxs01YsFvP1jmge\nDRMmTEBERAQMBgP27dsHYMAdF4lEWLBgAd8/EfHXv74Lo9E4Yj9paWk28UyLi4tx7do15OXl2UU5\n//z5M4qLi9HQ0ACj0chHFec4btjMlEajEUSElStXjlrneIAZzDEKx3EoLS3lDcdgrMdx3r9/jzVr\n1sDT0xMZGRmYOXMmBAIBDAbDsOc4PT09UV5ejvb2dtTX16OxsRFZWVmoqKiATqfj2yUkJGD9+vVD\nPmOk7IJeXl58GgpgYO1SLBbbGdnf7fRa77PmLxpJ81BpVx1BpVKhrq4OHR0dkEgkaGhowNKlS23y\nJnEcB51ON6Rhnjx58oh9+Pj42Pzz2LlzJxoaGpCXl4fc3Fybtps2bcKLFy+wbds2KBQKCIVCdHV1\n2XkRQ8FxHIqKingPYzBjPRukozCDOYbx9/cf9lhRfX09+vr6UFRUhLCwML68sbHRoedLpVJIpVKk\np6dDr9djz549qK2tRUJCAtzc3GA2m+1miY7i6urqUJ4bb29v3LlzB2az2SYvtnVG9OsM7HeaV6xY\nMSqdixYtgru7OwwGA75//47u7m5+uQAYMHbAgGH8XZK6P8XX1xdr166FVquFRqPho6M/ffoUHR0d\nSE1NRWpqKt/ekd14q06hUDjqbzYeYGuY4xiz2QwANrPQ3t5ePl2q1T3/9Z5jx47ZJc6yuoxfv36F\ns7Mz5s2bh6amJrujPaWlpdDr9X9tDEqlEhaLhV9isNLY2AiO46BUKh3SPBQjubEA4O7ujsjISNy4\ncQP19fXw8fGBXC630UdEqKmpsbnvzZs3yMzMxPv37x0dqg1btmyBh4cHcnJy+LKhvicAnD9/3m4s\nv16HhobCycnJTue3b9+wf/9+PH/+fFQ6xxpshjmOse4m5+XlISUlBR8/fsTp06exfPlynD59Glev\nXuVnqINd49evX+Py5ctIS0tDQEAA+vr6UFlZCaFQiKioKAADv07RaDRYv3490tLSMGHCBNTV1aGi\nogIHDx78a2OIjY1FYGAgDh48yB+xefToEQoKChAVFQWZTAYAw2r+3VqvWCxGa2srdDod/Pz8bGbh\ng1GpVEhPT8eHDx+gUqls6mQyGWJiYlBcXAxnZ2fMnz8fnZ2dOHnyJH7+/Mmvff4pnp6eSElJQV5e\nHvR6PRISEuDv7w+RSITy8nJMnToVLi4uqKqqwvTp0+Hq6orbt2/zZ0HFYjE6OjpQU1MDX19fyOVy\naDQanDt3Dh4eHliyZAk+fvyIsrIydHZ2/usjUGOGv5W5jfHfoaioiCQSCb19+3bEtlVVVRQTE0Nz\n5syhhIQEMhgMZLFYKCkpiWQyGR0/fpyam5tJIpHwGRp7e3vpyJEjFBkZScHBwaRUKmnz5s3U1tZm\n8+z79+/Txo0bae7cuRQcHExqtZquXLkyoiaNRkPR0dEOj/fLly904MABCgsLo6CgIIqKiqL8/Hwy\nmUx8G0c0azQaWrduHX/98OFDio2NJalUStu3byciIolEQidOnLDp32Qy0dy5cykwMJDa29vt9JlM\nJsrPz6fo6GgKCgoipVJJBw4cIKPROOy4rO/90qVLQ9abTCaKjIykhQsXUm9vLxERtbS0kFqtppCQ\nEIqKiqKSkhIiIiosLKSQkBBKTEwkIqLr169TeHg4BQcHU05ODhER/fz5k86cOUMqlYqkUinNmzeP\n0tPT6eXLl8PqHE+wJGgMBoPhIGwNk8FgMByEGUwGg8FwEGYwGQwGw0GYwWQwGAwHYQaTwWAwHIQZ\nTAaDwXAQZjAZDAbDQZjBZDAYDAdhBpPBYDAchBlMBoPBcJB/APlAXc65WDnEAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f976a265290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "MODEL_VERSION = \"v4\"\n",
    "MODEL_SUFFIX = \"%s.last_author\" % MODEL_VERSION\n",
    "\n",
    "results_full = fit_model(df_filtered_la, base_formula, include_prc=True,\n",
    "                            MODEL_VERSION=MODEL_VERSION, MODEL_SUFFIX=MODEL_SUFFIX, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "results_filtered = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" \n",
    "                        % (MODEL_VERSION, MODEL_SUFFIX, version))\n",
    "with open(\"Filtered_models.%s.filtered_%s.txt\" % (MODEL_SUFFIX, version), \"wb+\") as fp: \n",
    "    print_model(results_filtered, fp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Fit models version 2\n",
    "\n",
    "\n",
    "**Filters applied**\n",
    "\n",
    "\n",
    "* pub type = journal article\n",
    "* affiliation = USA\n",
    "* language = English\n",
    "* \\# source references = [10,x], where x represents the second kink in the fit vs. empirical plots (perhaps 100 for first authors and ~ 40 or 50 for last authors).\n",
    "* eth1 = ENGLISH, eth2 = UNKNOWN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "base_formula = (\"is_self_cite ~ I(auth_prev_papers == 0) + I(auth_prev_papers == 1) \"\n",
    "           \"+ np.log10(auth_prev_papers + 1) + I(np.log10(auth_prev_papers + 1)**2)\"\n",
    "           \"+ C(gender, levels=GENDERS)\"\n",
    "           #\"+ mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)\"\n",
    "           \"+ np.log10(source_ncites) \"\n",
    "           \"+ I(np.log10(source_ncites)**2) + I(np.log10(source_ncites)**3)\"\n",
    "           \"+ I(source_n_authors > 20) + np.log10(np.clip(source_n_authors, 0, 20))\"\n",
    "           \"+ I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)\"\n",
    "           \"+ np.log10(source_n_mesh_ex + 1) + I(source_n_mesh_ex == 0)\" \n",
    "           \"+ np.log10(sink_n_mesh_ex + 1) + I(sink_n_mesh_ex == 0)\"\n",
    "           \"+ I(year_span < 0) + I(year_span == 0) + mf.score_log_1(year_span) + I(mf.score_log_1(year_span)**2)\"\n",
    "           \"+ I(sink_prev_ncites == 0) + np.log10(sink_prev_ncites + 1) + I(np.log10(sink_prev_ncites + 1)**2)\"\n",
    "           \"+ I(jj_sim == 0) + np.log10(jj_sim + 1) + I(np.log10(jj_sim + 1)**2) + journal_same\"\n",
    "           \"+ sink_is_eng + sink_is_journal + sink_is_review + sink_is_case_rep + sink_is_let_ed_com\"\n",
    "           \"+ np.log10(np.nan_to_num(source_V_novelty) + 1)\"\n",
    "           \"+ np.log10(np.nan_to_num(sink_V_novelty) + 1) + I(np.log10(np.nan_to_num(sink_V_novelty) + 1)**2)\"\n",
    "          )\n",
    "\n",
    "version = 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### First author"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/content/smishra8/SOFTWARE/anaconda2/lib/python2.7/site-packages/ipykernel/__main__.py:10: UserWarning: Boolean Series key will be reindexed to match DataFrame index.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5019975, 54) (41618369, 54)\n"
     ]
    }
   ],
   "source": [
    "df_filtered = filtered_data(df, max_source_ncites=100, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Created figure with 1 rows and 1 columns\n",
      "(5019975, 54) is_self_cite ~ I(auth_prev_papers == 0) + I(auth_prev_papers == 1) + np.log10(auth_prev_papers + 1) + I(np.log10(auth_prev_papers + 1)**2)+ C(gender, levels=GENDERS)+ np.log10(source_ncites) + I(np.log10(source_ncites)**2) + I(np.log10(source_ncites)**3)+ I(source_n_authors > 20) + np.log10(np.clip(source_n_authors, 0, 20))+ I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)+ np.log10(source_n_mesh_ex + 1) + I(source_n_mesh_ex == 0)+ np.log10(sink_n_mesh_ex + 1) + I(sink_n_mesh_ex == 0)+ I(year_span < 0) + I(year_span == 0) + mf.score_log_1(year_span) + I(mf.score_log_1(year_span)**2)+ I(sink_prev_ncites == 0) + np.log10(sink_prev_ncites + 1) + I(np.log10(sink_prev_ncites + 1)**2)+ I(jj_sim == 0) + np.log10(jj_sim + 1) + I(np.log10(jj_sim + 1)**2) + journal_same+ sink_is_eng + sink_is_journal + sink_is_review + sink_is_case_rep + sink_is_let_ed_com+ np.log10(np.nan_to_num(source_V_novelty) + 1)+ np.log10(np.nan_to_num(sink_V_novelty) + 1) + I(np.log10(np.nan_to_num(sink_V_novelty) + 1)**2) True\n",
      "Created dataframes\n",
      "X.shape = (5019975, 36), y.shape = (5019975, 1)\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.177102\n",
      "         Iterations 11\n",
      "In eval measures function.\n",
      "{'fp': 9012.0, 'auc': 0.83563194754400105, 'rmse': 0.22123413304689701, 'prsquared': 0.19062317281306018, 'recall': 0.020903296074173888, 'precision': 0.3992, 'f_score': 0.03972640001061494, 'bic': 1778649.1271854057, 'tn': 4724501.0, 'df_model': 35.0, 'aic': 1778165.685506989, 'mae': 0.096619038492187823, 'tp': 5988.0, 'fpr': 0.0019038713953040797, 'tpr': 0.020903296074173888, 'llf': -889046.84275349451, 'fn': 280474.0, 'accuracy': 0.9423331789500944}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "        <td>Model:</td>              <td>Logit</td>      <td>Pseudo R-squared:</td>     <td>0.191</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <td>Dependent Variable:</td>   <td>is_self_cite</td>         <td>AIC:</td>        <td>1778165.6855</td>\n",
       "</tr>\n",
       "<tr>\n",
       "         <td>Date:</td>        <td>2016-07-26 12:06</td>       <td>BIC:</td>        <td>1778649.1272</td>\n",
       "</tr>\n",
       "<tr>\n",
       "   <td>No. Observations:</td>       <td>5019975</td>      <td>Log-Likelihood:</td>   <td>-8.8905e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "       <td>Df Model:</td>             <td>35</td>            <td>LL-Null:</td>       <td>-1.0984e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "     <td>Df Residuals:</td>         <td>5019939</td>       <td>LLR p-value:</td>       <td>0.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "      <td>Converged:</td>           <td>1.0000</td>           <td>Scale:</td>          <td>1.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "    <td>No. Iterations:</td>        <td>11.0000</td>             <td></td>                <td></td>      \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "                           <td></td>                            <th>Coef.</th>  <th>Std.Err.</th>     <th>z</th>     <th>P>|z|</th> <th>[0.025</th>  <th>0.975]</th> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>                                           <td>-3.5424</td>  <td>0.4819</td>   <td>-7.3503</td> <td>0.0000</td> <td>-4.4870</td> <td>-2.5978</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 0)[T.True]</th>                    <td>-1.2362</td>  <td>0.0297</td>  <td>-41.6214</td> <td>0.0000</td> <td>-1.2945</td> <td>-1.1780</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 1)[T.True]</th>                    <td>-0.1583</td>  <td>0.0166</td>   <td>-9.5147</td> <td>0.0000</td> <td>-0.1909</td> <td>-0.1257</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.F]</th>                      <td>-0.0333</td>  <td>0.0122</td>   <td>-2.7402</td> <td>0.0061</td> <td>-0.0571</td> <td>-0.0095</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.M]</th>                      <td>-0.0061</td>  <td>0.0117</td>   <td>-0.5164</td> <td>0.6056</td> <td>-0.0290</td> <td>0.0169</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_authors > 20)[T.True]</th>                    <td>-0.0473</td>  <td>0.0339</td>   <td>-1.3960</td> <td>0.1627</td> <td>-0.1137</td> <td>0.0191</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_mesh_ex == 0)[T.True]</th>                    <td>-0.6282</td>  <td>0.0365</td>  <td>-17.2302</td> <td>0.0000</td> <td>-0.6997</td> <td>-0.5568</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_n_mesh_ex == 0)[T.True]</th>                      <td>-0.1561</td>  <td>0.0436</td>   <td>-3.5751</td> <td>0.0004</td> <td>-0.2416</td> <td>-0.0705</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span < 0)[T.True]</th>                            <td>-0.3271</td>  <td>0.0414</td>   <td>-7.9000</td> <td>0.0000</td> <td>-0.4083</td> <td>-0.2460</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span == 0)[T.True]</th>                           <td>0.5802</td>   <td>0.0154</td>   <td>37.6863</td> <td>0.0000</td> <td>0.5500</td>  <td>0.6103</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_prev_ncites == 0)[T.True]</th>                    <td>0.1564</td>   <td>0.0108</td>   <td>14.4536</td> <td>0.0000</td> <td>0.1352</td>  <td>0.1776</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(jj_sim == 0)[T.True]</th>                              <td>-0.1629</td>  <td>0.0220</td>   <td>-7.3919</td> <td>0.0000</td> <td>-0.2060</td> <td>-0.1197</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>journal_same[T.True]</th>                                <td>0.5009</td>   <td>0.0075</td>   <td>66.8221</td> <td>0.0000</td> <td>0.4862</td>  <td>0.5156</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_eng[T.True]</th>                                 <td>3.6246</td>   <td>0.2680</td>   <td>13.5227</td> <td>0.0000</td> <td>3.0992</td>  <td>4.1499</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_journal[T.True]</th>                             <td>0.4962</td>   <td>0.0317</td>   <td>15.6474</td> <td>0.0000</td> <td>0.4341</td>  <td>0.5584</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_review[T.True]</th>                              <td>-0.6393</td>  <td>0.0074</td>  <td>-86.4486</td> <td>0.0000</td> <td>-0.6538</td> <td>-0.6248</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_case_rep[T.True]</th>                            <td>-0.8612</td>  <td>0.0172</td>  <td>-50.1023</td> <td>0.0000</td> <td>-0.8949</td> <td>-0.8275</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_let_ed_com[T.True]</th>                          <td>-0.4512</td>  <td>0.0314</td>  <td>-14.3747</td> <td>0.0000</td> <td>-0.5127</td> <td>-0.3897</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(auth_prev_papers + 1)</th>                      <td>1.9110</td>   <td>0.0200</td>   <td>95.7579</td> <td>0.0000</td> <td>1.8719</td>  <td>1.9501</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(auth_prev_papers + 1) ** 2)</th>              <td>-0.1558</td>  <td>0.0064</td>  <td>-24.2642</td> <td>0.0000</td> <td>-0.1684</td> <td>-0.1432</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_ncites)</th>                             <td>-7.3952</td>  <td>0.8157</td>   <td>-9.0664</td> <td>0.0000</td> <td>-8.9939</td> <td>-5.7965</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 2)</th>                     <td>5.4000</td>   <td>0.5505</td>   <td>9.8097</td>  <td>0.0000</td> <td>4.3211</td>  <td>6.4789</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 3)</th>                     <td>-1.3862</td>  <td>0.1219</td>  <td>-11.3686</td> <td>0.0000</td> <td>-1.6252</td> <td>-1.1472</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.clip(source_n_authors, 0, 20))</th>          <td>0.3029</td>   <td>0.0446</td>   <td>6.7977</td>  <td>0.0000</td> <td>0.2156</td>  <td>0.3902</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)</th>  <td>-0.3527</td>  <td>0.0332</td>  <td>-10.6089</td> <td>0.0000</td> <td>-0.4178</td> <td>-0.2875</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_n_mesh_ex + 1)</th>                      <td>-0.1505</td>  <td>0.0142</td>  <td>-10.6107</td> <td>0.0000</td> <td>-0.1783</td> <td>-0.1227</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_n_mesh_ex + 1)</th>                        <td>-0.2557</td>  <td>0.0147</td>  <td>-17.3982</td> <td>0.0000</td> <td>-0.2845</td> <td>-0.2269</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.score_log_1(year_span)</th>                           <td>0.1516</td>   <td>0.0382</td>   <td>3.9647</td>  <td>0.0001</td> <td>0.0767</td>  <td>0.2265</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(mf.score_log_1(year_span) ** 2)</th>                   <td>-0.8572</td>  <td>0.0231</td>  <td>-37.1872</td> <td>0.0000</td> <td>-0.9024</td> <td>-0.8120</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_prev_ncites + 1)</th>                      <td>-0.1370</td>  <td>0.0164</td>   <td>-8.3709</td> <td>0.0000</td> <td>-0.1690</td> <td>-0.1049</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(sink_prev_ncites + 1) ** 2)</th>              <td>-0.1806</td>  <td>0.0059</td>  <td>-30.4232</td> <td>0.0000</td> <td>-0.1922</td> <td>-0.1690</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(jj_sim + 1)</th>                                <td>0.2548</td>   <td>0.0273</td>   <td>9.3344</td>  <td>0.0000</td> <td>0.2013</td>  <td>0.3083</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(jj_sim + 1) ** 2)</th>                        <td>-0.0556</td>  <td>0.0081</td>   <td>-6.8353</td> <td>0.0000</td> <td>-0.0715</td> <td>-0.0396</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(source_V_novelty) + 1)</th>       <td>-0.1117</td>  <td>0.0036</td>  <td>-31.2724</td> <td>0.0000</td> <td>-0.1187</td> <td>-0.1047</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(sink_V_novelty) + 1)</th>         <td>0.1584</td>   <td>0.0207</td>   <td>7.6332</td>  <td>0.0000</td> <td>0.1177</td>  <td>0.1990</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)</th> <td>-0.0370</td>  <td>0.0035</td>  <td>-10.5280</td> <td>0.0000</td> <td>-0.0439</td> <td>-0.0301</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary2.Summary'>\n",
       "\"\"\"\n",
       "                                           Results: Logit\n",
       "====================================================================================================\n",
       "Model:                         Logit                       Pseudo R-squared:            0.191       \n",
       "Dependent Variable:            is_self_cite                AIC:                         1778165.6855\n",
       "Date:                          2016-07-26 12:06            BIC:                         1778649.1272\n",
       "No. Observations:              5019975                     Log-Likelihood:              -8.8905e+05 \n",
       "Df Model:                      35                          LL-Null:                     -1.0984e+06 \n",
       "Df Residuals:                  5019939                     LLR p-value:                 0.0000      \n",
       "Converged:                     1.0000                      Scale:                       1.0000      \n",
       "No. Iterations:                11.0000                                                              \n",
       "----------------------------------------------------------------------------------------------------\n",
       "                                                     Coef.  Std.Err.    z     P>|z|   [0.025  0.975]\n",
       "----------------------------------------------------------------------------------------------------\n",
       "Intercept                                           -3.5424   0.4819  -7.3503 0.0000 -4.4870 -2.5978\n",
       "I(auth_prev_papers == 0)[T.True]                    -1.2362   0.0297 -41.6214 0.0000 -1.2945 -1.1780\n",
       "I(auth_prev_papers == 1)[T.True]                    -0.1583   0.0166  -9.5147 0.0000 -0.1909 -0.1257\n",
       "C(gender, levels=GENDERS)[T.F]                      -0.0333   0.0122  -2.7402 0.0061 -0.0571 -0.0095\n",
       "C(gender, levels=GENDERS)[T.M]                      -0.0061   0.0117  -0.5164 0.6056 -0.0290  0.0169\n",
       "I(source_n_authors > 20)[T.True]                    -0.0473   0.0339  -1.3960 0.1627 -0.1137  0.0191\n",
       "I(source_n_mesh_ex == 0)[T.True]                    -0.6282   0.0365 -17.2302 0.0000 -0.6997 -0.5568\n",
       "I(sink_n_mesh_ex == 0)[T.True]                      -0.1561   0.0436  -3.5751 0.0004 -0.2416 -0.0705\n",
       "I(year_span < 0)[T.True]                            -0.3271   0.0414  -7.9000 0.0000 -0.4083 -0.2460\n",
       "I(year_span == 0)[T.True]                            0.5802   0.0154  37.6863 0.0000  0.5500  0.6103\n",
       "I(sink_prev_ncites == 0)[T.True]                     0.1564   0.0108  14.4536 0.0000  0.1352  0.1776\n",
       "I(jj_sim == 0)[T.True]                              -0.1629   0.0220  -7.3919 0.0000 -0.2060 -0.1197\n",
       "journal_same[T.True]                                 0.5009   0.0075  66.8221 0.0000  0.4862  0.5156\n",
       "sink_is_eng[T.True]                                  3.6246   0.2680  13.5227 0.0000  3.0992  4.1499\n",
       "sink_is_journal[T.True]                              0.4962   0.0317  15.6474 0.0000  0.4341  0.5584\n",
       "sink_is_review[T.True]                              -0.6393   0.0074 -86.4486 0.0000 -0.6538 -0.6248\n",
       "sink_is_case_rep[T.True]                            -0.8612   0.0172 -50.1023 0.0000 -0.8949 -0.8275\n",
       "sink_is_let_ed_com[T.True]                          -0.4512   0.0314 -14.3747 0.0000 -0.5127 -0.3897\n",
       "np.log10(auth_prev_papers + 1)                       1.9110   0.0200  95.7579 0.0000  1.8719  1.9501\n",
       "I(np.log10(auth_prev_papers + 1) ** 2)              -0.1558   0.0064 -24.2642 0.0000 -0.1684 -0.1432\n",
       "np.log10(source_ncites)                             -7.3952   0.8157  -9.0664 0.0000 -8.9939 -5.7965\n",
       "I(np.log10(source_ncites) ** 2)                      5.4000   0.5505   9.8097 0.0000  4.3211  6.4789\n",
       "I(np.log10(source_ncites) ** 3)                     -1.3862   0.1219 -11.3686 0.0000 -1.6252 -1.1472\n",
       "np.log10(np.clip(source_n_authors, 0, 20))           0.3029   0.0446   6.7977 0.0000  0.2156  0.3902\n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)  -0.3527   0.0332 -10.6089 0.0000 -0.4178 -0.2875\n",
       "np.log10(source_n_mesh_ex + 1)                      -0.1505   0.0142 -10.6107 0.0000 -0.1783 -0.1227\n",
       "np.log10(sink_n_mesh_ex + 1)                        -0.2557   0.0147 -17.3982 0.0000 -0.2845 -0.2269\n",
       "mf.score_log_1(year_span)                            0.1516   0.0382   3.9647 0.0001  0.0767  0.2265\n",
       "I(mf.score_log_1(year_span) ** 2)                   -0.8572   0.0231 -37.1872 0.0000 -0.9024 -0.8120\n",
       "np.log10(sink_prev_ncites + 1)                      -0.1370   0.0164  -8.3709 0.0000 -0.1690 -0.1049\n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)              -0.1806   0.0059 -30.4232 0.0000 -0.1922 -0.1690\n",
       "np.log10(jj_sim + 1)                                 0.2548   0.0273   9.3344 0.0000  0.2013  0.3083\n",
       "I(np.log10(jj_sim + 1) ** 2)                        -0.0556   0.0081  -6.8353 0.0000 -0.0715 -0.0396\n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)       -0.1117   0.0036 -31.2724 0.0000 -0.1187 -0.1047\n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          0.1584   0.0207   7.6332 0.0000  0.1177  0.1990\n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2) -0.0370   0.0035 -10.5280 0.0000 -0.0439 -0.0301\n",
       "====================================================================================================\n",
       "\n",
       "\"\"\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f976e349c90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nECdMxhhTiBMmY4wp9C/dCZ04+FKEIAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f976e349750>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "MODEL_VERSION = \"v4\"\n",
    "MODEL_SUFFIX = \"%s.first_author\" % MODEL_VERSION\n",
    "\n",
    "results_full = fit_model(df_filtered, base_formula, include_prc=True,\n",
    "                            MODEL_VERSION=MODEL_VERSION, MODEL_SUFFIX=MODEL_SUFFIX, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "results_filtered = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" \n",
    "                        % (MODEL_VERSION, MODEL_SUFFIX, version))\n",
    "with open(\"Filtered_models.%s.filtered_%s.txt\" % (MODEL_SUFFIX, version), \"wb+\") as fp: \n",
    "    print_model(results_filtered, fp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Last author"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/content/smishra8/SOFTWARE/anaconda2/lib/python2.7/site-packages/ipykernel/__main__.py:10: UserWarning: Boolean Series key will be reindexed to match DataFrame index.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4849982, 54) (41618369, 54)\n"
     ]
    }
   ],
   "source": [
    "df_filtered_la = filtered_data(df_la, max_source_ncites=50, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Created figure with 1 rows and 1 columns\n",
      "(4849982, 54) is_self_cite ~ I(auth_prev_papers == 0) + I(auth_prev_papers == 1) + np.log10(auth_prev_papers + 1) + I(np.log10(auth_prev_papers + 1)**2)+ C(gender, levels=GENDERS)+ np.log10(source_ncites) + I(np.log10(source_ncites)**2) + I(np.log10(source_ncites)**3)+ I(source_n_authors > 20) + np.log10(np.clip(source_n_authors, 0, 20))+ I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)+ np.log10(source_n_mesh_ex + 1) + I(source_n_mesh_ex == 0)+ np.log10(sink_n_mesh_ex + 1) + I(sink_n_mesh_ex == 0)+ I(year_span < 0) + I(year_span == 0) + mf.score_log_1(year_span) + I(mf.score_log_1(year_span)**2)+ I(sink_prev_ncites == 0) + np.log10(sink_prev_ncites + 1) + I(np.log10(sink_prev_ncites + 1)**2)+ I(jj_sim == 0) + np.log10(jj_sim + 1) + I(np.log10(jj_sim + 1)**2) + journal_same+ sink_is_eng + sink_is_journal + sink_is_review + sink_is_case_rep + sink_is_let_ed_com+ np.log10(np.nan_to_num(source_V_novelty) + 1)+ np.log10(np.nan_to_num(sink_V_novelty) + 1) + I(np.log10(np.nan_to_num(sink_V_novelty) + 1)**2) True\n",
      "Created dataframes\n",
      "X.shape = (4849982, 36), y.shape = (4849982, 1)\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.306621\n",
      "         Iterations 11\n",
      "In eval measures function.\n",
      "{'fp': 4956.0, 'auc': 0.73781916996890062, 'rmse': 0.30018802337589734, 'prsquared': 0.10896487289589485, 'recall': 0.008512153590011194, 'precision': 0.47555555555555556, 'f_score': 0.016724940965870926, 'bic': 2974763.9888064303, 'tn': 4317075.0, 'df_model': 35.0, 'aic': 2974281.7873265739, 'mae': 0.17963184024640519, 'tp': 4494.0, 'fpr': 0.0011466831218933876, 'tpr': 0.008512153590011194, 'llf': -1487104.8936632869, 'fn': 523457.0, 'accuracy': 0.8910484616231565}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "        <td>Model:</td>              <td>Logit</td>      <td>Pseudo R-squared:</td>     <td>0.109</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <td>Dependent Variable:</td>   <td>is_self_cite</td>         <td>AIC:</td>        <td>2974281.7873</td>\n",
       "</tr>\n",
       "<tr>\n",
       "         <td>Date:</td>        <td>2016-07-26 12:10</td>       <td>BIC:</td>        <td>2974763.9888</td>\n",
       "</tr>\n",
       "<tr>\n",
       "   <td>No. Observations:</td>       <td>4849982</td>      <td>Log-Likelihood:</td>   <td>-1.4871e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "       <td>Df Model:</td>             <td>35</td>            <td>LL-Null:</td>       <td>-1.6690e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "     <td>Df Residuals:</td>         <td>4849946</td>       <td>LLR p-value:</td>       <td>0.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "      <td>Converged:</td>           <td>1.0000</td>           <td>Scale:</td>          <td>1.0000</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "    <td>No. Iterations:</td>        <td>11.0000</td>             <td></td>                <td></td>      \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "                           <td></td>                             <th>Coef.</th>  <th>Std.Err.</th>     <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>   <th>0.975]</th> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>                                            <td>-0.4911</td>  <td>0.6950</td>   <td>-0.7065</td>  <td>0.4799</td>  <td>-1.8533</td>  <td>0.8712</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 0)[T.True]</th>                     <td>-1.2363</td>  <td>0.0544</td>  <td>-22.7108</td>  <td>0.0000</td>  <td>-1.3430</td>  <td>-1.1296</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(auth_prev_papers == 1)[T.True]</th>                     <td>-0.1887</td>  <td>0.0349</td>   <td>-5.4125</td>  <td>0.0000</td>  <td>-0.2570</td>  <td>-0.1204</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.F]</th>                       <td>0.0340</td>   <td>0.0108</td>   <td>3.1578</td>   <td>0.0016</td>  <td>0.0129</td>   <td>0.0551</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(gender, levels=GENDERS)[T.M]</th>                       <td>0.0045</td>   <td>0.0102</td>   <td>0.4441</td>   <td>0.6570</td>  <td>-0.0155</td>  <td>0.0246</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_authors > 20)[T.True]</th>                     <td>-0.0372</td>  <td>0.0320</td>   <td>-1.1620</td>  <td>0.2452</td>  <td>-0.0999</td>  <td>0.0255</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(source_n_mesh_ex == 0)[T.True]</th>                     <td>-0.9082</td>  <td>0.0276</td>  <td>-32.8883</td>  <td>0.0000</td>  <td>-0.9623</td>  <td>-0.8540</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_n_mesh_ex == 0)[T.True]</th>                       <td>-0.1062</td>  <td>0.0304</td>   <td>-3.4929</td>  <td>0.0005</td>  <td>-0.1658</td>  <td>-0.0466</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span < 0)[T.True]</th>                             <td>-0.3941</td>  <td>0.0438</td>   <td>-9.0025</td>  <td>0.0000</td>  <td>-0.4799</td>  <td>-0.3083</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(year_span == 0)[T.True]</th>                            <td>0.7484</td>   <td>0.0126</td>   <td>59.5842</td>  <td>0.0000</td>  <td>0.7238</td>   <td>0.7730</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(sink_prev_ncites == 0)[T.True]</th>                     <td>0.2924</td>   <td>0.0087</td>   <td>33.7215</td>  <td>0.0000</td>  <td>0.2754</td>   <td>0.3094</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(jj_sim == 0)[T.True]</th>                               <td>-0.1879</td>  <td>0.0169</td>  <td>-11.1327</td>  <td>0.0000</td>  <td>-0.2209</td>  <td>-0.1548</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>journal_same[T.True]</th>                                 <td>0.6156</td>   <td>0.0054</td>  <td>114.4935</td>  <td>0.0000</td>  <td>0.6051</td>   <td>0.6261</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_eng[T.True]</th>                                  <td>3.9592</td>   <td>0.2677</td>   <td>14.7895</td>  <td>0.0000</td>  <td>3.4345</td>   <td>4.4838</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_journal[T.True]</th>                              <td>0.5745</td>   <td>0.0270</td>   <td>21.3152</td>  <td>0.0000</td>  <td>0.5217</td>   <td>0.6273</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_review[T.True]</th>                               <td>-0.6413</td>  <td>0.0056</td>  <td>-113.6266</td> <td>0.0000</td>  <td>-0.6523</td>  <td>-0.6302</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_case_rep[T.True]</th>                             <td>-1.0461</td>  <td>0.0144</td>  <td>-72.4814</td>  <td>0.0000</td>  <td>-1.0744</td>  <td>-1.0178</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>sink_is_let_ed_com[T.True]</th>                           <td>-0.3730</td>  <td>0.0262</td>  <td>-14.2362</td>  <td>0.0000</td>  <td>-0.4244</td>  <td>-0.3217</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(auth_prev_papers + 1)</th>                       <td>2.4676</td>   <td>0.0225</td>  <td>109.7519</td>  <td>0.0000</td>  <td>2.4235</td>   <td>2.5117</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(auth_prev_papers + 1) ** 2)</th>               <td>-0.4013</td>  <td>0.0061</td>  <td>-65.2533</td>  <td>0.0000</td>  <td>-0.4133</td>  <td>-0.3892</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_ncites)</th>                             <td>-17.4836</td>  <td>1.4198</td>  <td>-12.3139</td>  <td>0.0000</td> <td>-20.2664</td> <td>-14.7008</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 2)</th>                      <td>13.4201</td>  <td>1.0390</td>   <td>12.9169</td>  <td>0.0000</td>  <td>11.3838</td>  <td>15.4564</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(source_ncites) ** 3)</th>                      <td>-3.4219</td>  <td>0.2508</td>  <td>-13.6445</td>  <td>0.0000</td>  <td>-3.9134</td>  <td>-2.9303</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.clip(source_n_authors, 0, 20))</th>           <td>-0.1828</td>  <td>0.0350</td>   <td>-5.2234</td>  <td>0.0000</td>  <td>-0.2514</td>  <td>-0.1142</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)</th>   <td>-0.0873</td>  <td>0.0261</td>   <td>-3.3514</td>  <td>0.0008</td>  <td>-0.1384</td>  <td>-0.0363</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(source_n_mesh_ex + 1)</th>                       <td>-0.2391</td>  <td>0.0108</td>  <td>-22.2102</td>  <td>0.0000</td>  <td>-0.2602</td>  <td>-0.2180</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_n_mesh_ex + 1)</th>                         <td>-0.2299</td>  <td>0.0107</td>  <td>-21.3941</td>  <td>0.0000</td>  <td>-0.2509</td>  <td>-0.2088</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>mf.score_log_1(year_span)</th>                            <td>0.8563</td>   <td>0.0291</td>   <td>29.4237</td>  <td>0.0000</td>  <td>0.7993</td>   <td>0.9133</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(mf.score_log_1(year_span) ** 2)</th>                    <td>-1.1549</td>  <td>0.0171</td>  <td>-67.6205</td>  <td>0.0000</td>  <td>-1.1883</td>  <td>-1.1214</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(sink_prev_ncites + 1)</th>                       <td>0.2151</td>   <td>0.0116</td>   <td>18.6162</td>  <td>0.0000</td>  <td>0.1925</td>   <td>0.2378</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(sink_prev_ncites + 1) ** 2)</th>               <td>-0.2013</td>  <td>0.0038</td>  <td>-52.3889</td>  <td>0.0000</td>  <td>-0.2088</td>  <td>-0.1937</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(jj_sim + 1)</th>                                 <td>0.3580</td>   <td>0.0217</td>   <td>16.5276</td>  <td>0.0000</td>  <td>0.3155</td>   <td>0.4004</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(jj_sim + 1) ** 2)</th>                         <td>-0.1427</td>  <td>0.0066</td>  <td>-21.5469</td>  <td>0.0000</td>  <td>-0.1556</td>  <td>-0.1297</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(source_V_novelty) + 1)</th>        <td>-0.1569</td>  <td>0.0026</td>  <td>-59.9178</td>  <td>0.0000</td>  <td>-0.1620</td>  <td>-0.1517</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>np.log10(np.nan_to_num(sink_V_novelty) + 1)</th>          <td>0.1994</td>   <td>0.0141</td>   <td>14.1080</td>  <td>0.0000</td>  <td>0.1717</td>   <td>0.2271</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)</th>  <td>-0.0491</td>  <td>0.0025</td>  <td>-19.9738</td>  <td>0.0000</td>  <td>-0.0540</td>  <td>-0.0443</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary2.Summary'>\n",
       "\"\"\"\n",
       "                                             Results: Logit\n",
       "========================================================================================================\n",
       "Model:                           Logit                        Pseudo R-squared:             0.109       \n",
       "Dependent Variable:              is_self_cite                 AIC:                          2974281.7873\n",
       "Date:                            2016-07-26 12:10             BIC:                          2974763.9888\n",
       "No. Observations:                4849982                      Log-Likelihood:               -1.4871e+06 \n",
       "Df Model:                        35                           LL-Null:                      -1.6690e+06 \n",
       "Df Residuals:                    4849946                      LLR p-value:                  0.0000      \n",
       "Converged:                       1.0000                       Scale:                        1.0000      \n",
       "No. Iterations:                  11.0000                                                                \n",
       "--------------------------------------------------------------------------------------------------------\n",
       "                                                     Coef.   Std.Err.     z     P>|z|   [0.025   0.975] \n",
       "--------------------------------------------------------------------------------------------------------\n",
       "Intercept                                            -0.4911   0.6950   -0.7065 0.4799  -1.8533   0.8712\n",
       "I(auth_prev_papers == 0)[T.True]                     -1.2363   0.0544  -22.7108 0.0000  -1.3430  -1.1296\n",
       "I(auth_prev_papers == 1)[T.True]                     -0.1887   0.0349   -5.4125 0.0000  -0.2570  -0.1204\n",
       "C(gender, levels=GENDERS)[T.F]                        0.0340   0.0108    3.1578 0.0016   0.0129   0.0551\n",
       "C(gender, levels=GENDERS)[T.M]                        0.0045   0.0102    0.4441 0.6570  -0.0155   0.0246\n",
       "I(source_n_authors > 20)[T.True]                     -0.0372   0.0320   -1.1620 0.2452  -0.0999   0.0255\n",
       "I(source_n_mesh_ex == 0)[T.True]                     -0.9082   0.0276  -32.8883 0.0000  -0.9623  -0.8540\n",
       "I(sink_n_mesh_ex == 0)[T.True]                       -0.1062   0.0304   -3.4929 0.0005  -0.1658  -0.0466\n",
       "I(year_span < 0)[T.True]                             -0.3941   0.0438   -9.0025 0.0000  -0.4799  -0.3083\n",
       "I(year_span == 0)[T.True]                             0.7484   0.0126   59.5842 0.0000   0.7238   0.7730\n",
       "I(sink_prev_ncites == 0)[T.True]                      0.2924   0.0087   33.7215 0.0000   0.2754   0.3094\n",
       "I(jj_sim == 0)[T.True]                               -0.1879   0.0169  -11.1327 0.0000  -0.2209  -0.1548\n",
       "journal_same[T.True]                                  0.6156   0.0054  114.4935 0.0000   0.6051   0.6261\n",
       "sink_is_eng[T.True]                                   3.9592   0.2677   14.7895 0.0000   3.4345   4.4838\n",
       "sink_is_journal[T.True]                               0.5745   0.0270   21.3152 0.0000   0.5217   0.6273\n",
       "sink_is_review[T.True]                               -0.6413   0.0056 -113.6266 0.0000  -0.6523  -0.6302\n",
       "sink_is_case_rep[T.True]                             -1.0461   0.0144  -72.4814 0.0000  -1.0744  -1.0178\n",
       "sink_is_let_ed_com[T.True]                           -0.3730   0.0262  -14.2362 0.0000  -0.4244  -0.3217\n",
       "np.log10(auth_prev_papers + 1)                        2.4676   0.0225  109.7519 0.0000   2.4235   2.5117\n",
       "I(np.log10(auth_prev_papers + 1) ** 2)               -0.4013   0.0061  -65.2533 0.0000  -0.4133  -0.3892\n",
       "np.log10(source_ncites)                             -17.4836   1.4198  -12.3139 0.0000 -20.2664 -14.7008\n",
       "I(np.log10(source_ncites) ** 2)                      13.4201   1.0390   12.9169 0.0000  11.3838  15.4564\n",
       "I(np.log10(source_ncites) ** 3)                      -3.4219   0.2508  -13.6445 0.0000  -3.9134  -2.9303\n",
       "np.log10(np.clip(source_n_authors, 0, 20))           -0.1828   0.0350   -5.2234 0.0000  -0.2514  -0.1142\n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   -0.0873   0.0261   -3.3514 0.0008  -0.1384  -0.0363\n",
       "np.log10(source_n_mesh_ex + 1)                       -0.2391   0.0108  -22.2102 0.0000  -0.2602  -0.2180\n",
       "np.log10(sink_n_mesh_ex + 1)                         -0.2299   0.0107  -21.3941 0.0000  -0.2509  -0.2088\n",
       "mf.score_log_1(year_span)                             0.8563   0.0291   29.4237 0.0000   0.7993   0.9133\n",
       "I(mf.score_log_1(year_span) ** 2)                    -1.1549   0.0171  -67.6205 0.0000  -1.1883  -1.1214\n",
       "np.log10(sink_prev_ncites + 1)                        0.2151   0.0116   18.6162 0.0000   0.1925   0.2378\n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)               -0.2013   0.0038  -52.3889 0.0000  -0.2088  -0.1937\n",
       "np.log10(jj_sim + 1)                                  0.3580   0.0217   16.5276 0.0000   0.3155   0.4004\n",
       "I(np.log10(jj_sim + 1) ** 2)                         -0.1427   0.0066  -21.5469 0.0000  -0.1556  -0.1297\n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)        -0.1569   0.0026  -59.9178 0.0000  -0.1620  -0.1517\n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)           0.1994   0.0141   14.1080 0.0000   0.1717   0.2271\n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)  -0.0491   0.0025  -19.9738 0.0000  -0.0540  -0.0443\n",
       "========================================================================================================\n",
       "\n",
       "\"\"\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f976a2e6f90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUwAAAFMCAYAAACgboVfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3XdUFNfbB/DvUlSkBFGxUhIbURArLXYQESxoxIKKSlRA\nsITELolRlNiwUCSIBBBMiA27IiDYMGBMjBHsIiBKEVRABIR5/+DHvK4Ud2F3Zxeezzmew9ydnf2O\nug9T7tzLYxiGASGEkE+S4zoAIYTICiqYhBAiICqYhBAiICqYhBAiICqYhBAiICqYhBAiIE4LJsMw\n2LNnD7788kv4+vp+cv3bt29j9uzZMDQ0hImJCdavX493795JICkhhHBYMAsKCjB//nycPn0a8vLy\nn1w/NzcXjo6O6Nq1K44cOYJdu3bh2rVr8PDwEGmuyZMnY/LkySLdpqTRPkgH2gfpIar94Kxgnjhx\nAoqKijhy5Ajk5D4d48CBA2jRogU2btyI7t27w8TEBCtXrsSpU6eQmZkpgcSEkOaOs4JpYWGBgIAA\nqKioCLT+9evXYWRkBAUFBbbNzMwMAJCYmCiWjIQQ8iHOCmaXLl2EWv/p06c13qOkpIS2bdsiLS1N\nhMkIIaR2MnOXvLi4GEpKSjXaW7dujaKiIg4SEUKaG4VPr9J01XYR+MGDB+jcuTOePn0q0M0oaVRW\nVgYAMn1tl/ZBOjSFfQCA0tJSPH36FLa2tjXumRw9elTg7chMwVRVVa31SLKwsBCqqqoi+5zOnTsj\nLS0NlpaWItsmV8zNzbmO0Gi0D9KhKewDAJSUlEBZWbnB75eZgqmjo4OMjAy+tjdv3qCgoADdunVr\n0DZr+83y9OlTWFpaIiIiAh07dmzQdglpasrKyvD48WPcv38f9+/fx7179/DgwQOkp6dD1CNEKioq\nolWrVlBSUkKrVq2goKAABQUFyMvLQ1FREfLy8uyyvLx8vb1s3rx5g4cPH2LAgAF49uwZAgMDoaOj\n0+BsMlMwhw4dirCwMJSVlaFFixYAgPj4eMjLy2PIkCEi+5zq0/COHTuia9euItsuIbKCYRg8evQI\n169fZ//cunUL79+/F2o7HTp0gLa2Njp27Ii2bdtCQ0OD7091m6qqKpSVlaGsrIzWrVtDUVFRJPuR\nlJSEcePG4fDhw9DT04O5uXmjL7NxVjBfv36N8vJy9rfT27dvkZeXBwDQ0NDAzp07kZKSgv379wMA\nZs6ciYiICKxZswaLFy/GixcvsGPHDkyfPh3t27fnajcIkXkMw+C///5DTEwM4uLikJiYiJcvX37y\nfS1btoSenh709fXRo0cP6OjoQFtbG9ra2ujatStatWolgfS1u3XrFsaPH4/g4GBYWFiI7BosZwXT\nzc0NN27cYJd//fVXBAcHg8fjITY2Fnl5eXw7qa6ujpCQEHh6emLixIlQUVHBhAkT4O7uzkV8QmRa\nZmYmYmJicOHCBcTGxiI7O7ve9bt3744BAwagT58+0NfXR58+fdCtWze+ftHSIjU1FVZWVvD19cW4\nceNEum0eTVHBLzMzE+bm5oiNjaVTctKkPHnyBIcPH8ahQ4eQnJxc53pqamowNjaGiYkJTExMYGRk\nhHbt2kkwaeP8+uuvUFBQwOzZs9k2UX2vpe/XAyFEZB49esQWyb/++qvWdVq1aoWhQ4di9OjRsLCw\ngKGhoUCPK0urefPmiW3bVDAJaWIePHiAQ4cO4dChQ/jnn39qXUdPTw+2trYYPXo0zMzMOL3eKEuo\nYBLSBLx+/RqRkZH49ddfcf369VrX0dfXx5QpU2BnZ4fevXtLOGHTQAWTEBnFMAzi4+MRFBSEo0eP\n1jo2bN++fWFnZ4cpU6ZAT0+Pg5TiVVBQgGfPnkFfX18in0cFkxAZ8+bNG4SFhcHf3x+pqak1Xu/V\nqxdmz54NOzs79OzZk4OEklFYWIixY8fCwsICnp6eEvlMKpiEyIjnz59j9+7d2Lt3L968ecP3mpqa\nGmbMmIG5c+fC2NgYPB6Po5SS8fbtW4wbNw79+vXDxo0bJfa5VDAJkXL37t3D9u3b2SfdPmRsbIxF\nixbBzs6u1tG8mqLS0lJMmjQJOjo68Pf3l+gvByqYhEipGzduYPPmzYiKiuJ7XltBQQEzZszA0qVL\nMXDgQA4TSh7DMJg+fTo+++wzBAcHS7z7ExVMQqTM7du34eHhgePHj/O1KysrY+HChVi2bBm0tbU5\nSsctHo8HFxcXjBgxgpOnjKhgEiIl7t+/jx9//BGRkZF8R5Tt27fH0qVL4eLiAg0NDQ4TSgcuh16k\ngkkIx9LS0rBhwwaEhoaisrKSbdfU1MTq1avh5OTUbK5PSjsqmIRwJCsrC56enggKCkJ5eTnb3qZN\nG6xYsQJubm4CTxJIJIMKJiESVlxcDC8vL+zYsYOvs7mKigrc3d3x7bffQl1dncOE0sPLywuWlpZS\nc3OLCiYhEsIwDA4fPozvvvuOb/YAJSUlLF68GMuXL5epUYHEzcvLCwcOHMA333zDdRQWFUxCJODh\nw4dwdXVFdHQ02yYvLw9nZ2esXbsWnTp14jCd9Nm9ezf279+PS5cuQVNTk+s4LCqYhIhRaWkptm3b\nBk9PT5SWlrLtI0eOhI+PD/r06cNhOukUGBgIb29vXLp0CZ07d+Y6Dh8qmISISXx8PFxcXHD37l22\nrVOnTti1axfs7Oya/OOLDZGWloZNmzYhNja2UZOViQsVTEJErKCgAN9++y1CQ0PZNh6PBzc3N3h6\nekJNTY3DdNJNV1cXKSkpjZoKV5yoYBIiQqdPn8bChQuRlZXFtg0YMAC//PILBg0axGEy2SGtxRIA\nZHccekKkyOvXr+Ho6Ihx48axxVJFRQW7d+9GUlISFcsmgo4wCWmk6OhofPPNN3yznI4aNQrBwcFS\neR1Omrx7906mpsegI0xCGqiwsBDOzs4YM2YMWyxbt24NPz8/XLhwgYrlJyQnJ6NPnz4oKiriOorA\n6AiTkAa4ePEiHB0dkZaWxrYNHToUv/76K7p168ZdMBnx77//Yty4cQgKCpKpxz/pCJMQIZSVlWH5\n8uUYNWoUWyxbtWqFnTt3Ij4+noqlAO7evQsrKyv4+Phg/PjxXMcRCh1hEiKgx48fY/r06UhOTmbb\nTE1NERIS0qTnzhGlR48eYfTo0fDy8sLUqVO5jiM0OsIkRAB//PEH+vfvzxZLBQUFeHl54fLly1Qs\nhXDnzh2sXbsWc+bM4TpKg9ARJiH1ePv2LZYtW4Z9+/axbbq6uvjtt99gYmLCYTLZNGHCBK4jNAoV\nTELq8PjxY0yaNAn//vsv2/b1118jKCiIhl9rpuiUnJBanD9/HoMGDWKLZcuWLbF3714cOnSIimUz\nRgWTkA8wDIMtW7bA2toaBQUFAIAvvvgCf/75J5ydnWnADCEUFhYiKSmJ6xgiRafkhPxPUVER5s2b\nh8OHD7NtVlZWOHjwINq0acNhMtnz9u1bjB8/Hn379oWRkRHXcUSGjjAJQdUAvyYmJnzFcs2aNTh1\n6hQVSyGVlpZi0qRJ0NbWxq5du7iOI1J0hEmavQsXLmDq1Kl49eoVgKrRckJDQ/H1119znEz2lJeX\nY9q0aVBTU0NwcDDk5JrWMVnT2htChLRv3z6MHTuWLZY9evTAn3/+ScWygb755htUVFQgIiICCgpN\n73is6e0RIQKorKzEmjVrsGXLFrbNysoKv/32G90Fb4SFCxdi0KBBaNGiBddRxIIKJml2SkpK4ODg\nwHe9ctGiRdi9e3eTPCqSpCFDhnAdQazofwdpVnJycjBhwgT8+eefAKqmjvD29sbSpUupyxD5JCqY\npNl4+PAhrKys8OjRIwBV84EfPHgQtra2HCcjsoLzmz4hISGwsLCAgYEBbGxscPr06XrXT0xMxMyZ\nMzF48GAMHDgQrq6uePr0qYTSElmVlJQEU1NTtlh26NABly5domLZCDt27EBcXBzXMSSK04IZEREB\nb29vuLm54eTJk5g6dSqWL1+Oq1ev1rr+f//9hwULFqBv3744dOgQwsPD2c7GJSUlEk5PZMWpU6cw\ncuRI5OXlAQB69eqF69ev0zw7jbBnzx7s3bsXenp6XEeRLIZDw4YNY7y8vPjaXF1dmVmzZtW6/pYt\nWxhjY2O+ttTUVKZXr17MpUuXRJIpIyOD6dmzJ5ORkSGS7RFuBQYGMnJycgwABgBjZmbG5OXlcR1L\npgUGBjLa2tpMWloa11EEJqrvNWdHmI8ePUJ2djbMzMz42s3MzHDz5k2UlZXVeA+Px6vREVZRUZF9\njZBqDMNgw4YNWLhwISorKwEAtra2iImJQdu2bTlOJ7vCw8Px008/ITY2tlnOWcRZwUxPTwePx0OX\nLl342rW0tFBZWYmMjIwa75k8eTJKSkoQHByM0tJSlJSUwN/fH59//jmNTUhYlZWVWLZsGX788Ue2\nbdGiRTh8+DCUlJQ4TCbbcnNz8cMPPyA6Ohrdu3fnOg4nOLtLXlxcDKBqlr0PVS/XNpNct27d4Ofn\nhyVLlmD79u0AgM8//xxBQUHUf44AqHo0z9HREeHh4Wybp6cn1qxZQ2chjdS+fXukpKTI1LS4osb5\nXXJh3L9/H+7u7pg8eTL++OMPhIaGonPnznBycmILMGm+3r17h6+//potljweD3v37sXatWupWIpI\ncy6WAIdHmKqqqgBqHklWL9c29aavry+0tLSwZs0atu3LL7/EkCFDcOTIETg4OAiVYfLkyTXaart2\nSqRfSUkJJk6ciAsXLgComnMnPDwc06ZN4zgZkSbOzs41Hts8evSowO/n7AhTR0cHDMMgPT2drz0t\nLQ0KCgrQ1tau8Z7Hjx/jiy++4GtTUVGBhoYG9cVsxt6+fYtx48axxVJJSQknTpygYtlIb9++5TqC\n1OHsCFNXVxdaWlq4fPkyzM3N2faEhASYmpqyd78/1KFDB3Yu6GqFhYXIzc1Fhw4dhM5Q22+WzMxM\nvjxEuhUVFWH8+PGIj48HUHUNvLrfJWm4f//9F+PGjcNff/2F9u3bcx1HZAICAtC1a9cGv5/Ta5iu\nrq44cuQIoqKikJWVhcDAQCQlJcHV1RVA1ZME33zzDbv+7Nmz8e+//2LXrl14/PgxUlNTsWrVKigo\nKGDs2LFc7QbhSGFhIaytrdliqaysjLNnz1KxbKS7d+/CysoK27Zta1LFUhQ4vbVsa2uLkpIS+Pr6\nIicnB7q6uvDz84OhoSEAIC8vD5mZmez6I0aMgI+PD/z8/BAcHAxFRUX07dsXoaGh0NLS4mo3CAfe\nvHmDsWPH4tq1awCqromfPXsWX331FcfJZNujR48wevRoeHl50SWN2oikG30TQk/6SL+CggLG2NiY\nfXpHTU2NSUxM5DqWzEtPT2d0dXUZf39/rqOIHKdP+rx+/Rrv378XbeUmRACvXr3C6NGj2eHZ1NXV\nERMTQw8uiEBaWhqWLVsGFxcXrqNILYELZnJyMhwcHGBoaAhTU1M8fPgQABAVFYXIyEixBSSkWmFh\nIcaOHYsbN24AANq0aYPY2FgMHjyY42RNw9ChQ7F06VKuY0g1gQrmtWvXMHfuXGRnZ8POzg4Mw7Cv\n5eTkYP369Thz5ozYQhJSPW3r9evXAQAaGhqIi4vDgAEDOE5GmhOBCqavry+GDx+Os2fPYt26dXwF\nc+HChZg5cyaCgoLEFpI0b9XTtiYkJAAA1NTUEB0djX79+nGcjDQ3AhXM1NRU2Nvb1zllpqWlJXuK\nTogolZeXY+rUqYiOjgbw/12HBg4cyHEy2VZSUoKLFy9yHUPmCNyt6MOjyo+VlJTU2tGckMaoqKjA\nrFmzcOLECQBVzzGfPHmyxpCARDilpaWYPHkyOnbsSH1WhSTQEaahoSGCg4Nrfc66qKgIe/bsYftO\nEiIKlZWVmD9/Pv744w8AVeOeHjt2jL7gjVReXo7p06dDWVkZ+/bt4zqOzBHoCNPNzQ1z586FtbU1\nhg8fDh6Ph9DQULx//x7x8fEoLS1FWFiYuLOSZoJhGHz//fcICQkBAMjLy+OPP/6AlZUVt8FkXEVF\nBRwcHFBWVoZjx47RkIgNINAR5qBBg3Dw4EHo6uri0KFDYBgGx44dYy+8h4eH0wV4IjJbt27Fzp07\nAVQN0XbgwAGarEwEli5ditzcXBw5cqTGiD1EMAL/iunbty+CgoLw/v17vHr1CkBVPzh5eXkwDIP3\n79/TbyzSaPv378eqVavYZV9fX8yYMYPDRE2Ho6MjevXq1ezHtGwMgY4wzc3N8eDBAwBV4wy2a9cO\n7dq1g7y8PADg3LlzNMIPabSoqCgsXLiQXf7xxx+xaNEiDhM1LQMGDICysjLXMWRavYeEWVlZAIBn\nz54hKyur1r/siooK/PXXX8jPzxdPQtIsXLp0CdOnT2cnLHNxceGbk4cQaVBvwaw+auTxeHB2dq5z\nPYZh6PE00mApKSmYOHEiSktLAQBTp06Fj48PTStBpE69BfPatWu4ceMGFi9ejKlTp0JTU7PW9TQ1\nNWFtbS2WgKRpe/78OcaOHcteFx85ciTCwsLYyz2kYfz8/NCxY0d8/fXXXEdpUuotmG3atMHo0aPh\n5uaGadOm1TmYaGFhIdLT09G7d2+xhCRNU2FhIWxsbNhpSvT19XHs2DG0bNmS42Sybf/+/di6dSv7\nKCkRHYFu+ri5udU78nJycjLmzJkjslCk6at+5PHvv/8GAHTp0gVnzpzBZ599xnEy2Xbw4EH88MMP\niImJga6uLtdxmhyB+wH9/vvvOH36NF68eMFemAeqnsjIycmhoeyJwBiGgYuLC86dOwegarT0M2fO\n0Kj5jXTs2DF89913iImJQY8ePbiO0yQJdIR54MABrF+/Hunp6dDU1MSzZ8+grq4OBQUFPH/+HFZW\nVvDz8xN3VtJEbN68Gfv37wdQ1U3t6NGj6Nu3L8epZFtRURFWrVqFM2fOoE+fPlzHabIEOsKMjIyE\nvb09PDw8wOPxoKenh02bNkFPTw8JCQnw9vZGu3btxJ2VNAFHjhzBunXr2OX9+/fDwsKCw0RNg4qK\nCm7fvk1P8IiZQEeYmZmZsLa25uvmUX1aPnz4cIwfPx6bNm0ST0LSZPz9999wcHBgl9evX8+3TBqH\niqX4CVQwFRUVUVJSwi6rqakhNzeXXR4wYAA7xwohtXnx4gUmTpyIt2/fAgCmTZuGH374geNUhAhH\n4ME39uzZgzt37gAAvvjiC3bYLaBq0ndC6vLu3TtMmjQJGRkZAICBAwciODiYOqY3QmFhIdcRmiWB\nCubSpUuRnp4OHx8fAMCUKVMQGxuLYcOGYfLkydi6dSuGDh0q1qBENjEMg4ULF7Jz8XTq1AnHjx9H\n69atOU4mu+7du4cvv/wST5484TpKsyPQTR89PT2cPn2aPUKYMmUKioqKcOLECbx//x729vY02xyp\n1bZt23DgwAEAQMuWLREVFYUuXbpwnEp2PX78GBYWFvD09MTnn3/OdZxmR+B+mNUjFFWbO3cu5s6d\nK45MpIk4ceIE31BtwcHBMDIy4jCRbMvIyIC5uTlWr15N3z2OfPKUvKKiAi9fvvzkhmJiYkQSiDQN\nt2/fxsyZM9m5oNasWQN7e3uOU8muFy9ewNzcHG5ubjTkHYfqLZj+/v4wMjLCkCFDMHDgQOzZs6fG\nOvfv38fcuXOxePFisYUksuXVq1ewtbVFUVERAGDixInYuHEjx6lkW05ODpycnPDdd99xHaVZq/OU\n/NixY9izZw/69++Pfv364d69e9i7dy+UlJSwYMECFBQUYPfu3Th8+DDk5eXh6OgoydxESjEMA0dH\nRzx+/BgAYGBggPDw8DqnaCaC6du3Lz0NJQXqLJiRkZGwtLTkO6rcvXs3Dh48CEVFRfj7+6OoqAjj\nx4/HsmXL0KlTJ4kEJtJt9+7dOHbsGICq/rpHjx6FiooKx6kIEY06f+3fu3cPkyZN4mubPn06nj9/\nji1btsDAwABHjx7Fli1bqFgSAMD169exfPlydjk4OBjdu3fnMBEholVnwSwpKakxAlGHDh0AAFu2\nbMH+/fuhp6cn3nREZrx8+RLTpk3D+/fvAVT13aXBaxumtLQUp06d4joGqUW93YrqehKjZ8+eYglD\nZFNlZSUcHBzYgYCNjIywdetWjlPJpvLyckyfPh3y8vKwsbGhp6GkDM2LSxpt27ZtOHPmDICqUfoj\nIyNpIIgGqKiowJw5c1BWVoZjx45RsZRCDTrCpH9IUu3SpUtYu3YtuxwWFkYjfTdAZWUlFi5ciOzs\nbJw6dYp+4Uipegumh4dHjal1eTwe1q5dW+NZYB6Ph9DQUNEnJFIrJycH06dPR0VFBQBgxYoVGDdu\nHMepZJOHhwfu3r2L8+fPQ0lJies4pA51FszOnTujoKAABQUFfO2dOnVCfn4+zUPezFVUVGDmzJl4\n/vw5AGDIkCHw9PTkOJXscnBwwIoVK6gLlpSrs2DGxcVJMgeRMZs2bWIfh23Xrh1+//13KCoqcpxK\ndvXq1YvrCEQA9PgFEVpsbCzWr18PoOpSTEREBI1ARJoFKphEKFlZWbC3t2cH1Vi3bh0sLS05TkWI\nZHBeMENCQmBhYQEDAwPY2Njg9OnT9a5fVFQEDw8PGBsbY8CAAZg/fz47TicRr4qKCtjb2yMnJwcA\nMHLkSPz4448cp5I9wcHBCAwM5DoGaQBOC2ZERAS8vb3h5uaGkydPYurUqVi+fDmuXr1a53tcXFzw\n9OlThIWF4eDBgyguLoaLi4sEUzdfW7duRUJCAgCgY8eOOHjwIOTl5TlOJVt+++03eHh4YMSIEVxH\nIQ3Aacf1wMBA2Nvbw9bWFgCgq6uL5ORkBAQE4Kuvvqqx/uXLl/Hff//h4sWLUFdXBwBs374dd+7c\nQVlZGfVdE6ObN2/yTVoWERGBjh07cphI9hw7dgzffvstYmJi6Gk5GcXZEeajR4+QnZ0NMzMzvnYz\nMzPcvHkTZWVlNd5z8eJFGBsbs8USALp06QJLS0sqlmJUUlKCWbNmsc+Ju7u7Y9SoURynki3nzp2D\nk5MTzpw5A319fa7jkAYSuGA+efIEq1evhrW1NQYNGoS7d+8CqCpi1adpwkhPTwePx6txd1VLSwuV\nlZW1Xpe8f/8+dHV1sW/fPowZMwampqZwd3enPqFitmrVKqSmpgKoGt+S5qAXTnl5OVatWoXjx49j\nwIABXMchjSBQwbxz5w6+/vprxMXFQVdXF8XFxexrN27cwKJFi3Dt2jWhPrh6Gx8/MVS9XD1a94fy\n8/Nx9uxZ3L9/H97e3vDy8sKtW7fg4OCAyspKoT6fCObChQvsmKgtWrRAeHg4WrVqxXEq2aKoqIgb\nN27A1NSU6yikkQS6hunt7Q09PT0EBgZCRUWFb1i35cuX48WLF/D3969xei1q79+/R6tWrbB161b2\nefaWLVti3rx5uHLlCoYNGybU9iZPnlyjrbZLAc1Vfn4+32Rbnp6eNOp3Ayko0Dg30sDZ2bnG5buj\nR48K/H6BjjBv3bqFBQsW1PnY1pQpU3Dnzh2BPxQAVFVVAdQ8kqxeru2zlJWV0atXL77BPwYOHAge\nj4d79+4J9fmkfgzDwMXFBVlZWQCA4cOHw93dneNUhHBLoF975eXlIh8QQEdHBwzDID09HT169GDb\n09LSoKCgAG1t7Vrf8/Gz7ZWVlWAYpkHP4Nb2myUzMxPm5uZCb6upiYiIwB9//AGgaqqJ0NBQ6kIk\noFevXvHdmCTSIyAgAF27dm3w+wU6wvzyyy/ZL8/HKisrERQUJPSzsLq6utDS0sLly5f52hMSEmBq\nalrrc8lDhw7FrVu38OrVK7bt5s2b4PF49CyuCKWnp8PV1ZVd9vX1hY6ODoeJZMeTJ0/Qt29f3Lp1\ni+soRAwEKpgLFizA2bNnMWPGDAQHB4PH4+HcuXPYtWsXxo4di8TERDg7Owv94a6urjhy5AiioqKQ\nlZWFwMBAJCUlsV/WHTt24JtvvmHXHz9+PDp16oQlS5bg4cOH+PPPP/HTTz9h4MCBdPdRRCorKzF3\n7ly8efMGAGBnZ4dZs2ZxnEo2VJ+drFq1CoaGhlzHIeLACOjChQuMtbU106tXL74/Y8eOZS5cuCDo\nZmo4ePAgY25uzhgYGDDjx49n4uPj2ddWrVrFWFpa8q3/4sULxs3NjRkwYAAzaNAgZtWqVUxhYWGD\nP/9jGRkZTM+ePZmMjAyRbVOW+Pv7MwAYAEynTp2Yly9fch1JJrx48YLp2bMns23bNq6jkFqI6nvN\nY5j/jaIgoBcvXiA7OxtA1eNx1ROjNRXVRwmxsbGNutYhi54+fQp9fX32xtvp06dhbW3NcSrp9/Ll\nS4wYMQJ2dnZ8T0MR6SGq77VAN31cXFwwfvx4jBo1Ch07dqRH4poghmGwYMECtlg6ODhQsRTQmzdv\nMHv2bL4phknTJFDB/Oeff3Dx4kUoKSlh1KhRsLGxwdChQ2nA2CZk//79uHDhAoCqM4edO3dynEh2\nfP7551ixYgXXMYgECFQwr169iuTkZERHRyMmJganT5+GmpoaLC0tYWNjAxMTE5oYTYZlZmbiu+++\nY5cDAgKgoaHBYSJCpJNABVNOTg7GxsYwNjaGh4cH/vnnH5w/fx4xMTE4fPgw2rVrBysrK6xbt07c\neYmIMQyDhQsXsnfFZ8yYgYkTJ3KcihDp1KDRivr164eVK1fiwoUL8Pb2hpKSEiIiIkSdjUhAWFgY\nzp49CwDQ1NRknxsntXv//j0iIyMh5L1S0kQI/YArwzBITk7GhQsXEBsbi+fPn0NNTQ1TpkwRRz4i\nRllZWVi2bBm77Ofnh3bt2nGYSLpVVFRgzpw5yM/Px9dff03PhzdDAv2Ll5aW4sqVK4iJicHFixfx\n+vVrqKqqwtzcHNbW1jAzM6PH5mQM879nxaufmpoyZQr90qtHZWUlnJyc8Pz5c5w+fZqKZTMl0L+6\nsbExSktLoayszFck6S657Prtt99w4sQJAEDbtm3h5+fHcSLpxTAMli5dipSUFERHR4t8XAUiOwQq\nmNVFcujQoTSyeROQnZ2NxYsXs8s+Pj7Q1NTkMJF027p1KxITExEbG9ugQV5I0yFQwdyxY4e4cxAJ\nWrJkCTtpCmi8AAAgAElEQVRK/cSJEzF9+nSOE0k3e3t7zJ8/H5999hnXUQjH6iyY5ubmCAgIQI8e\nPQQa7ozH4yEmJkak4YjonTlzhh15Sl1dHXv37qU+tJ+gpaXFdQQiJeosmEZGRlBWVgYADB48mL5U\nTUBRUREWLVrELm/btg2dOnXiMBEhsqXOgunl5cX+/PPPP9e7kcLCQhQWFoouFRGLH3/8EU+fPgVQ\nNbaoo6Mjx4kIkS0CDyCckpJS5+uJiYnUJUXK3bx5E7t27QJQNSnXL7/8Ajk5zmZZllq///47Nm/e\nzHUMIqXqvemTnJwMoKpbxZ07d/hmi6xWUVGB8+fP1/oakQ7v37/HwoUL2Zk1V69ejS+//JLjVNIn\nKioKy5Yto2vxpE71FsxFixahqKgIPB6v3nH+GIaBhYWFyMMR0fD19cVff/0FAOjZsydWr17NcSLp\nc+7cOSxcuBBnz56Fvr4+13GIlKq3YCYlJSE1NRWTJ0+Gm5sbunTpUmMdHo+H9u3bw8TERGwhScOl\np6fzDYryyy+/0LziH4mPj4eDgwOioqIwcOBAruMQKVZvweTxeOjduze8vLwwatQo6ocmYxiGgZub\nG3u5ZN68eRgxYgS3oaRMZWUlPDw8EBkZCTMzM67jEClXZ8HMysqCpqYmFBQUYGxsjOLi4k9ep+zc\nubPIA5KGO378OE6ePAkAaNeuHbZt28ZxIukjJyeHhIQEugFGBFJvx/XDhw+jT58+GDVqlED9MFNT\nU0UajjRcSUkJ30hE3t7eaNu2LYeJpBcVSyKoOgvm5s2b2cmCNm/eTB3XZcyWLVvYPpfDhg2jqXIJ\nEYE6C+akSZPYnydPniyRMEQ0njx5gi1btgCoOnry8fGhX3j/k5eXh7Zt29LfB2kQgc9FTp06hevX\nr7PLV69exaxZszB+/Hjs27dPLOFIw7i7u+Pdu3cAqrqG9e3bl+NE0iEzMxNGRkZISEjgOgqRUQIV\nzMjISKxYsQKZmZkAgIyMDDg7OyMzMxNdu3aFj48PwsPDxRqUCOb8+fOIiooCUHWjZ8OGDRwnkg7Z\n2dmwsLCAi4sL9RQgDSbQ8G4RERGYP38++/hjZGQkeDweIiMj0aFDBwQHB+OPP/6g62QcKysrw5Il\nS9hlLy8vtGnThsNE0uHly5cYPXo0ZsyYQXOHk0YR6AgzIyMDw4YNY5fj4uJgZmaGDh06AAD69+/P\nHn0S7uzevRv3798HUDXCFA2uAbx+/RpWVlYYO3ZsvU+rESIIgQpmy5YtUV5eDqCqf+bjx48xcuRI\n9vV3797RRXSOZWVl8Z1++/j4UHcZVM1HZWdnh59//pn+j5JGE+iUXE9PDxEREWjbti327t2Lli1b\nYvTo0ezrFy5cwBdffCG2kOTTVqxYgaKiIgCAo6MjjI2NOU4kHTQ1NbFixQquY5AmQqCCuXjxYixY\nsACxsbFgGAbffvstNDQ0AAA7d+7EwYMHsX37drEGJXW7fPkyOy/8Z599xjeWKSFEdAQqmAMHDsS5\nc+dw69YttG3bFgMGDGBf69WrF/bs2QNLS0uxhSR1q6io4LvRs2HDBprQjBAxEXhyZU1NTb7T8GrW\n1tYiDUSEEx4ejn/++QcAoK+vzzcFRXNTUVGBAwcOwMHBga7fErEQuGA+ePAA+/fvx40bN5CTkwM5\nOTl06NABpqammD9/PvsYJZGct2/fYu3atezyjh07oKAg8D9pk1JZWQlnZ2c8evQI06dPpyHsiFgI\n9O26desWHBwcwOPxYGhoiH79+oFhGOTk5ODo0aM4c+YMfvvtN3Tr1k3ceckHdu7ciWfPngEALC0t\nm+1lEYZhsGzZMty5cwfR0dFULInYCFQw9+zZAwMDA/j5+dUYEzM/Px8uLi7YuXMnfH19xRKS1JSd\nnc1OTsfj8Zrt0G0Mw2D16tW4evUqYmNjoaKiwnUk0oQJdKHn1q1bcHR0rHUAYQ0NDTg6OiIpKUnk\n4Ujd1q9fz3YjmjdvXrN9XjwwMBCnT59GdHQ01NXVuY5DmjiBCmZZWRmUlJTqfF1dXZ0d7IGIX2pq\nKjvgSevWrZv18+J2dna4cOECjfVJJEKggqmjo4P4+Pg6X4+Li4OOjo6oMpFPWLlyJSoqKgAA33//\nfa1zLTUXGhoa6NixI9cxSDMh0DVMe3t7/PTTT3j+/DlGjRrFPkP+4sULXLhwARcvXmzWRzmSdPHi\nRXbaiQ4dOtBgEoRIkEAFc8aMGXj9+jUCAwMRHR3NPpPLMAxUVVWxcuVK2NnZNShASEgIwsPDkZ2d\nDW1tbSxatAg2NjYCvXfjxo2IiIjAgQMHMHjw4AZ9viyprKzE999/zy5v2LCBbnIQIkECd9pzdnbG\n3Llzcfv2beTk5IDH46FDhw4wMDBAixYtGvThERER8Pb2xoYNG9CvXz8kJCRg+fLlUFdXx1dffVXv\ne//9918cOnSoWQ2ocPDgQdy8eRMA0Lt372Y3GtHx48eRkJAAb29vrqOQZkqoXs6tWrUS6ZFcYGAg\n7O3tYWtrCwDQ1dVFcnIyAgIC6i2YlZWVWL9+PWxtbXHo0CGR5ZFmJSUlWLNmDbu8bdu2ZtVJ/fz5\n81iwYAHOnDnDdRTSjNV70+fGjRuYO3cuBgwYgP79+2PWrFkiG97/0aNHyM7OrjEXtJmZGW7evImy\nsrI63xsWFoaSkhI4OjqCYRiR5JF2u3fvRkZGBoCqGT3Hjh3LcSLJiY+Px6xZs3Ds2DEMGjSI6zik\nGauzYP7999+YO3cuUlNTYWRkBDMzM3ZqilOnTjX6g9PT08Hj8Wrc4dXS0kJlZSVbHD724sUL+Pj4\nYP369VBUVGx0Dlnw6tUrdlIzHo+H7du3N5tLEdevX4ednR0iIyM/eZmGEHGr85wuICAA3bt3R2ho\nKNthvby8HCtXrsT27dsxbty4Rn1wcXExgKp+hB+qXq7ulP0xT09PjB49GsbGxuxjgU3d9u3b8erV\nKwBVN+D69evHcSLJ2bRpE0JDQzFq1CiuoxBSd8H8+++/8cMPP/A93aOoqIilS5fCysoKz549k3j/\nv7i4OCQnJ+PcuXMi2V5t0wfXdymAC7m5udi1axcAQF5eHj/99BPHiSTrxIkTzeZomoifs7NzjZvU\nR48eFfj9dZ6Sv3nzBtra2jXatbW1wTBMnUeAglJVVQVQ80iyevnj7jIlJSXw9PTEypUr2Ym9msP1\ny59//pk9Gp83bx66d+/OcSLJomJJpEm9t1nl5eVrtH3YB7MxdHR0wDAM0tPT0aNHD7Y9LS0NCgoK\nNYr1f//9h+fPn+OHH36Ah4cHX545c+ZAS0sL58+fFypDbb9ZMjMzYW5uLuTeiMezZ8/g7+8PAGjR\nogXffhNChBcQENCooSg565eiq6sLLS0tXL58ma9AJSQkwNTUtMYNHQMDA/YJl2o5OTlwdHSEl5cX\n+vfvL5HckrRp0yb2GX0nJ6daj/ibktzcXLRp06ZZdZcisqXe/5kpKSl4+/Ztra/duXMHhYWFfG3C\n9tF0dXWFh4cH+vXrByMjI5w6dQpJSUns/DQ7duxASkoK9u/fj1atWtU4Ha0eEKRLly5N7ln2J0+e\nICgoCEDVfn7YB7MpysnJwfDhw+Hl5cX2yyVE2tRbMOubx3ndunXszwzDgMfjITU1VagPt7W1RUlJ\nCXx9fZGTkwNdXV34+fnB0NAQAJCXl/fJ+c6b6jWuDRs2sFMbL168uEkPMJGfn4/Ro0dj2rRpVCyJ\nVOMxdVyMPHbsmNAbmzRpUqMDca36GmZsbCxn027cu3cPvXv3RmVlJVRVVfHkyZMmO3zZ69evYWFh\ngREjRmDr1q1N9hcg4Zaovtd1HmE2heInq3788UdUVlYCANzd3ZtssSwuLoaNjQ2MjIyoWBKZQFPr\nSZl///0XkZGRAKrGevz22285TiQ+FRUVmDhxInx8fKhYEplAtyOlzIddh1asWFHrtCBNhZqaGo3n\nSWQKHWFKkRs3buDEiRMAqgYHdnNz4zgRIeRDVDClyKZNm9ifV69eDWVlZQ7TEEI+RgVTSvz333+I\niooCAGhqamLhwoUcJxIthmEQGBgodc/qEyIMoQrms2fPcObMGYSEhODly5cA6h5ViAjHy8uL/dnd\n3b3eWTplDcMwWLZsGUJCQqhgEpkm0E2f9+/f46effsKRI0dQWVkJHo8HExMTtG3bFj4+Prh9+zYC\nAwNpfpkGevjwIX7//XcAQJs2beDi4sJxItFhGAZr1qzBlStXEBsbS/9HiEwT6AgzICAAJ06cwKJF\ni3D06FG+gTesrKyQnp7ODhJBhPfzzz+z/S6XLFkCNTU1jhOJzqZNm3Dy5EmcP38e6urqXMchpFEE\nKpjHjx+Hq6sr3Nzc0Lt3b77X+vfvjyVLluD06dNiCdjUZWRkICwsDEDVkHZLlizhOJHo/P777wgL\nC0NMTAzatWvHdRxCGk2gU/Lnz59jwIABdb7evXt39pomEc62bdvYZ8ZdXFygoaHBcSLRmTBhAoYN\nG9akn4MnzYtAR5ht2rTBkydP6nw9NTW1SX3RJSU7Oxv79u0DUDUjp7u7O8eJRKt169bo3Lkz1zEI\nERmBCuaoUaOwa9cuXLlyhW3j8XgoKytDVFQUtm/fDgsLC7GFbKq8vb3Z8S7nz59PR2KESDmBTsm/\n++473LlzBwsWLGAnKXNwcEBRUREqKiqgr6/fpJ95Fof8/Hz2RpmCggI9IkiIDBCoYKqpqSEyMhLn\nz5/H1atXkZ2dDQDo1KkTTE1NMWbMmFqnsyB18/HxYfuwOjg4yPxo6tHR0Th48CBCQkK4jkKI2Ag8\n+Ia8vDysra1hbW0tzjzNQmFhIXbv3g0AkJOTw6pVqzhO1DgJCQmYOXNmg8ZQJUSWCFQwk5OTBdqY\nsFNUNFd79+5FQUEBAGDatGl8k8DJmuvXr2PKlCn4/fffMWTIEK7jECJWAhXM2bNnCzReobBTVDRH\nJSUl8Pb2Zpdlea6emzdvYsKECQgNDZWamTYJESeBCuaHzzlXYxgGubm5uHjxIlq0aAEnJyeRh2uK\nwsPD2WvAtra20NfX5zhRw+3evRt79+6lyzSk2RCoYNY3XYWTkxNWrFiBW7du4auvvhJZsKaIYRjs\n2bOHXZb1O+MhISE0UjppVkQyvNukSZPYwSNI3RISEvDff/8BAAYNGgRTU1OOEzUOFUvS3IikYObn\n59eYo5zU9OHR5ZIlS6jgECJjBDolrx7Y9mPl5eXIyspCZGQkevXqJdJgTU1aWhqOHz8OoGqA4KlT\np3KcSDh5eXlQVlZuUuN0EiIsgQrmqlWrwOPxUMcU5ujatSvWrVsn0mBNjb+/PzuEm5OTE1q2bMlx\nIsHl5+fDwsICS5cuxbx587iOQwhnBCqY1cOPfUxOTg5qamro1q0bPelTj+LiYnaQDQUFBTg7O3Oc\nSHBv3ryBlZUVLCwsMHfuXK7jEMIpgQpmSUkJ+vbtizZt2og7T5MUERGBV69eAQDs7OxkZgSf4uJi\n2NjYYNCgQdi2bRtdcyXNnkA3fdzd3fH06VNxZ2mSPu5KJCsDBL979w62trbo3r07fH19qVgSAgEL\n5qRJk2gCqwa6ePEi7ty5A6Dq0VFjY2OOEwmGx+PBxsYGQUFBkJOjyUUJAQQ8JW/dujUyMzNhZmYG\nQ0NDaGhoQEGB/608Hg+bN28WS0hZJqtdiVq2bIlly5ZxHYMQqSJQwQwMDGR/vnr1aq3rUMGs6cmT\nJzh58iSAqq5EdnZ2HCcihDSGQAXz7t274s7RJH3YlcjZ2VmmuhIRQmqq8+JUVFQUXr9+LcksTUpx\ncTGCgoIASH9XIoZhsHfvXnpai5BPqLNgrl69GpmZmZLM0qSEh4ezXYmmTp2KTp06cZyobuvWrcO+\nfftQUVHBdRRCpFqdp+R1PdVDPk2WuhJt2rQJUVFRSEhIgLq6OtdxCJFqAk9RQQQXFxeHlJQUAICR\nkZHUdiXauXMnQkNDkZCQgHbt2nEdhxCpV2/BzMvLQ1ZWlsAbk5UnWMTtw6PLxYsXc5ikbufOncOe\nPXuQkJAg1ZcLCJEm9RZMYW9U0BQVwOPHj9muRB06dJDarkQWFha4cuUKunTpwnUUQmRGvQXTzs4O\nmpqaksrSJOzdu5e9/ivNXYkUFBSoWBIipHoL5rRp09CnTx+xBggJCWHnudHW1saiRYtgY2NT5/rX\nrl2Dj48P7t+/DxUVFZiZmeH7779H27ZtxZpTEGVlZey83AoKCjTPESFNDKcPCUdERMDb2xtubm44\nefIkpk6diuXLl9f5NNHNmzexcOFC9OvXD0eOHMG2bdtw8+ZNqXmE78SJE8jLywMAjB8/XqquDVKv\nB0Iaj9OCGRgYCHt7e9ja2kJXVxdz5szBqFGjEBAQUOv6oaGh6NmzJ1auXAldXV0YGRlhyZIluHHj\nBl68eCHh9DVVd1QHgPnz53OYhN+lS5cwceJEKpqENFKdp+STJk0S6/iXjx49QnZ2NszMzPjazczM\nsGnTJpSVlaFFixZ8r23ZsgUlJSV8bdWn4gUFBejYsaPY8n7K06dPER0dDQDo0qULxowZw1mWD/35\n55+YMmUKfvvtN5kZ+IMQaVXnEaaXl5dYuwmlp6eDx+PVuPGgpaWFyspKZGRk1HhPq1atahTxuLg4\nqKiooFu3bmLLKoiQkBD2CM7R0VEqRqD/559/MGHCBPz6668wNzfnOg4hMo+zU/Li4mIAVUPHfah6\nuaio6JPbSExMRHh4OJydnWscjUpSRUUFgoODAVSN2uTo6MhZlmopKSkYO3Ys/P39672JRggRnMw+\n6XPt2jW4urpizJgx+Oabbxq0jcmTJ9doa8ggyTExMUhPTwcAmJubQ1dXt0F5RCkoKAjbtm3D119/\nzXUUQqRGbQdXR48eFfj9nBVMVVVVADWPJKuXVVRU6nxvXFwcli1bBhsbG2zatEl8IQUkjTd7duzY\nQdcsCRExzgqmjo4OGIZBeno6evTowbanpaVBQUEB2tratb4vOTkZS5cuxcyZM7Fq1apGZajtN0tm\nZqZQ1/tyc3PZ+cY1NDRga2vbqEyiQsWSkJoCAgLQtWvXBr+fs2uYurq60NLSwuXLl/naExISYGpq\nCkVFxRrvyc3NhZubG6ZMmdLoYikqBw8eRHl5OQBg9uzZUvtkDyGk8Ti9hunq6goPDw/069cPRkZG\nOHXqFJKSkhAREQGg6rQyJSUF+/fvBwDs3r0bLVu2hJOTE9tBvJqqqionxao6KwDO5u3Oz88Hj8ej\naZAJETNOC6atrS1KSkrg6+uLnJwc6Orqws/PD4aGhgCqRkv6cBDjxMRE5ObmYuTIkTW25eXlJfHT\n4QcPHiA5ORkA0Lt3bza3JL158wZWVlaYPn063N3dJf75hDQnnN8lnzFjBmbMmFHra15eXnzLsbGx\nkogksA+PLmfOnCnx64bFxcUYN24cBg4ciG+//Vain01Ic0QTTjcQwzB8BdPe3l6in//u3TvY2tri\niy++gJ+fH93kIUQCqGA2UHJyMh4+fAgA+OqrryTa97K8vBx2dnbQ0NBAUFAQ5OTon5EQSaBvWgNF\nRkayP0v66FJOTg7W1tYIDw+HggLnV1UIaTbo29YADMPg8OHDAKqK15QpUyT6+fLy8nBxcZHoZxJC\n6AizQZKSkthHIUeMGEGj0hPSTFDBbIBDhw6xP0vrnD2EENGjgikkhmHYgiknJ1frAB6i5u/vj5yc\nHLF/DiGkflQwhfTh6fjw4cPFfjq+efNm6jZEiJSgmz5C+vB0fOrUqWL9rF27diEkJAQJCQlo3769\nWD+LEPJpVDCF8PHdcXGejv/yyy/YtWsXLl26JFWTqRHSnFHBFEJycjKePn0KQLyn44mJifD09ER8\nfHydw9wRQiSPCqYQJHV33MTEBElJSXRkSYiUoZs+ApLk3XEej0fFkhApRAVTQB+fjnfo0IHjRIQQ\nSaOCKSBxno5XT89LCJFuVDAFIM6743/++SdGjBiBiooKkW2TECIedNNHAHfu3EFaWhoAYMiQISI7\nHf/nn38wYcIEBAcHQ15eXiTbJISIDx1hCuD06dPsz+PGjRPJNlNSUjB27Fj4+fnBxsZGJNskhIgX\nFUwBiLpgPnz4EJaWlti2bZvEh4YjhDQcFcxPKCgowLVr1wAAn3/+OfT09Bq9zUOHDuGHH37ArFmz\nGr0tQojk0DXMT4iNjWVvyIwdO1Ykg2CsXr260dsghEgeHWF+wvnz59mfraysOExCCOEaFcx6MAzD\nFkxFRcVa50MnhDQfVDDrcffuXWRkZAComhlSRUVF6G28efMGWVlZoo5GCOEAFcx6fHg6PmbMGKHf\n//btW4wbNw4BAQGijEUI4QgVzHrExMSwPwtbMN+9ewdbW1vo6upi/fr1Ik5GCOECFcw6VFZW4vLl\nywAADQ0NGBoaCvze8vJyTJ06Ferq6ggODoacHP01E9IU0De5DqmpqXjz5g0AYOjQoQIXvYqKCrZ/\nZXh4OBQUqOcWIU0FFcw6XL9+nf15+PDhAr9PTk4O1tbW+OOPP9CiRQtxRCOEcIQOf+rQ0ILJ4/Ew\nZ84ccUQihHCMjjDrkJycDAD47LPPhLp+SQhpuqhg1uH169cAqoZzo6HXCCEAFcxPGjZsWL2v7927\nF0+ePJFQGkIIl6hgfoKxsXGdr+3evRs7duxAy5YtJZiIEMIVuunzCf3796+1PTAwEDt37kRCQgI6\nd+4s4VSEEC5QwaxHr169oKamVqP9wIED2LBhA+Lj46Gjo8NBMkIIF+iUvB7FxcWIjo7mm9UxJSUF\nK1asQHR0NLp3785hOkKIpFHBrEdmZibGjBkDAwMDPHjwAADQu3dv3Lp1C7179+Y4HSFE0jgvmCEh\nIbCwsICBgQFsbGz45s+pze3btzF79mwYGhrCxMQE69evx7t378Sa8c6dOzAzM2OLpqamplg/jxAi\nnTgtmBEREfD29oabmxtOnjyJqVOnYvny5bh69Wqt6+fm5sLR0RFdu3bFkSNHsGvXLly7dg0eHh5i\nz5qXl4dJkybxnZ4TQpoXTgtmYGAg7O3t2WHQ5syZg1GjRtU5fuSBAwfQokULbNy4Ed27d4eJiQlW\nrlyJU6dOITMzU+x579y5wzfkGyGkeeGsYD569AjZ2dkwMzPjazczM8PNmzdRVlZW4z3Xr1+HkZER\n3whA1e9PTEwUb+D/8ff3l8jnEEKkD2cFMz09HTweD126dOFr19LSQmVlJTs1xIeePn1aY30lJSW0\nbdsWaWlp4ozLSkhIkMjnEEKkD2cFs7i4GADQunVrvvbq5aKiolrfo6SkVKO9devWta4vDoWFhRL5\nHEKI9OH8LrmsUVVV5ToCIYQjnD3pU114Pj4yrF6ubYZGVVXVWo8kCwsLG1TIJk+eXKOttLQUAOoc\nKd3U1FQiN5gIIaLz4sULAICTk1ONsR+OHj0q8HY4K5g6OjpgGAbp6eno0aMH256WlgYFBQVoa2vX\n+p6Pr22+efMGBQUF6Natm0hyPX36FABq/XwAePjwIczNzUXyWYQQyXr//n2jBsvhrGDq6upCS0sL\nly9f5itACQkJMDU1haKiYo33DB06FGFhYSgrK2Onf4iPj4e8vDyGDBkidIbafrPY2tqipKQEgYGB\nMjsOprOzMwDI9PS+tA/SoSnsA1B1ZPn+/XscP34crVq1avB2OB18w9XVFR4eHujXrx+MjIxw6tQp\nJCUlISIiAgCwY8cOpKSkYP/+/QCAmTNnIiIiAmvWrMHixYvx4sUL7NixA9OnT0f79u1FkklOTg7K\nysoyPahG9S+Trl27cpyk4WgfpENT2AcAaNmyJVq2bNmoYglwXDCrj+Z8fX2Rk5MDXV1d+Pn5sVNC\n5OXl8V0vVFdXR0hICDw9PTFx4kSoqKhgwoQJcHd352oXCCHNCOfDu82YMQMzZsyo9TUvL68abT17\n9kRYWJi4YxFCSA3UrYgQQgTEY2g0CUIIEQgdYRJCiICoYBJCiICoYBJCiICoYBJCiICoYBJCiICo\nYBJCiICoYBJCiICoYBJCiICoYBJCiICoYBJCiICaXcEMCQmBhYUFDAwMYGNjg9OnT9e7/u3btzF7\n9mwYGhrCxMQE69evx7t37ySUtnbC7sO1a9cwY8YMDBw4EMOHD8fq1avx8uVLCaWtm7D78aGNGzdC\nT08PycnJYkz4acLuQ1FRETw8PGBsbIwBAwZg/vz5tU74J0nC7kNiYiJmzpyJwYMHY+DAgXB1dWUH\n3uYKwzDYs2cPvvzyS/j6+n5y/QZ/r5lmJDw8nDEwMGCOHTvGPHnyhAkJCWG+/PJL5sqVK7Wun5OT\nwwwaNIhZtWoV8+DBAyYxMZEZPXo08/3330s4+f8Tdh/++usvpk+fPszPP//MPHnyhPnzzz8ZS0tL\nZtasWRJOzk/Y/fjQrVu3GAMDA0ZPT49JSkqSQNraNWQfZs2axcyePZu5e/cuk5qaykyfPp2xsbGR\nYGp+wu7D7du3+f4/paSkMA4ODszIkSOZt2/fSjh9lfz8fMbR0ZGxtLRk+vTpw/j4+NS7fmO+182q\nYA4bNozx8vLia3N1da2zeOzYsYMxMzNjysvL2baYmBhGT0+PycjIEGvWugi7D0uWLGEmTZrE13bq\n1ClGT0+Pef78udhyfoqw+1GtoqKCmTRpEuPh4cF5wRR2Hy5dusT069ePKSgoYNsyMzOZ8+fPM6Wl\npWLNWhdh92HLli2MsbExX1tqairTq1cv5tKlS2LLWZ+QkBDGycmJKSwsZAwMDD5ZMBvzvW42p+SP\nHj1CdnY2zMzM+NrNzMxw8+ZNlJWV1XjP9evXYWRkxDchWvX7ExMTxRu4Fg3Zhy1btrAj1ldr27Yt\nAKCgoEB8YevRkP2oFhYWhpKSEjg6OoLhcKCthuzDxYsXYWxsDHV1dbatS5cusLS0ZEc2l6SG7AOP\nx4OcHH/ZqJ5OhsfjiS9sPSwsLBAQEFDrxIm1acz3utkUzPT0dPB4PHTp0oWvXUtLC5WVlbVeR3r6\n9DBNl5oAABCSSURBVGmN9ZWUlNC2bVukpaWJM26tGrIPrVq1Qps2bfja4uLioKKiIrKJ44TVkP0A\nqmb+8/Hxwfr162ud80mSGrIP9+/fh66uLvbt24cxY8bA1NQU7u7uyM/Pl1RsPg3Zh8mTJ6OkpATB\nwcEoLS1FSUkJ/P398fnnn8PExERS0fl8nP9TGvO9bjYFs7i4GADQunVrvvbq5dqm7y0uLoaSklKN\n9tatW9e6vrg1ZB8+lpiYiPDwcDg7O3NyVAM0fD88PT0xevRoGBsbizegABqyD/n5+Th79izu378P\nb29veHl54datW3BwcEBlZaX4Q3+kIfvQrVs3+Pn5wd/fH/3798fAgQNx9+5dBAUF1Tk1tbRpzPda\nNvaQiMS1a9fg6uqKMWPG4JtvvuE6jlDi4uKQnJyMc+fOcR2lwd6/f49WrVph69at7Olry5YtMW/e\nPFy5cgXDhg3jOOGn3b9/H+7u7pg8eTImTJjAzrDq5OSEyMhIKCsrcx1RrJrNEaaqqiqAmr81q5dr\nu/6hqqpa62+cwsJCdnuS1JB9qBYXFwdnZ2dYWVlhx44d4gspAGH3o6SkBJ6enli5ciV7eYHL65dA\nw/4tlJWV0atXL75rfQMHDgSPx8O9e/fEmLZ2DdkHX19faGlpYc2aNdDX18fgwYOxc+dOZGZm4siR\nI+IPLQKN+V43m4Kpo6MDhmGQnp7O156WlgYFBQVoa2vX+p6Pr+O8efMGBQUFnFz/a8g+AEBycjKW\nLl0Ke3t7eHl51bhoL2nC7sd///2H58+f44cffkCfPn3Qp08fjBkzBjweD3PmzMGYMWMkGR9Aw/8/\nvX79mq+tsrISDMMIfMNClBqyD48fP8YXX3zB16aiogINDQ3O+2IKqjHf62ZTMHV1daGlpYXLly/z\ntSckJMDU1LTWmwhDhw5FcnIy393C+Ph4yMvLY8iQIWLP/LGG7ENubi7c3NwwZcoUrFq1SlJR6yXs\nfhgYGODkyZOIiorC8ePHcfz4cezbtw8Mw2Dz5s0IDAyUZHwADf//dOvWLbx69Yptu3nzJng8Hnr1\n6iX2zB9ryD506NChxo2RwsJC5ObmokOHDuKMKzKN+l43pv+TrDl27Bijr6/PHDt2jHn27Bnzyy+/\nMH369GH++ecfhmEYZvv27YyjoyO7fkFBAWNqasp89913TFpaGnP9+nVm2LBhzIYNG7jaBaH3Ye3a\ntczQoUOZ58+fM7m5uXx/3r17x9VuCL0fH8vMzGR69erFaT9MYfehtLSUsbKyYmbPns08ePCAuX79\nOmNpacnMnDmTq10Qeh8uXrzI6OnpMTt37mQePXrEpKSkMIsWLWL69evHpKenc7IPr169YnJzc5mc\nnBzGwMCA2bJlC/t/vKKiQqTf62Z108fW1hYlJSXw9fVFTk4OdHV14efnB0NDQwBAXl4eMjMz2fXV\n1dUREhICT09PTJw4ESoqKpgwYQLc3d252gWh9yExMRG5ubkYOXJkjW15eXnB1tZWYtk/JOx+1Iar\nfn/VhN2HFi1asP+fpk2bBjk5OVhYWGDt2rVc7YLQ+zBixAj4+PjAz88PwcHBUFRURN++fREaGgot\nLS1O9sHNzQ03btxgl3/99VcEBweDx+MhNjZWpN9rmmaXEEIE1GyuYRJCSGNRwSSEEAFRwSSEEAFR\nwSSEEAFRwSSEEAFRwSSEEAFRwSSEEAE1q47rzYWvr2+985rweDxcuXKFHUj4U5KSkuDg4IADBw5g\n8ODBoopZp9mzZ9eYq0dBQQGampowNzfH4sWLoaamJpbP5fF4CAsLq3MdPT09uLm5wc3NTeSf/7Hq\nv/ePycnJQUNDA4MHD4aLiwt69uwp9iykChXMJorH4yEgIADt2rWr9fWPBxUWZHuSpKmpiYCAAHZU\nouLiYvz999/45ZdfkJSUhMOHD4t8EOGNGzfyLUdHR+Pnn39GXFwc23bkyBFoamqK9HM/ZfHixRgx\nYgS7XFZWhocPHyIwMBDTp0/HoUOHhB4M5tdff0VcXBwOHDgg4rRNGxXMJqxHjx7o3Lkz1zEaRFFR\nEb179+ZrMzIygqamJlavXo0zZ85g4sSJIv1MXV1dvuWkpKQavyj69Okj0s8URJcuXWp8bv/+/TF4\n8GBYWVkhLCwMP/30k1DbTE5O5vzRUllE1zCbubdv32LHjh0YOXIk9PX1MWLECKxatQq5ubn1vi8o\nKAjW1tbo168fTExM4OTkhLt37/Ktc+PGDcydOxf9+/dH//79YW9vj2vXrjUqb/WR1u3bt9m2/Px8\nrFu3DkOGDIG+vj6GDx+OjRs31hjz8FOZZ8+ezZ4Cz549G+Hh4Xj27Bn09PSwevVqAFWn5L6+vigu\nLoahoSHb/qG4uDjo6enhwoULAKqOCLdv345Ro0ax+TZv3syOeN5Qurq6UFdXR1ZWFl97ZGQkJkyY\nAAMDA5iYmMDR0ZHv72vUqFGIi4tDUlIS37S0DMNg3759sLKygr6+PszMzKRmSmZpQQWzmfPw8MCB\nAwfg4uKC8PBwLFmyBHFxcVi8eDHfeh8OOXDgwAHs2rULU6dORXBwMH7++WcUFxdj7ty5ePv2LQDg\n33//xdy5c8Hj8bBnzx74+PhAXV0dCxYsaNRc4tXTIFRP6VBWVoY5c+YgPj4eS5cuRUhICFxcXHDi\nxAk4OzsLlflDGzZsQO/evaGpqYmjR4/W+PtQVlbG0KFDER8fX2Mw4+joaKioqGD48OEAgG+//RYH\nDx6Eg4MDQkJC4OTkhKioKCxZsqTBfw8AkJ2djdevX0NHR4dti4yMxI8//oivvvoKoaGh8PLyQl5e\nHubPn4+8vDwAwC+//IL27dujT58+OHLkCKZNmwagajCW3bt3Y9y4cf/X3rmHNPn9cfz9eJmXIYak\nKBoUmkzUJSpak9KpmcsCpRCpkaalUiQG3Sw0U5JAE3KVNwKpFCUoMRQSp5ahaUoJll2MkpScZWqY\nMpt+vn/Inp9rXpbf4PdFzwv2x3POeZ7zPnvgs3PbeaOsrAxnzpxBW1sb4uLi8OvXr3+ldbXAhuRr\nGCKCkZERjh07hujoaACAl5cX+vv7UVJSgsHBwQUNplpbW+Hq6oq4uDg+TSwW4/79+5iYmIClpSUK\nCgpgb2+P4uJi3jtIIpFAJpNBoVAsubCyFF1dXQD+NzSuq6tDX18fbt68yZ/I5OvrCyJCVlYW2tvb\n4e/vb5Dm+WzatAlCoRDj4+N6UwNadu/eDaVSic7OTn4xTKPRoKmpCTt37oRAIEB3dzeUSiUuXbrE\nByZfX19YWloiLS0Nz58//+OFtOnpabx9+xY5OTkQCASIiYnh80ZHRxEREYGzZ8/yaQKBAAkJCWhp\naUFUVBQ2b94MU1NTCIVCvm0qlQoVFRWIj4/nF7R8fX3h5OQEuVyO2tra/9vJVv8lWA9zFbPcQVQc\nxyE3NxeJiYk66U5OTgCAL1++LHjf+vXr8f79e9y7dw9qtRoAYGNjgyNHjsDOzg4ajQYdHR0IDAzU\nMVozMjLiD9H9U6amptDa2ors7Gw4OjoiIiICANDe3g4TExM9P5ygoCAQEX/s13KaV4JUKoWZmRk/\n9AbmLFzHx8exZ88eAMDTp0/BcRzCwsIW1Pfy5ctl60lLS4NIJOI/YrEY0dHRMDExQVlZGVxcXPiy\nycnJehYk2mPXFnufWt0zMzN6On18fGBlZWWQzrUA62GuUogIISEhC+bZ29ujubkZAPDmzRuUlpai\no6MD379/x8zMDIC5YDo/4M5fIDh16hQ+ffqEjIwMZGdnw8fHB+Hh4YiKioJAIMDY2Bimp6dRUVGB\n8vJynbo5jgPHcRgfH4e1tfWi+rVzh/MxMjKCv78/srKyYG5uDgAYHh6GjY0NjI2Ndcra2try+YZo\nXgkWFhYIDAyEUqnE+fPnAcwNx21sbLBt2za+fiLir3//LlQq1bL1pKSk6JxnWlRUhMePHyMvL0/v\nlPPR0VEUFRWhsbERKpWKP1Wc47glnSlVKhWICPv371+xzrUAC5irFI7jUFJSwgeO+Wi34wwNDeHA\ngQOwtrbGyZMn4eLiAoFAAKVSueQ+Tmtra9y5cwc9PT1oaGhAU1MTLl68iPLyclRVVfHlIiMjERsb\nu+AzlnMXtLOz420ogLm5S1tbW70gu9hKr/Y+rX/RcpoXsl01BJlMhvr6evT29kIkEqGxsRHh4eE6\nvkkcx6GqqmrBwLxu3bpl63BwcND58Th9+jQaGxuRl5eH3NxcnbLx8fH48OEDjh8/Dl9fXwiFQgwP\nD+uNIhaC4zgoFAp+hDGf1e4GaSgsYK5inJ2dl9xW1NDQgKmpKSgUCgQEBPDpTU1NBj3fw8MDHh4e\nSE1NRXV1Nc6dO4e6ujpERkbCzMwMGo1Gr5doKKampgb53Njb2+PZs2fQaDQ6vtjaHtHvPbDFNO/b\nt29FOoOCgmBubg6lUomfP3/i27dv/HQBMBfsgLnAuJhJ3Z/i6OiIgwcPoqysDHK5nD8d/d27d+jt\n7UVSUhKSkpL48oasxmt1CoXCFb+ztQCbw1zDaDQaANDphU5OTvJ2qdrh+e/3XLlyRc84Sztk/PHj\nB4yNjeHn54fm5ma9rT0lJSWorq7+a22QSCSYmZnhpxi0NDU1geM4SCQSgzQvxHLDWAAwNzeHVCpF\nS0sLGhoa4ODgAB8fHx19RISamhqd+z5//oyMjAwMDQ0Z2lQdkpOTYWVlhZycHD5tofcJAHfv3tVr\ny+/X/v7+MDIy0tM5MTGBCxcuoK+vb0U6Vxush7mG0a4m5+XlITExESMjIygsLMTevXtRWFiIR48e\n8T3U+UPj/v5+PHz4ECkpKXB1dcXU1BQqKiogFAoRHBwMYO7fKXK5HLGxsUhJSYGFhQXq6+tRXl6O\nzMzMv9aGsLAwuLm5ITMzk99i8/r1a1y7dg3BwcEQi8UAsKTmxeZ6bW1t0dXVhaqqKjg5Oen0wucj\nk8mQmpqKr1+/QiaT6eSJxWKEhoaiqKgIxsbG2Lp1KwYGBnDjxg3Mzs7yc59/irW1NRITE5GXl4fq\n6mpERkbC2dkZNjY2uHPnDjZs2AATExNUVlZi48aNMDU1RVtbG78X1NbWFr29vaipqYGjoyN8fHwg\nl8tx+/ZtWFlZYdeuXRgZGUFpaSkGBgb+9RaoVcPfcm5j/HdQKBQkEolocHBw2bKVlZUUGhpKW7Zs\nocjISFIqlTQzM0MJCQkkFovp6tWr1N7eTiKRiHdonJycpMuXL5NUKiVPT0+SSCR09OhR6u7u1nn2\nixcv6PDhw+Tt7U2enp4UFRVFtbW1y2qSy+UUEhJicHvHxsYoPT2dAgICyN3dnYKDgyk/P5/UajVf\nxhDNcrmcDh06xF+/evWKwsLCyMPDg06cOEFERCKRiK5fv65Tv1qtJm9vb3Jzc6Oenh49fWq1mvLz\n8ykkJITc3d1JIpFQeno6qVSqJdul/d4fPHiwYL5arSapVEo7duygyclJIiLq7OykqKgo8vLyouDg\nYCouLiYiooKCAvLy8qKYmBgiInry5Alt376dPD09KScnh4iIZmdn6datWySTycjDw4P8/PwoNTWV\nPn78uKTOtQQzQWMwGAwDYXOYDAaDYSAsYDIYDIaBsIDJYDAYBsICJoPBYBgIC5gMBoNhICxgMhgM\nhoGwgMlgMBgGwgImg8FgGAgLmAwGg2EgLGAyGAyGgfwDpr9mOSTdzLcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f976a0c0a50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "MODEL_VERSION = \"v4\"\n",
    "MODEL_SUFFIX = \"%s.last_author\" % MODEL_VERSION\n",
    "\n",
    "results_full = fit_model(df_filtered_la, base_formula, include_prc=True,\n",
    "                            MODEL_VERSION=MODEL_VERSION, MODEL_SUFFIX=MODEL_SUFFIX, version=version)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "results_filtered = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" \n",
    "                        % (MODEL_VERSION, MODEL_SUFFIX, version))\n",
    "with open(\"Filtered_models.%s.filtered_%s.txt\" % (MODEL_SUFFIX, version), \"wb+\") as fp: \n",
    "    print_model(results_filtered, fp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build combined tables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def merge_results(results, keys=(\"Full\", \"Filtered 1\", \"Filtered 2\"), columns=(\"Coef.\", \"Std.Err.\", \"P>|z|\")):\n",
    "    df_list = []\n",
    "    columns = list(columns)\n",
    "    for r in results:\n",
    "        df_list.append(r[-1][0][1][1].tables[1][columns])\n",
    "    df_t = pd.concat(df_list, keys=keys, axis=1)\n",
    "    df_t = df_t.reindex(results[0][-1][0][1][1].tables[1].index)\n",
    "    return df_t\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### First author"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "MODEL_VERSION = \"v4\"\n",
    "MODEL_SUFFIX = \"%s.first_author\" % MODEL_VERSION\n",
    "\n",
    "results_full = load(\"results_all_model/%s/results_all_model.%s.pkl\" % (MODEL_VERSION, MODEL_SUFFIX))\n",
    "results_flt_1 = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" % (MODEL_VERSION, MODEL_SUFFIX, 1))\n",
    "results_flt_2 = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" % (MODEL_VERSION, MODEL_SUFFIX, 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"3\" halign=\"left\">Full</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 1</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Intercept</th>\n",
       "      <td>-1.957e+00</td>\n",
       "      <td>0.031</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-4.047</td>\n",
       "      <td>0.289</td>\n",
       "      <td>2.037e-44</td>\n",
       "      <td>-3.542</td>\n",
       "      <td>0.482</td>\n",
       "      <td>1.978e-13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(sink_prev_ncites == 0)[T.True]</th>\n",
       "      <td>1.265e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.064e-235</td>\n",
       "      <td>0.166</td>\n",
       "      <td>0.007</td>\n",
       "      <td>4.064e-120</td>\n",
       "      <td>0.156</td>\n",
       "      <td>0.011</td>\n",
       "      <td>2.378e-47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(sink_prev_ncites + 1)</th>\n",
       "      <td>-4.433e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.147</td>\n",
       "      <td>0.011</td>\n",
       "      <td>7.272e-39</td>\n",
       "      <td>-0.137</td>\n",
       "      <td>0.016</td>\n",
       "      <td>5.717e-17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(sink_prev_ncites + 1) ** 2)</th>\n",
       "      <td>-1.698e-01</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.226</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.181</td>\n",
       "      <td>0.006</td>\n",
       "      <td>2.708e-203</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(auth_prev_papers == 0)[T.True]</th>\n",
       "      <td>-1.421e+00</td>\n",
       "      <td>0.009</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.232</td>\n",
       "      <td>0.018</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.236</td>\n",
       "      <td>0.030</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(auth_prev_papers == 1)[T.True]</th>\n",
       "      <td>-2.357e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.159</td>\n",
       "      <td>0.010</td>\n",
       "      <td>2.285e-58</td>\n",
       "      <td>-0.158</td>\n",
       "      <td>0.017</td>\n",
       "      <td>1.823e-21</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(auth_prev_papers + 1)</th>\n",
       "      <td>1.610e+00</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>1.862</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>1.911</td>\n",
       "      <td>0.020</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(auth_prev_papers + 1) ** 2)</th>\n",
       "      <td>-9.062e-02</td>\n",
       "      <td>0.002</td>\n",
       "      <td>1.091e-294</td>\n",
       "      <td>-0.127</td>\n",
       "      <td>0.004</td>\n",
       "      <td>9.929e-187</td>\n",
       "      <td>-0.156</td>\n",
       "      <td>0.006</td>\n",
       "      <td>4.687e-130</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_journal[T.True]</th>\n",
       "      <td>2.661e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>6.473e-129</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_review[T.True]</th>\n",
       "      <td>-7.869e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>5.938e-199</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_case_rep[T.True]</th>\n",
       "      <td>-9.404e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_let_ed_com[T.True]</th>\n",
       "      <td>-4.295e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_journal[T.True]</th>\n",
       "      <td>3.283e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.814e-181</td>\n",
       "      <td>0.548</td>\n",
       "      <td>0.022</td>\n",
       "      <td>5.741e-140</td>\n",
       "      <td>0.496</td>\n",
       "      <td>0.032</td>\n",
       "      <td>3.460e-55</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_review[T.True]</th>\n",
       "      <td>-7.826e-01</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.741</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.639</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_case_rep[T.True]</th>\n",
       "      <td>-6.369e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.882</td>\n",
       "      <td>0.012</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.861</td>\n",
       "      <td>0.017</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_let_ed_com[T.True]</th>\n",
       "      <td>-2.838e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>9.580e-138</td>\n",
       "      <td>-0.421</td>\n",
       "      <td>0.021</td>\n",
       "      <td>1.665e-86</td>\n",
       "      <td>-0.451</td>\n",
       "      <td>0.031</td>\n",
       "      <td>7.459e-47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(year_span &lt; 0)[T.True]</th>\n",
       "      <td>-5.564e-01</td>\n",
       "      <td>0.016</td>\n",
       "      <td>2.011e-275</td>\n",
       "      <td>-0.445</td>\n",
       "      <td>0.028</td>\n",
       "      <td>1.636e-55</td>\n",
       "      <td>-0.327</td>\n",
       "      <td>0.041</td>\n",
       "      <td>2.789e-15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(year_span == 0)[T.True]</th>\n",
       "      <td>5.382e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.446</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.580</td>\n",
       "      <td>0.015</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.score_log_1(year_span)</th>\n",
       "      <td>5.387e-01</td>\n",
       "      <td>0.015</td>\n",
       "      <td>2.008e-275</td>\n",
       "      <td>-0.204</td>\n",
       "      <td>0.026</td>\n",
       "      <td>6.462e-15</td>\n",
       "      <td>0.152</td>\n",
       "      <td>0.038</td>\n",
       "      <td>7.348e-05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(mf.score_log_1(year_span) ** 2)</th>\n",
       "      <td>-1.281e+00</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.767</td>\n",
       "      <td>0.016</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.857</td>\n",
       "      <td>0.023</td>\n",
       "      <td>1.099e-302</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>journal_same[T.True]</th>\n",
       "      <td>4.567e-01</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.509</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.501</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(jj_sim == 0)[T.True]</th>\n",
       "      <td>-2.805e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.735e-259</td>\n",
       "      <td>-0.172</td>\n",
       "      <td>0.015</td>\n",
       "      <td>4.417e-31</td>\n",
       "      <td>-0.163</td>\n",
       "      <td>0.022</td>\n",
       "      <td>1.448e-13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(jj_sim + 1)</th>\n",
       "      <td>6.298e-02</td>\n",
       "      <td>0.010</td>\n",
       "      <td>3.421e-10</td>\n",
       "      <td>0.205</td>\n",
       "      <td>0.019</td>\n",
       "      <td>3.512e-28</td>\n",
       "      <td>0.255</td>\n",
       "      <td>0.027</td>\n",
       "      <td>1.016e-20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(jj_sim + 1) ** 2)</th>\n",
       "      <td>5.949e-03</td>\n",
       "      <td>0.003</td>\n",
       "      <td>4.459e-02</td>\n",
       "      <td>-0.044</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.126e-15</td>\n",
       "      <td>-0.056</td>\n",
       "      <td>0.008</td>\n",
       "      <td>8.183e-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_ncites == 1)[T.True]</th>\n",
       "      <td>-2.997e-01</td>\n",
       "      <td>0.022</td>\n",
       "      <td>1.212e-41</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(source_ncites)</th>\n",
       "      <td>-1.647e+00</td>\n",
       "      <td>0.034</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-3.223</td>\n",
       "      <td>0.562</td>\n",
       "      <td>1.002e-08</td>\n",
       "      <td>-7.395</td>\n",
       "      <td>0.816</td>\n",
       "      <td>1.230e-19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(source_ncites) ** 2)</th>\n",
       "      <td>8.198e-01</td>\n",
       "      <td>0.023</td>\n",
       "      <td>4.438e-269</td>\n",
       "      <td>2.579</td>\n",
       "      <td>0.379</td>\n",
       "      <td>9.953e-12</td>\n",
       "      <td>5.400</td>\n",
       "      <td>0.550</td>\n",
       "      <td>1.023e-22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(source_ncites) ** 3)</th>\n",
       "      <td>-2.294e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.771</td>\n",
       "      <td>0.084</td>\n",
       "      <td>3.470e-20</td>\n",
       "      <td>-1.386</td>\n",
       "      <td>0.122</td>\n",
       "      <td>5.992e-30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.UNKNOWN]</th>\n",
       "      <td>-4.468e-02</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.435e-12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.UK]</th>\n",
       "      <td>-5.242e-03</td>\n",
       "      <td>0.003</td>\n",
       "      <td>1.058e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.JAPAN]</th>\n",
       "      <td>2.669e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.GERMANY]</th>\n",
       "      <td>2.382e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>9.888e-09</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.FRANCE]</th>\n",
       "      <td>-4.425e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>3.363e-17</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.ITALY]</th>\n",
       "      <td>-1.010e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>3.885e-54</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.CANADA]</th>\n",
       "      <td>-3.874e-03</td>\n",
       "      <td>0.004</td>\n",
       "      <td>3.608e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.CHINA]</th>\n",
       "      <td>-4.120e-01</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.AUSTRALIA]</th>\n",
       "      <td>8.788e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.664e-63</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.SPAIN]</th>\n",
       "      <td>-1.223e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.850e-83</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.NETHERLANDS]</th>\n",
       "      <td>6.808e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>1.152e-21</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.SWEDEN]</th>\n",
       "      <td>1.074e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>4.490e-58</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.INDIA]</th>\n",
       "      <td>-4.363e-02</td>\n",
       "      <td>0.010</td>\n",
       "      <td>3.026e-05</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.OTHER]</th>\n",
       "      <td>-6.440e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>8.308e-128</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.nan_to_num(source_V_novelty) + 1)</th>\n",
       "      <td>-1.058e-01</td>\n",
       "      <td>0.001</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.096</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.112</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.108e-214</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.nan_to_num(sink_V_novelty) + 1)</th>\n",
       "      <td>9.484e-02</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.753e-35</td>\n",
       "      <td>0.200</td>\n",
       "      <td>0.014</td>\n",
       "      <td>2.878e-49</td>\n",
       "      <td>0.158</td>\n",
       "      <td>0.021</td>\n",
       "      <td>2.289e-14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)</th>\n",
       "      <td>-2.774e-02</td>\n",
       "      <td>0.001</td>\n",
       "      <td>1.412e-100</td>\n",
       "      <td>-0.044</td>\n",
       "      <td>0.002</td>\n",
       "      <td>7.102e-82</td>\n",
       "      <td>-0.037</td>\n",
       "      <td>0.004</td>\n",
       "      <td>6.418e-26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_eng[T.True]</th>\n",
       "      <td>6.494e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_eng[T.True]</th>\n",
       "      <td>-1.846e-01</td>\n",
       "      <td>0.010</td>\n",
       "      <td>6.033e-81</td>\n",
       "      <td>2.075</td>\n",
       "      <td>0.084</td>\n",
       "      <td>2.546e-133</td>\n",
       "      <td>3.625</td>\n",
       "      <td>0.268</td>\n",
       "      <td>1.149e-41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_n_authors &gt; 20)[T.True]</th>\n",
       "      <td>1.722e-04</td>\n",
       "      <td>0.015</td>\n",
       "      <td>9.906e-01</td>\n",
       "      <td>-0.037</td>\n",
       "      <td>0.025</td>\n",
       "      <td>1.343e-01</td>\n",
       "      <td>-0.047</td>\n",
       "      <td>0.034</td>\n",
       "      <td>1.627e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.clip(source_n_authors, 0, 20))</th>\n",
       "      <td>8.239e-02</td>\n",
       "      <td>0.018</td>\n",
       "      <td>3.324e-06</td>\n",
       "      <td>0.242</td>\n",
       "      <td>0.030</td>\n",
       "      <td>1.357e-15</td>\n",
       "      <td>0.303</td>\n",
       "      <td>0.045</td>\n",
       "      <td>1.063e-11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)</th>\n",
       "      <td>-2.563e-01</td>\n",
       "      <td>0.013</td>\n",
       "      <td>1.194e-85</td>\n",
       "      <td>-0.336</td>\n",
       "      <td>0.023</td>\n",
       "      <td>3.234e-50</td>\n",
       "      <td>-0.353</td>\n",
       "      <td>0.033</td>\n",
       "      <td>2.710e-26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[0]</th>\n",
       "      <td>-1.426e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.087e-04</td>\n",
       "      <td>-0.033</td>\n",
       "      <td>0.006</td>\n",
       "      <td>2.974e-08</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[1]</th>\n",
       "      <td>5.313e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>2.200e-36</td>\n",
       "      <td>-0.038</td>\n",
       "      <td>0.007</td>\n",
       "      <td>1.021e-07</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[2]</th>\n",
       "      <td>4.562e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>5.136e-29</td>\n",
       "      <td>0.180</td>\n",
       "      <td>0.006</td>\n",
       "      <td>9.166e-200</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[3]</th>\n",
       "      <td>-1.021e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.235e-60</td>\n",
       "      <td>-0.147</td>\n",
       "      <td>0.009</td>\n",
       "      <td>7.110e-61</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[4]</th>\n",
       "      <td>4.401e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.985e-22</td>\n",
       "      <td>0.116</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.022e-50</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[5]</th>\n",
       "      <td>-7.411e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.704e-63</td>\n",
       "      <td>-0.066</td>\n",
       "      <td>0.008</td>\n",
       "      <td>4.727e-16</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[6]</th>\n",
       "      <td>-5.946e-02</td>\n",
       "      <td>0.006</td>\n",
       "      <td>7.073e-23</td>\n",
       "      <td>-0.092</td>\n",
       "      <td>0.010</td>\n",
       "      <td>7.999e-20</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[7]</th>\n",
       "      <td>-6.379e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>5.197e-34</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.007</td>\n",
       "      <td>4.308e-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[8]</th>\n",
       "      <td>1.355e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.099e-172</td>\n",
       "      <td>0.030</td>\n",
       "      <td>0.012</td>\n",
       "      <td>8.899e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[9]</th>\n",
       "      <td>-5.473e-02</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.299e-20</td>\n",
       "      <td>-0.028</td>\n",
       "      <td>0.009</td>\n",
       "      <td>3.004e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[10]</th>\n",
       "      <td>3.040e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>3.702e-06</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.014</td>\n",
       "      <td>6.575e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[11]</th>\n",
       "      <td>-1.869e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.492e-146</td>\n",
       "      <td>-0.005</td>\n",
       "      <td>0.012</td>\n",
       "      <td>6.557e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[12]</th>\n",
       "      <td>1.261e-02</td>\n",
       "      <td>0.018</td>\n",
       "      <td>4.875e-01</td>\n",
       "      <td>-0.062</td>\n",
       "      <td>0.026</td>\n",
       "      <td>1.588e-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[13]</th>\n",
       "      <td>-1.807e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.110</td>\n",
       "      <td>0.008</td>\n",
       "      <td>4.316e-46</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_n_mesh_ex == 0)[T.True]</th>\n",
       "      <td>-6.666e-01</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.511</td>\n",
       "      <td>0.024</td>\n",
       "      <td>5.006e-100</td>\n",
       "      <td>-0.628</td>\n",
       "      <td>0.036</td>\n",
       "      <td>1.575e-66</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(sink_n_mesh_ex == 0)[T.True]</th>\n",
       "      <td>-9.292e-02</td>\n",
       "      <td>0.015</td>\n",
       "      <td>1.355e-09</td>\n",
       "      <td>0.019</td>\n",
       "      <td>0.028</td>\n",
       "      <td>4.991e-01</td>\n",
       "      <td>-0.156</td>\n",
       "      <td>0.044</td>\n",
       "      <td>3.500e-04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(source_n_mesh_ex + 1)</th>\n",
       "      <td>-1.704e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>2.015e-212</td>\n",
       "      <td>-0.107</td>\n",
       "      <td>0.009</td>\n",
       "      <td>2.730e-30</td>\n",
       "      <td>-0.150</td>\n",
       "      <td>0.014</td>\n",
       "      <td>2.657e-26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(sink_n_mesh_ex + 1)</th>\n",
       "      <td>-1.681e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.170e-202</td>\n",
       "      <td>-0.143</td>\n",
       "      <td>0.010</td>\n",
       "      <td>1.055e-48</td>\n",
       "      <td>-0.256</td>\n",
       "      <td>0.015</td>\n",
       "      <td>8.514e-68</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(gender, levels=GENDERS)[T.F]</th>\n",
       "      <td>-2.852e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>1.196e-17</td>\n",
       "      <td>-0.052</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.845e-19</td>\n",
       "      <td>-0.033</td>\n",
       "      <td>0.012</td>\n",
       "      <td>6.140e-03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(gender, levels=GENDERS)[T.M]</th>\n",
       "      <td>-3.870e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>2.422e-36</td>\n",
       "      <td>-0.030</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.872e-08</td>\n",
       "      <td>-0.006</td>\n",
       "      <td>0.012</td>\n",
       "      <td>6.056e-01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                         Full           \\\n",
       "                                                        Coef. Std.Err.   \n",
       "Intercept                                          -1.957e+00    0.031   \n",
       "I(sink_prev_ncites == 0)[T.True]                    1.265e-01    0.004   \n",
       "np.log10(sink_prev_ncites + 1)                     -4.433e-01    0.006   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)             -1.698e-01    0.002   \n",
       "I(auth_prev_papers == 0)[T.True]                   -1.421e+00    0.009   \n",
       "I(auth_prev_papers == 1)[T.True]                   -2.357e-01    0.005   \n",
       "np.log10(auth_prev_papers + 1)                      1.610e+00    0.007   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)             -9.062e-02    0.002   \n",
       "source_is_journal[T.True]                           2.661e-01    0.011   \n",
       "source_is_review[T.True]                           -7.869e-02    0.003   \n",
       "source_is_case_rep[T.True]                         -9.404e-01    0.007   \n",
       "source_is_let_ed_com[T.True]                       -4.295e-01    0.011   \n",
       "sink_is_journal[T.True]                             3.283e-01    0.011   \n",
       "sink_is_review[T.True]                             -7.826e-01    0.003   \n",
       "sink_is_case_rep[T.True]                           -6.369e-01    0.006   \n",
       "sink_is_let_ed_com[T.True]                         -2.838e-01    0.011   \n",
       "I(year_span < 0)[T.True]                           -5.564e-01    0.016   \n",
       "I(year_span == 0)[T.True]                           5.382e-01    0.006   \n",
       "mf.score_log_1(year_span)                           5.387e-01    0.015   \n",
       "I(mf.score_log_1(year_span) ** 2)                  -1.281e+00    0.010   \n",
       "journal_same[T.True]                                4.567e-01    0.003   \n",
       "I(jj_sim == 0)[T.True]                             -2.805e-01    0.008   \n",
       "np.log10(jj_sim + 1)                                6.298e-02    0.010   \n",
       "I(np.log10(jj_sim + 1) ** 2)                        5.949e-03    0.003   \n",
       "I(source_ncites == 1)[T.True]                      -2.997e-01    0.022   \n",
       "np.log10(source_ncites)                            -1.647e+00    0.034   \n",
       "I(np.log10(source_ncites) ** 2)                     8.198e-01    0.023   \n",
       "I(np.log10(source_ncites) ** 3)                    -2.294e-01    0.005   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN... -4.468e-02    0.006   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]   -5.242e-03    0.003   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...  2.669e-01    0.006   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...  2.382e-02    0.004   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR... -4.425e-02    0.005   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT... -1.010e-01    0.007   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA... -3.874e-03    0.004   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH... -4.120e-01    0.010   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...  8.788e-02    0.005   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP... -1.223e-01    0.006   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...  6.808e-02    0.007   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...  1.074e-01    0.007   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN... -4.363e-02    0.010   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT... -6.440e-02    0.003   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)      -1.058e-01    0.001   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)         9.484e-02    0.008   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *... -2.774e-02    0.001   \n",
       "source_is_eng[T.True]                               6.494e-01    0.007   \n",
       "sink_is_eng[T.True]                                -1.846e-01    0.010   \n",
       "I(source_n_authors > 20)[T.True]                    1.722e-04    0.015   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))          8.239e-02    0.018   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2) -2.563e-01    0.013   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.426e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  5.313e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.562e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.021e-01    0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.401e-02    0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -7.411e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -5.946e-02    0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -6.379e-02    0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.355e-01    0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -5.473e-02    0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.040e-02    0.007   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.869e-01    0.007   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.261e-02    0.018   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.807e-01    0.004   \n",
       "I(source_n_mesh_ex == 0)[T.True]                   -6.666e-01    0.013   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                     -9.292e-02    0.015   \n",
       "np.log10(source_n_mesh_ex + 1)                     -1.704e-01    0.005   \n",
       "np.log10(sink_n_mesh_ex + 1)                       -1.681e-01    0.006   \n",
       "C(gender, levels=GENDERS)[T.F]                     -2.852e-02    0.003   \n",
       "C(gender, levels=GENDERS)[T.M]                     -3.870e-02    0.003   \n",
       "\n",
       "                                                               Filtered 1  \\\n",
       "                                                         P>|z|      Coef.   \n",
       "Intercept                                            0.000e+00     -4.047   \n",
       "I(sink_prev_ncites == 0)[T.True]                    1.064e-235      0.166   \n",
       "np.log10(sink_prev_ncites + 1)                       0.000e+00     -0.147   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)               0.000e+00     -0.226   \n",
       "I(auth_prev_papers == 0)[T.True]                     0.000e+00     -1.232   \n",
       "I(auth_prev_papers == 1)[T.True]                     0.000e+00     -0.159   \n",
       "np.log10(auth_prev_papers + 1)                       0.000e+00      1.862   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)              1.091e-294     -0.127   \n",
       "source_is_journal[T.True]                           6.473e-129        NaN   \n",
       "source_is_review[T.True]                            5.938e-199        NaN   \n",
       "source_is_case_rep[T.True]                           0.000e+00        NaN   \n",
       "source_is_let_ed_com[T.True]                         0.000e+00        NaN   \n",
       "sink_is_journal[T.True]                             1.814e-181      0.548   \n",
       "sink_is_review[T.True]                               0.000e+00     -0.741   \n",
       "sink_is_case_rep[T.True]                             0.000e+00     -0.882   \n",
       "sink_is_let_ed_com[T.True]                          9.580e-138     -0.421   \n",
       "I(year_span < 0)[T.True]                            2.011e-275     -0.445   \n",
       "I(year_span == 0)[T.True]                            0.000e+00      0.446   \n",
       "mf.score_log_1(year_span)                           2.008e-275     -0.204   \n",
       "I(mf.score_log_1(year_span) ** 2)                    0.000e+00     -0.767   \n",
       "journal_same[T.True]                                 0.000e+00      0.509   \n",
       "I(jj_sim == 0)[T.True]                              1.735e-259     -0.172   \n",
       "np.log10(jj_sim + 1)                                 3.421e-10      0.205   \n",
       "I(np.log10(jj_sim + 1) ** 2)                         4.459e-02     -0.044   \n",
       "I(source_ncites == 1)[T.True]                        1.212e-41        NaN   \n",
       "np.log10(source_ncites)                              0.000e+00     -3.223   \n",
       "I(np.log10(source_ncites) ** 2)                     4.438e-269      2.579   \n",
       "I(np.log10(source_ncites) ** 3)                      0.000e+00     -0.771   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...   3.435e-12        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]     1.058e-01        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...   0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...   9.888e-09        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...   3.363e-17        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...   3.885e-54        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...   3.608e-01        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...   0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...   1.664e-63        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...   1.850e-83        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...   1.152e-21        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...   4.490e-58        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...   3.026e-05        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...  8.308e-128        NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)        0.000e+00     -0.096   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          1.753e-35      0.200   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...  1.412e-100     -0.044   \n",
       "source_is_eng[T.True]                                0.000e+00        NaN   \n",
       "sink_is_eng[T.True]                                  6.033e-81      2.075   \n",
       "I(source_n_authors > 20)[T.True]                     9.906e-01     -0.037   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))           3.324e-06      0.242   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   1.194e-85     -0.336   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.087e-04     -0.033   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   2.200e-36     -0.038   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   5.136e-29      0.180   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.235e-60     -0.147   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.985e-22      0.116   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.704e-63     -0.066   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   7.073e-23     -0.092   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   5.197e-34      0.013   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.099e-172      0.030   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.299e-20     -0.028   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   3.702e-06      0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  2.492e-146     -0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   4.875e-01     -0.062   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   0.000e+00     -0.110   \n",
       "I(source_n_mesh_ex == 0)[T.True]                     0.000e+00     -0.511   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                       1.355e-09      0.019   \n",
       "np.log10(source_n_mesh_ex + 1)                      2.015e-212     -0.107   \n",
       "np.log10(sink_n_mesh_ex + 1)                        3.170e-202     -0.143   \n",
       "C(gender, levels=GENDERS)[T.F]                       1.196e-17     -0.052   \n",
       "C(gender, levels=GENDERS)[T.M]                       2.422e-36     -0.030   \n",
       "\n",
       "                                                                         \\\n",
       "                                                   Std.Err.       P>|z|   \n",
       "Intercept                                             0.289   2.037e-44   \n",
       "I(sink_prev_ncites == 0)[T.True]                      0.007  4.064e-120   \n",
       "np.log10(sink_prev_ncites + 1)                        0.011   7.272e-39   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                0.004   0.000e+00   \n",
       "I(auth_prev_papers == 0)[T.True]                      0.018   0.000e+00   \n",
       "I(auth_prev_papers == 1)[T.True]                      0.010   2.285e-58   \n",
       "np.log10(auth_prev_papers + 1)                        0.013   0.000e+00   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                0.004  9.929e-187   \n",
       "source_is_journal[T.True]                               NaN         NaN   \n",
       "source_is_review[T.True]                                NaN         NaN   \n",
       "source_is_case_rep[T.True]                              NaN         NaN   \n",
       "source_is_let_ed_com[T.True]                            NaN         NaN   \n",
       "sink_is_journal[T.True]                               0.022  5.741e-140   \n",
       "sink_is_review[T.True]                                0.005   0.000e+00   \n",
       "sink_is_case_rep[T.True]                              0.012   0.000e+00   \n",
       "sink_is_let_ed_com[T.True]                            0.021   1.665e-86   \n",
       "I(year_span < 0)[T.True]                              0.028   1.636e-55   \n",
       "I(year_span == 0)[T.True]                             0.010   0.000e+00   \n",
       "mf.score_log_1(year_span)                             0.026   6.462e-15   \n",
       "I(mf.score_log_1(year_span) ** 2)                     0.016   0.000e+00   \n",
       "journal_same[T.True]                                  0.005   0.000e+00   \n",
       "I(jj_sim == 0)[T.True]                                0.015   4.417e-31   \n",
       "np.log10(jj_sim + 1)                                  0.019   3.512e-28   \n",
       "I(np.log10(jj_sim + 1) ** 2)                          0.006   3.126e-15   \n",
       "I(source_ncites == 1)[T.True]                           NaN         NaN   \n",
       "np.log10(source_ncites)                               0.562   1.002e-08   \n",
       "I(np.log10(source_ncites) ** 2)                       0.379   9.953e-12   \n",
       "I(np.log10(source_ncites) ** 3)                       0.084   3.470e-20   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]        NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...      NaN         NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)         0.002   0.000e+00   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)           0.014   2.878e-49   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...    0.002   7.102e-82   \n",
       "source_is_eng[T.True]                                   NaN         NaN   \n",
       "sink_is_eng[T.True]                                   0.084  2.546e-133   \n",
       "I(source_n_authors > 20)[T.True]                      0.025   1.343e-01   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))            0.030   1.357e-15   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)    0.023   3.234e-50   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.006   2.974e-08   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   1.021e-07   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.006  9.166e-200   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.009   7.110e-61   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   1.022e-50   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   4.727e-16   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.010   7.999e-20   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   4.308e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.012   8.899e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.009   3.004e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.014   6.575e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.012   6.557e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.026   1.588e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   4.316e-46   \n",
       "I(source_n_mesh_ex == 0)[T.True]                      0.024  5.006e-100   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                        0.028   4.991e-01   \n",
       "np.log10(source_n_mesh_ex + 1)                        0.009   2.730e-30   \n",
       "np.log10(sink_n_mesh_ex + 1)                          0.010   1.055e-48   \n",
       "C(gender, levels=GENDERS)[T.F]                        0.006   1.845e-19   \n",
       "C(gender, levels=GENDERS)[T.M]                        0.005   1.872e-08   \n",
       "\n",
       "                                                   Filtered 2           \\\n",
       "                                                        Coef. Std.Err.   \n",
       "Intercept                                              -3.542    0.482   \n",
       "I(sink_prev_ncites == 0)[T.True]                        0.156    0.011   \n",
       "np.log10(sink_prev_ncites + 1)                         -0.137    0.016   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                 -0.181    0.006   \n",
       "I(auth_prev_papers == 0)[T.True]                       -1.236    0.030   \n",
       "I(auth_prev_papers == 1)[T.True]                       -0.158    0.017   \n",
       "np.log10(auth_prev_papers + 1)                          1.911    0.020   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                 -0.156    0.006   \n",
       "source_is_journal[T.True]                                 NaN      NaN   \n",
       "source_is_review[T.True]                                  NaN      NaN   \n",
       "source_is_case_rep[T.True]                                NaN      NaN   \n",
       "source_is_let_ed_com[T.True]                              NaN      NaN   \n",
       "sink_is_journal[T.True]                                 0.496    0.032   \n",
       "sink_is_review[T.True]                                 -0.639    0.007   \n",
       "sink_is_case_rep[T.True]                               -0.861    0.017   \n",
       "sink_is_let_ed_com[T.True]                             -0.451    0.031   \n",
       "I(year_span < 0)[T.True]                               -0.327    0.041   \n",
       "I(year_span == 0)[T.True]                               0.580    0.015   \n",
       "mf.score_log_1(year_span)                               0.152    0.038   \n",
       "I(mf.score_log_1(year_span) ** 2)                      -0.857    0.023   \n",
       "journal_same[T.True]                                    0.501    0.007   \n",
       "I(jj_sim == 0)[T.True]                                 -0.163    0.022   \n",
       "np.log10(jj_sim + 1)                                    0.255    0.027   \n",
       "I(np.log10(jj_sim + 1) ** 2)                           -0.056    0.008   \n",
       "I(source_ncites == 1)[T.True]                             NaN      NaN   \n",
       "np.log10(source_ncites)                                -7.395    0.816   \n",
       "I(np.log10(source_ncites) ** 2)                         5.400    0.550   \n",
       "I(np.log10(source_ncites) ** 3)                        -1.386    0.122   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]          NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...        NaN      NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)          -0.112    0.004   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)             0.158    0.021   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...     -0.037    0.004   \n",
       "source_is_eng[T.True]                                     NaN      NaN   \n",
       "sink_is_eng[T.True]                                     3.625    0.268   \n",
       "I(source_n_authors > 20)[T.True]                       -0.047    0.034   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))              0.303    0.045   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)     -0.353    0.033   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "I(source_n_mesh_ex == 0)[T.True]                       -0.628    0.036   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                         -0.156    0.044   \n",
       "np.log10(source_n_mesh_ex + 1)                         -0.150    0.014   \n",
       "np.log10(sink_n_mesh_ex + 1)                           -0.256    0.015   \n",
       "C(gender, levels=GENDERS)[T.F]                         -0.033    0.012   \n",
       "C(gender, levels=GENDERS)[T.M]                         -0.006    0.012   \n",
       "\n",
       "                                                                \n",
       "                                                         P>|z|  \n",
       "Intercept                                            1.978e-13  \n",
       "I(sink_prev_ncites == 0)[T.True]                     2.378e-47  \n",
       "np.log10(sink_prev_ncites + 1)                       5.717e-17  \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)              2.708e-203  \n",
       "I(auth_prev_papers == 0)[T.True]                     0.000e+00  \n",
       "I(auth_prev_papers == 1)[T.True]                     1.823e-21  \n",
       "np.log10(auth_prev_papers + 1)                       0.000e+00  \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)              4.687e-130  \n",
       "source_is_journal[T.True]                                  NaN  \n",
       "source_is_review[T.True]                                   NaN  \n",
       "source_is_case_rep[T.True]                                 NaN  \n",
       "source_is_let_ed_com[T.True]                               NaN  \n",
       "sink_is_journal[T.True]                              3.460e-55  \n",
       "sink_is_review[T.True]                               0.000e+00  \n",
       "sink_is_case_rep[T.True]                             0.000e+00  \n",
       "sink_is_let_ed_com[T.True]                           7.459e-47  \n",
       "I(year_span < 0)[T.True]                             2.789e-15  \n",
       "I(year_span == 0)[T.True]                            0.000e+00  \n",
       "mf.score_log_1(year_span)                            7.348e-05  \n",
       "I(mf.score_log_1(year_span) ** 2)                   1.099e-302  \n",
       "journal_same[T.True]                                 0.000e+00  \n",
       "I(jj_sim == 0)[T.True]                               1.448e-13  \n",
       "np.log10(jj_sim + 1)                                 1.016e-20  \n",
       "I(np.log10(jj_sim + 1) ** 2)                         8.183e-12  \n",
       "I(source_ncites == 1)[T.True]                              NaN  \n",
       "np.log10(source_ncites)                              1.230e-19  \n",
       "I(np.log10(source_ncites) ** 2)                      1.023e-22  \n",
       "I(np.log10(source_ncites) ** 3)                      5.992e-30  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]           NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...         NaN  \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)       1.108e-214  \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          2.289e-14  \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...   6.418e-26  \n",
       "source_is_eng[T.True]                                      NaN  \n",
       "sink_is_eng[T.True]                                  1.149e-41  \n",
       "I(source_n_authors > 20)[T.True]                     1.627e-01  \n",
       "np.log10(np.clip(source_n_authors, 0, 20))           1.063e-11  \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   2.710e-26  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "I(source_n_mesh_ex == 0)[T.True]                     1.575e-66  \n",
       "I(sink_n_mesh_ex == 0)[T.True]                       3.500e-04  \n",
       "np.log10(source_n_mesh_ex + 1)                       2.657e-26  \n",
       "np.log10(sink_n_mesh_ex + 1)                         8.514e-68  \n",
       "C(gender, levels=GENDERS)[T.F]                       6.140e-03  \n",
       "C(gender, levels=GENDERS)[T.M]                       6.056e-01  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_t = merge_results([results_full, results_flt_1, results_flt_2])\n",
    "with pd.option_context(\"display.max_columns\", 20, \"display.max_rows\", 200, \"precision\", 3):\n",
    "    display(df_t)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Last author"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "MODEL_VERSION = \"v4\"\n",
    "MODEL_SUFFIX = \"%s.last_author\" % MODEL_VERSION\n",
    "\n",
    "results_full_la = load(\"results_all_model/%s/results_all_model.%s.pkl\" % (MODEL_VERSION, MODEL_SUFFIX))\n",
    "results_flt_1_la = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" % (MODEL_VERSION, MODEL_SUFFIX, 1))\n",
    "results_flt_2_la = load(\"results_all_model/%s/results_all_model.%s.filtered_%s.pkl\" % (MODEL_VERSION, MODEL_SUFFIX, 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"3\" halign=\"left\">Full</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 1</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Intercept</th>\n",
       "      <td>-2.790</td>\n",
       "      <td>2.674e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>8.550e-01</td>\n",
       "      <td>0.488</td>\n",
       "      <td>7.964e-02</td>\n",
       "      <td>-0.491</td>\n",
       "      <td>0.695</td>\n",
       "      <td>4.799e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(year_span &lt; 0)[T.True]</th>\n",
       "      <td>-0.383</td>\n",
       "      <td>1.459e-02</td>\n",
       "      <td>4.270e-152</td>\n",
       "      <td>-3.821e-01</td>\n",
       "      <td>0.032</td>\n",
       "      <td>9.889e-33</td>\n",
       "      <td>-0.394</td>\n",
       "      <td>0.044</td>\n",
       "      <td>2.207e-19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(year_span == 0)[T.True]</th>\n",
       "      <td>0.735</td>\n",
       "      <td>4.656e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>7.146e-01</td>\n",
       "      <td>0.009</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.748</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.score_log_1(year_span)</th>\n",
       "      <td>1.232</td>\n",
       "      <td>1.131e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>8.259e-01</td>\n",
       "      <td>0.022</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.856</td>\n",
       "      <td>0.029</td>\n",
       "      <td>2.733e-190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(mf.score_log_1(year_span) ** 2)</th>\n",
       "      <td>-1.402</td>\n",
       "      <td>6.820e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.160e+00</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.155</td>\n",
       "      <td>0.017</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(auth_prev_papers == 0)[T.True]</th>\n",
       "      <td>-1.232</td>\n",
       "      <td>1.639e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.259e+00</td>\n",
       "      <td>0.039</td>\n",
       "      <td>7.020e-231</td>\n",
       "      <td>-1.236</td>\n",
       "      <td>0.054</td>\n",
       "      <td>3.506e-114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(auth_prev_papers == 1)[T.True]</th>\n",
       "      <td>-0.187</td>\n",
       "      <td>1.058e-02</td>\n",
       "      <td>3.454e-70</td>\n",
       "      <td>-2.089e-01</td>\n",
       "      <td>0.024</td>\n",
       "      <td>5.352e-18</td>\n",
       "      <td>-0.189</td>\n",
       "      <td>0.035</td>\n",
       "      <td>6.216e-08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(auth_prev_papers + 1)</th>\n",
       "      <td>2.172</td>\n",
       "      <td>7.733e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>2.458e+00</td>\n",
       "      <td>0.016</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>2.468</td>\n",
       "      <td>0.022</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(auth_prev_papers + 1) ** 2)</th>\n",
       "      <td>-0.340</td>\n",
       "      <td>2.154e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-4.035e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.401</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_journal[T.True]</th>\n",
       "      <td>0.229</td>\n",
       "      <td>1.021e-02</td>\n",
       "      <td>1.707e-111</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_review[T.True]</th>\n",
       "      <td>-0.249</td>\n",
       "      <td>2.125e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_case_rep[T.True]</th>\n",
       "      <td>-0.948</td>\n",
       "      <td>5.222e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_let_ed_com[T.True]</th>\n",
       "      <td>-0.343</td>\n",
       "      <td>9.952e-03</td>\n",
       "      <td>9.331e-261</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_journal[T.True]</th>\n",
       "      <td>0.324</td>\n",
       "      <td>9.445e-03</td>\n",
       "      <td>1.501e-257</td>\n",
       "      <td>5.370e-01</td>\n",
       "      <td>0.020</td>\n",
       "      <td>3.782e-163</td>\n",
       "      <td>0.575</td>\n",
       "      <td>0.027</td>\n",
       "      <td>8.207e-101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_review[T.True]</th>\n",
       "      <td>-0.641</td>\n",
       "      <td>2.142e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-6.616e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.641</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_case_rep[T.True]</th>\n",
       "      <td>-0.731</td>\n",
       "      <td>4.949e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.038e+00</td>\n",
       "      <td>0.011</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.046</td>\n",
       "      <td>0.014</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_let_ed_com[T.True]</th>\n",
       "      <td>-0.292</td>\n",
       "      <td>9.374e-03</td>\n",
       "      <td>3.266e-213</td>\n",
       "      <td>-3.867e-01</td>\n",
       "      <td>0.019</td>\n",
       "      <td>9.806e-90</td>\n",
       "      <td>-0.373</td>\n",
       "      <td>0.026</td>\n",
       "      <td>5.465e-46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>journal_same[T.True]</th>\n",
       "      <td>0.534</td>\n",
       "      <td>2.218e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>6.124e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.616</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(jj_sim == 0)[T.True]</th>\n",
       "      <td>-0.296</td>\n",
       "      <td>6.332e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.677e-01</td>\n",
       "      <td>0.013</td>\n",
       "      <td>2.733e-40</td>\n",
       "      <td>-0.188</td>\n",
       "      <td>0.017</td>\n",
       "      <td>8.701e-29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(jj_sim + 1)</th>\n",
       "      <td>0.126</td>\n",
       "      <td>8.068e-03</td>\n",
       "      <td>4.277e-55</td>\n",
       "      <td>3.594e-01</td>\n",
       "      <td>0.016</td>\n",
       "      <td>1.723e-107</td>\n",
       "      <td>0.358</td>\n",
       "      <td>0.022</td>\n",
       "      <td>2.322e-61</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(jj_sim + 1) ** 2)</th>\n",
       "      <td>-0.050</td>\n",
       "      <td>2.456e-03</td>\n",
       "      <td>1.148e-91</td>\n",
       "      <td>-1.456e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>3.399e-185</td>\n",
       "      <td>-0.143</td>\n",
       "      <td>0.007</td>\n",
       "      <td>5.665e-103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(sink_prev_ncites == 0)[T.True]</th>\n",
       "      <td>0.240</td>\n",
       "      <td>3.101e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>2.968e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.292</td>\n",
       "      <td>0.009</td>\n",
       "      <td>2.798e-249</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(sink_prev_ncites + 1)</th>\n",
       "      <td>-0.079</td>\n",
       "      <td>4.424e-03</td>\n",
       "      <td>4.191e-71</td>\n",
       "      <td>1.991e-01</td>\n",
       "      <td>0.009</td>\n",
       "      <td>6.238e-119</td>\n",
       "      <td>0.215</td>\n",
       "      <td>0.012</td>\n",
       "      <td>2.376e-77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(sink_prev_ncites + 1) ** 2)</th>\n",
       "      <td>-0.169</td>\n",
       "      <td>1.530e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-2.088e-01</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.201</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_ncites == 1)[T.True]</th>\n",
       "      <td>-0.405</td>\n",
       "      <td>2.079e-02</td>\n",
       "      <td>2.477e-84</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(source_ncites)</th>\n",
       "      <td>-2.221</td>\n",
       "      <td>3.101e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.776e+01</td>\n",
       "      <td>1.055</td>\n",
       "      <td>1.635e-63</td>\n",
       "      <td>-17.484</td>\n",
       "      <td>1.420</td>\n",
       "      <td>7.628e-35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(source_ncites) ** 2)</th>\n",
       "      <td>1.463</td>\n",
       "      <td>2.148e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>1.354e+01</td>\n",
       "      <td>0.772</td>\n",
       "      <td>6.040e-69</td>\n",
       "      <td>13.420</td>\n",
       "      <td>1.039</td>\n",
       "      <td>3.616e-38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(source_ncites) ** 3)</th>\n",
       "      <td>-0.387</td>\n",
       "      <td>4.793e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-3.438e+00</td>\n",
       "      <td>0.186</td>\n",
       "      <td>4.052e-76</td>\n",
       "      <td>-3.422</td>\n",
       "      <td>0.251</td>\n",
       "      <td>2.177e-42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.UNKNOWN]</th>\n",
       "      <td>-0.158</td>\n",
       "      <td>5.211e-03</td>\n",
       "      <td>3.526e-203</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.UK]</th>\n",
       "      <td>-0.090</td>\n",
       "      <td>2.373e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.JAPAN]</th>\n",
       "      <td>-0.092</td>\n",
       "      <td>6.048e-03</td>\n",
       "      <td>9.138e-53</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.GERMANY]</th>\n",
       "      <td>-0.169</td>\n",
       "      <td>3.061e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.FRANCE]</th>\n",
       "      <td>-0.116</td>\n",
       "      <td>3.820e-03</td>\n",
       "      <td>1.689e-203</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.ITALY]</th>\n",
       "      <td>-0.298</td>\n",
       "      <td>4.877e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.CANADA]</th>\n",
       "      <td>-0.069</td>\n",
       "      <td>3.035e-03</td>\n",
       "      <td>3.086e-113</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.CHINA]</th>\n",
       "      <td>-0.432</td>\n",
       "      <td>7.300e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.AUSTRALIA]</th>\n",
       "      <td>-0.099</td>\n",
       "      <td>4.043e-03</td>\n",
       "      <td>4.252e-132</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.SPAIN]</th>\n",
       "      <td>-0.184</td>\n",
       "      <td>4.889e-03</td>\n",
       "      <td>6.632e-310</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.NETHERLANDS]</th>\n",
       "      <td>-0.179</td>\n",
       "      <td>5.204e-03</td>\n",
       "      <td>5.058e-260</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.SWEDEN]</th>\n",
       "      <td>-0.083</td>\n",
       "      <td>4.978e-03</td>\n",
       "      <td>6.425e-63</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.INDIA]</th>\n",
       "      <td>-0.206</td>\n",
       "      <td>8.064e-03</td>\n",
       "      <td>1.078e-143</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.OTHER]</th>\n",
       "      <td>-0.213</td>\n",
       "      <td>2.054e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.nan_to_num(source_V_novelty) + 1)</th>\n",
       "      <td>-0.139</td>\n",
       "      <td>9.809e-04</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.345e-01</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.157</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.nan_to_num(sink_V_novelty) + 1)</th>\n",
       "      <td>0.048</td>\n",
       "      <td>5.268e-03</td>\n",
       "      <td>5.457e-20</td>\n",
       "      <td>1.570e-01</td>\n",
       "      <td>0.010</td>\n",
       "      <td>1.388e-52</td>\n",
       "      <td>0.199</td>\n",
       "      <td>0.014</td>\n",
       "      <td>3.391e-45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)</th>\n",
       "      <td>-0.022</td>\n",
       "      <td>9.206e-04</td>\n",
       "      <td>1.213e-127</td>\n",
       "      <td>-4.152e-02</td>\n",
       "      <td>0.002</td>\n",
       "      <td>3.386e-118</td>\n",
       "      <td>-0.049</td>\n",
       "      <td>0.002</td>\n",
       "      <td>9.311e-89</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_n_authors &gt; 20)[T.True]</th>\n",
       "      <td>-0.050</td>\n",
       "      <td>1.263e-02</td>\n",
       "      <td>7.036e-05</td>\n",
       "      <td>2.323e-02</td>\n",
       "      <td>0.024</td>\n",
       "      <td>3.258e-01</td>\n",
       "      <td>-0.037</td>\n",
       "      <td>0.032</td>\n",
       "      <td>2.452e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.clip(source_n_authors, 0, 20))</th>\n",
       "      <td>-0.444</td>\n",
       "      <td>1.368e-02</td>\n",
       "      <td>1.041e-230</td>\n",
       "      <td>-1.366e-01</td>\n",
       "      <td>0.026</td>\n",
       "      <td>1.286e-07</td>\n",
       "      <td>-0.183</td>\n",
       "      <td>0.035</td>\n",
       "      <td>1.756e-07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)</th>\n",
       "      <td>0.066</td>\n",
       "      <td>1.004e-02</td>\n",
       "      <td>3.481e-11</td>\n",
       "      <td>-1.128e-01</td>\n",
       "      <td>0.019</td>\n",
       "      <td>4.115e-09</td>\n",
       "      <td>-0.087</td>\n",
       "      <td>0.026</td>\n",
       "      <td>8.039e-04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_eng[T.True]</th>\n",
       "      <td>0.701</td>\n",
       "      <td>6.660e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_eng[T.True]</th>\n",
       "      <td>-0.031</td>\n",
       "      <td>8.947e-03</td>\n",
       "      <td>5.789e-04</td>\n",
       "      <td>2.656e+00</td>\n",
       "      <td>0.102</td>\n",
       "      <td>4.895e-149</td>\n",
       "      <td>3.959</td>\n",
       "      <td>0.268</td>\n",
       "      <td>1.711e-49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_n_mesh_ex == 0)[T.True]</th>\n",
       "      <td>-0.892</td>\n",
       "      <td>9.581e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-7.059e-01</td>\n",
       "      <td>0.020</td>\n",
       "      <td>6.641e-269</td>\n",
       "      <td>-0.908</td>\n",
       "      <td>0.028</td>\n",
       "      <td>3.226e-237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(sink_n_mesh_ex == 0)[T.True]</th>\n",
       "      <td>0.005</td>\n",
       "      <td>1.108e-02</td>\n",
       "      <td>6.763e-01</td>\n",
       "      <td>-9.296e-02</td>\n",
       "      <td>0.022</td>\n",
       "      <td>2.792e-05</td>\n",
       "      <td>-0.106</td>\n",
       "      <td>0.030</td>\n",
       "      <td>4.778e-04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(source_n_mesh_ex + 1)</th>\n",
       "      <td>-0.251</td>\n",
       "      <td>4.138e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.537e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>7.027e-84</td>\n",
       "      <td>-0.239</td>\n",
       "      <td>0.011</td>\n",
       "      <td>2.737e-109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(sink_n_mesh_ex + 1)</th>\n",
       "      <td>-0.092</td>\n",
       "      <td>4.139e-03</td>\n",
       "      <td>1.767e-108</td>\n",
       "      <td>-1.789e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.370e-112</td>\n",
       "      <td>-0.230</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.515e-101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[0]</th>\n",
       "      <td>0.062</td>\n",
       "      <td>2.555e-03</td>\n",
       "      <td>6.084e-130</td>\n",
       "      <td>5.838e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>5.079e-36</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[1]</th>\n",
       "      <td>0.082</td>\n",
       "      <td>3.304e-03</td>\n",
       "      <td>1.645e-134</td>\n",
       "      <td>-2.033e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.542e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[2]</th>\n",
       "      <td>0.034</td>\n",
       "      <td>3.646e-03</td>\n",
       "      <td>4.623e-21</td>\n",
       "      <td>1.815e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.474e-188</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[3]</th>\n",
       "      <td>-0.048</td>\n",
       "      <td>5.959e-03</td>\n",
       "      <td>6.342e-16</td>\n",
       "      <td>1.517e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.340e-43</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[4]</th>\n",
       "      <td>0.032</td>\n",
       "      <td>4.138e-03</td>\n",
       "      <td>4.544e-15</td>\n",
       "      <td>1.617e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.828e-83</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[5]</th>\n",
       "      <td>-0.038</td>\n",
       "      <td>3.142e-03</td>\n",
       "      <td>6.573e-33</td>\n",
       "      <td>-2.034e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.611e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[6]</th>\n",
       "      <td>0.026</td>\n",
       "      <td>4.390e-03</td>\n",
       "      <td>5.446e-09</td>\n",
       "      <td>5.540e-02</td>\n",
       "      <td>0.008</td>\n",
       "      <td>3.340e-11</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[7]</th>\n",
       "      <td>0.020</td>\n",
       "      <td>4.486e-03</td>\n",
       "      <td>6.955e-06</td>\n",
       "      <td>7.448e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.266e-30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[8]</th>\n",
       "      <td>0.070</td>\n",
       "      <td>3.496e-03</td>\n",
       "      <td>2.459e-89</td>\n",
       "      <td>9.378e-02</td>\n",
       "      <td>0.009</td>\n",
       "      <td>1.986e-24</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[9]</th>\n",
       "      <td>-0.085</td>\n",
       "      <td>5.591e-03</td>\n",
       "      <td>8.268e-52</td>\n",
       "      <td>-3.558e-02</td>\n",
       "      <td>0.009</td>\n",
       "      <td>1.047e-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[10]</th>\n",
       "      <td>-0.004</td>\n",
       "      <td>4.629e-03</td>\n",
       "      <td>4.259e-01</td>\n",
       "      <td>-2.958e-02</td>\n",
       "      <td>0.011</td>\n",
       "      <td>6.362e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[11]</th>\n",
       "      <td>-0.076</td>\n",
       "      <td>6.431e-03</td>\n",
       "      <td>1.624e-32</td>\n",
       "      <td>9.743e-02</td>\n",
       "      <td>0.016</td>\n",
       "      <td>2.776e-09</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[12]</th>\n",
       "      <td>-0.046</td>\n",
       "      <td>1.422e-02</td>\n",
       "      <td>1.126e-03</td>\n",
       "      <td>-1.005e-01</td>\n",
       "      <td>0.023</td>\n",
       "      <td>7.966e-06</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[13]</th>\n",
       "      <td>-0.057</td>\n",
       "      <td>3.398e-03</td>\n",
       "      <td>5.873e-63</td>\n",
       "      <td>6.383e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>9.465e-22</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(gender, levels=GENDERS)[T.F]</th>\n",
       "      <td>0.026</td>\n",
       "      <td>2.934e-03</td>\n",
       "      <td>7.435e-19</td>\n",
       "      <td>3.812e-04</td>\n",
       "      <td>0.006</td>\n",
       "      <td>9.488e-01</td>\n",
       "      <td>0.034</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.590e-03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(gender, levels=GENDERS)[T.M]</th>\n",
       "      <td>-0.002</td>\n",
       "      <td>2.597e-03</td>\n",
       "      <td>3.487e-01</td>\n",
       "      <td>-1.142e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>3.326e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.010</td>\n",
       "      <td>6.570e-01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                     Full             \\\n",
       "                                                    Coef.   Std.Err.   \n",
       "Intercept                                          -2.790  2.674e-02   \n",
       "I(year_span < 0)[T.True]                           -0.383  1.459e-02   \n",
       "I(year_span == 0)[T.True]                           0.735  4.656e-03   \n",
       "mf.score_log_1(year_span)                           1.232  1.131e-02   \n",
       "I(mf.score_log_1(year_span) ** 2)                  -1.402  6.820e-03   \n",
       "I(auth_prev_papers == 0)[T.True]                   -1.232  1.639e-02   \n",
       "I(auth_prev_papers == 1)[T.True]                   -0.187  1.058e-02   \n",
       "np.log10(auth_prev_papers + 1)                      2.172  7.733e-03   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)             -0.340  2.154e-03   \n",
       "source_is_journal[T.True]                           0.229  1.021e-02   \n",
       "source_is_review[T.True]                           -0.249  2.125e-03   \n",
       "source_is_case_rep[T.True]                         -0.948  5.222e-03   \n",
       "source_is_let_ed_com[T.True]                       -0.343  9.952e-03   \n",
       "sink_is_journal[T.True]                             0.324  9.445e-03   \n",
       "sink_is_review[T.True]                             -0.641  2.142e-03   \n",
       "sink_is_case_rep[T.True]                           -0.731  4.949e-03   \n",
       "sink_is_let_ed_com[T.True]                         -0.292  9.374e-03   \n",
       "journal_same[T.True]                                0.534  2.218e-03   \n",
       "I(jj_sim == 0)[T.True]                             -0.296  6.332e-03   \n",
       "np.log10(jj_sim + 1)                                0.126  8.068e-03   \n",
       "I(np.log10(jj_sim + 1) ** 2)                       -0.050  2.456e-03   \n",
       "I(sink_prev_ncites == 0)[T.True]                    0.240  3.101e-03   \n",
       "np.log10(sink_prev_ncites + 1)                     -0.079  4.424e-03   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)             -0.169  1.530e-03   \n",
       "I(source_ncites == 1)[T.True]                      -0.405  2.079e-02   \n",
       "np.log10(source_ncites)                            -2.221  3.101e-02   \n",
       "I(np.log10(source_ncites) ** 2)                     1.463  2.148e-02   \n",
       "I(np.log10(source_ncites) ** 3)                    -0.387  4.793e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN... -0.158  5.211e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]   -0.090  2.373e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA... -0.092  6.048e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE... -0.169  3.061e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR... -0.116  3.820e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT... -0.298  4.877e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA... -0.069  3.035e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH... -0.432  7.300e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU... -0.099  4.043e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP... -0.184  4.889e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE... -0.179  5.204e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW... -0.083  4.978e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN... -0.206  8.064e-03   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT... -0.213  2.054e-03   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)      -0.139  9.809e-04   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)         0.048  5.268e-03   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *... -0.022  9.206e-04   \n",
       "I(source_n_authors > 20)[T.True]                   -0.050  1.263e-02   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))         -0.444  1.368e-02   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)  0.066  1.004e-02   \n",
       "source_is_eng[T.True]                               0.701  6.660e-03   \n",
       "sink_is_eng[T.True]                                -0.031  8.947e-03   \n",
       "I(source_n_mesh_ex == 0)[T.True]                   -0.892  9.581e-03   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                      0.005  1.108e-02   \n",
       "np.log10(source_n_mesh_ex + 1)                     -0.251  4.138e-03   \n",
       "np.log10(sink_n_mesh_ex + 1)                       -0.092  4.139e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  0.062  2.555e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  0.082  3.304e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  0.034  3.646e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -0.048  5.959e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  0.032  4.138e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -0.038  3.142e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  0.026  4.390e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  0.020  4.486e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  0.070  3.496e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -0.085  5.591e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -0.004  4.629e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -0.076  6.431e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -0.046  1.422e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -0.057  3.398e-03   \n",
       "C(gender, levels=GENDERS)[T.F]                      0.026  2.934e-03   \n",
       "C(gender, levels=GENDERS)[T.M]                     -0.002  2.597e-03   \n",
       "\n",
       "                                                               Filtered 1  \\\n",
       "                                                         P>|z|      Coef.   \n",
       "Intercept                                            0.000e+00  8.550e-01   \n",
       "I(year_span < 0)[T.True]                            4.270e-152 -3.821e-01   \n",
       "I(year_span == 0)[T.True]                            0.000e+00  7.146e-01   \n",
       "mf.score_log_1(year_span)                            0.000e+00  8.259e-01   \n",
       "I(mf.score_log_1(year_span) ** 2)                    0.000e+00 -1.160e+00   \n",
       "I(auth_prev_papers == 0)[T.True]                     0.000e+00 -1.259e+00   \n",
       "I(auth_prev_papers == 1)[T.True]                     3.454e-70 -2.089e-01   \n",
       "np.log10(auth_prev_papers + 1)                       0.000e+00  2.458e+00   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)               0.000e+00 -4.035e-01   \n",
       "source_is_journal[T.True]                           1.707e-111        NaN   \n",
       "source_is_review[T.True]                             0.000e+00        NaN   \n",
       "source_is_case_rep[T.True]                           0.000e+00        NaN   \n",
       "source_is_let_ed_com[T.True]                        9.331e-261        NaN   \n",
       "sink_is_journal[T.True]                             1.501e-257  5.370e-01   \n",
       "sink_is_review[T.True]                               0.000e+00 -6.616e-01   \n",
       "sink_is_case_rep[T.True]                             0.000e+00 -1.038e+00   \n",
       "sink_is_let_ed_com[T.True]                          3.266e-213 -3.867e-01   \n",
       "journal_same[T.True]                                 0.000e+00  6.124e-01   \n",
       "I(jj_sim == 0)[T.True]                               0.000e+00 -1.677e-01   \n",
       "np.log10(jj_sim + 1)                                 4.277e-55  3.594e-01   \n",
       "I(np.log10(jj_sim + 1) ** 2)                         1.148e-91 -1.456e-01   \n",
       "I(sink_prev_ncites == 0)[T.True]                     0.000e+00  2.968e-01   \n",
       "np.log10(sink_prev_ncites + 1)                       4.191e-71  1.991e-01   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)               0.000e+00 -2.088e-01   \n",
       "I(source_ncites == 1)[T.True]                        2.477e-84        NaN   \n",
       "np.log10(source_ncites)                              0.000e+00 -1.776e+01   \n",
       "I(np.log10(source_ncites) ** 2)                      0.000e+00  1.354e+01   \n",
       "I(np.log10(source_ncites) ** 3)                      0.000e+00 -3.438e+00   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...  3.526e-203        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]     0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...   9.138e-53        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...   0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...  1.689e-203        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...   0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...  3.086e-113        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...   0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...  4.252e-132        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...  6.632e-310        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...  5.058e-260        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...   6.425e-63        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...  1.078e-143        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...   0.000e+00        NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)        0.000e+00 -1.345e-01   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          5.457e-20  1.570e-01   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...  1.213e-127 -4.152e-02   \n",
       "I(source_n_authors > 20)[T.True]                     7.036e-05  2.323e-02   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))          1.041e-230 -1.366e-01   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   3.481e-11 -1.128e-01   \n",
       "source_is_eng[T.True]                                0.000e+00        NaN   \n",
       "sink_is_eng[T.True]                                  5.789e-04  2.656e+00   \n",
       "I(source_n_mesh_ex == 0)[T.True]                     0.000e+00 -7.059e-01   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                       6.763e-01 -9.296e-02   \n",
       "np.log10(source_n_mesh_ex + 1)                       0.000e+00 -1.537e-01   \n",
       "np.log10(sink_n_mesh_ex + 1)                        1.767e-108 -1.789e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  6.084e-130  5.838e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.645e-134 -2.033e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   4.623e-21  1.815e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   6.342e-16  1.517e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   4.544e-15  1.617e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   6.573e-33 -2.034e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   5.446e-09  5.540e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   6.955e-06  7.448e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   2.459e-89  9.378e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   8.268e-52 -3.558e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   4.259e-01 -2.958e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.624e-32  9.743e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.126e-03 -1.005e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   5.873e-63  6.383e-02   \n",
       "C(gender, levels=GENDERS)[T.F]                       7.435e-19  3.812e-04   \n",
       "C(gender, levels=GENDERS)[T.M]                       3.487e-01 -1.142e-02   \n",
       "\n",
       "                                                                         \\\n",
       "                                                   Std.Err.       P>|z|   \n",
       "Intercept                                             0.488   7.964e-02   \n",
       "I(year_span < 0)[T.True]                              0.032   9.889e-33   \n",
       "I(year_span == 0)[T.True]                             0.009   0.000e+00   \n",
       "mf.score_log_1(year_span)                             0.022   0.000e+00   \n",
       "I(mf.score_log_1(year_span) ** 2)                     0.013   0.000e+00   \n",
       "I(auth_prev_papers == 0)[T.True]                      0.039  7.020e-231   \n",
       "I(auth_prev_papers == 1)[T.True]                      0.024   5.352e-18   \n",
       "np.log10(auth_prev_papers + 1)                        0.016   0.000e+00   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                0.004   0.000e+00   \n",
       "source_is_journal[T.True]                               NaN         NaN   \n",
       "source_is_review[T.True]                                NaN         NaN   \n",
       "source_is_case_rep[T.True]                              NaN         NaN   \n",
       "source_is_let_ed_com[T.True]                            NaN         NaN   \n",
       "sink_is_journal[T.True]                               0.020  3.782e-163   \n",
       "sink_is_review[T.True]                                0.004   0.000e+00   \n",
       "sink_is_case_rep[T.True]                              0.011   0.000e+00   \n",
       "sink_is_let_ed_com[T.True]                            0.019   9.806e-90   \n",
       "journal_same[T.True]                                  0.004   0.000e+00   \n",
       "I(jj_sim == 0)[T.True]                                0.013   2.733e-40   \n",
       "np.log10(jj_sim + 1)                                  0.016  1.723e-107   \n",
       "I(np.log10(jj_sim + 1) ** 2)                          0.005  3.399e-185   \n",
       "I(sink_prev_ncites == 0)[T.True]                      0.006   0.000e+00   \n",
       "np.log10(sink_prev_ncites + 1)                        0.009  6.238e-119   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                0.003   0.000e+00   \n",
       "I(source_ncites == 1)[T.True]                           NaN         NaN   \n",
       "np.log10(source_ncites)                               1.055   1.635e-63   \n",
       "I(np.log10(source_ncites) ** 2)                       0.772   6.040e-69   \n",
       "I(np.log10(source_ncites) ** 3)                       0.186   4.052e-76   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]        NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...      NaN         NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)         0.002   0.000e+00   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)           0.010   1.388e-52   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...    0.002  3.386e-118   \n",
       "I(source_n_authors > 20)[T.True]                      0.024   3.258e-01   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))            0.026   1.286e-07   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)    0.019   4.115e-09   \n",
       "source_is_eng[T.True]                                   NaN         NaN   \n",
       "sink_is_eng[T.True]                                   0.102  4.895e-149   \n",
       "I(source_n_mesh_ex == 0)[T.True]                      0.020  6.641e-269   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                        0.022   2.792e-05   \n",
       "np.log10(source_n_mesh_ex + 1)                        0.008   7.027e-84   \n",
       "np.log10(sink_n_mesh_ex + 1)                          0.008  1.370e-112   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.005   5.079e-36   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   2.542e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.006  3.474e-188   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.011   1.340e-43   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   1.828e-83   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   2.611e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   3.340e-11   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   2.266e-30   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.009   1.986e-24   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.009   1.047e-04   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.011   6.362e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.016   2.776e-09   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.023   7.966e-06   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   9.465e-22   \n",
       "C(gender, levels=GENDERS)[T.F]                        0.006   9.488e-01   \n",
       "C(gender, levels=GENDERS)[T.M]                        0.005   3.326e-02   \n",
       "\n",
       "                                                   Filtered 2           \\\n",
       "                                                        Coef. Std.Err.   \n",
       "Intercept                                              -0.491    0.695   \n",
       "I(year_span < 0)[T.True]                               -0.394    0.044   \n",
       "I(year_span == 0)[T.True]                               0.748    0.013   \n",
       "mf.score_log_1(year_span)                               0.856    0.029   \n",
       "I(mf.score_log_1(year_span) ** 2)                      -1.155    0.017   \n",
       "I(auth_prev_papers == 0)[T.True]                       -1.236    0.054   \n",
       "I(auth_prev_papers == 1)[T.True]                       -0.189    0.035   \n",
       "np.log10(auth_prev_papers + 1)                          2.468    0.022   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                 -0.401    0.006   \n",
       "source_is_journal[T.True]                                 NaN      NaN   \n",
       "source_is_review[T.True]                                  NaN      NaN   \n",
       "source_is_case_rep[T.True]                                NaN      NaN   \n",
       "source_is_let_ed_com[T.True]                              NaN      NaN   \n",
       "sink_is_journal[T.True]                                 0.575    0.027   \n",
       "sink_is_review[T.True]                                 -0.641    0.006   \n",
       "sink_is_case_rep[T.True]                               -1.046    0.014   \n",
       "sink_is_let_ed_com[T.True]                             -0.373    0.026   \n",
       "journal_same[T.True]                                    0.616    0.005   \n",
       "I(jj_sim == 0)[T.True]                                 -0.188    0.017   \n",
       "np.log10(jj_sim + 1)                                    0.358    0.022   \n",
       "I(np.log10(jj_sim + 1) ** 2)                           -0.143    0.007   \n",
       "I(sink_prev_ncites == 0)[T.True]                        0.292    0.009   \n",
       "np.log10(sink_prev_ncites + 1)                          0.215    0.012   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                 -0.201    0.004   \n",
       "I(source_ncites == 1)[T.True]                             NaN      NaN   \n",
       "np.log10(source_ncites)                               -17.484    1.420   \n",
       "I(np.log10(source_ncites) ** 2)                        13.420    1.039   \n",
       "I(np.log10(source_ncites) ** 3)                        -3.422    0.251   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]          NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...        NaN      NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)          -0.157    0.003   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)             0.199    0.014   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...     -0.049    0.002   \n",
       "I(source_n_authors > 20)[T.True]                       -0.037    0.032   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))             -0.183    0.035   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)     -0.087    0.026   \n",
       "source_is_eng[T.True]                                     NaN      NaN   \n",
       "sink_is_eng[T.True]                                     3.959    0.268   \n",
       "I(source_n_mesh_ex == 0)[T.True]                       -0.908    0.028   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                         -0.106    0.030   \n",
       "np.log10(source_n_mesh_ex + 1)                         -0.239    0.011   \n",
       "np.log10(sink_n_mesh_ex + 1)                           -0.230    0.011   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "C(gender, levels=GENDERS)[T.F]                          0.034    0.011   \n",
       "C(gender, levels=GENDERS)[T.M]                          0.005    0.010   \n",
       "\n",
       "                                                                \n",
       "                                                         P>|z|  \n",
       "Intercept                                            4.799e-01  \n",
       "I(year_span < 0)[T.True]                             2.207e-19  \n",
       "I(year_span == 0)[T.True]                            0.000e+00  \n",
       "mf.score_log_1(year_span)                           2.733e-190  \n",
       "I(mf.score_log_1(year_span) ** 2)                    0.000e+00  \n",
       "I(auth_prev_papers == 0)[T.True]                    3.506e-114  \n",
       "I(auth_prev_papers == 1)[T.True]                     6.216e-08  \n",
       "np.log10(auth_prev_papers + 1)                       0.000e+00  \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)               0.000e+00  \n",
       "source_is_journal[T.True]                                  NaN  \n",
       "source_is_review[T.True]                                   NaN  \n",
       "source_is_case_rep[T.True]                                 NaN  \n",
       "source_is_let_ed_com[T.True]                               NaN  \n",
       "sink_is_journal[T.True]                             8.207e-101  \n",
       "sink_is_review[T.True]                               0.000e+00  \n",
       "sink_is_case_rep[T.True]                             0.000e+00  \n",
       "sink_is_let_ed_com[T.True]                           5.465e-46  \n",
       "journal_same[T.True]                                 0.000e+00  \n",
       "I(jj_sim == 0)[T.True]                               8.701e-29  \n",
       "np.log10(jj_sim + 1)                                 2.322e-61  \n",
       "I(np.log10(jj_sim + 1) ** 2)                        5.665e-103  \n",
       "I(sink_prev_ncites == 0)[T.True]                    2.798e-249  \n",
       "np.log10(sink_prev_ncites + 1)                       2.376e-77  \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)               0.000e+00  \n",
       "I(source_ncites == 1)[T.True]                              NaN  \n",
       "np.log10(source_ncites)                              7.628e-35  \n",
       "I(np.log10(source_ncites) ** 2)                      3.616e-38  \n",
       "I(np.log10(source_ncites) ** 3)                      2.177e-42  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]           NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...         NaN  \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)        0.000e+00  \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          3.391e-45  \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...   9.311e-89  \n",
       "I(source_n_authors > 20)[T.True]                     2.452e-01  \n",
       "np.log10(np.clip(source_n_authors, 0, 20))           1.756e-07  \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   8.039e-04  \n",
       "source_is_eng[T.True]                                      NaN  \n",
       "sink_is_eng[T.True]                                  1.711e-49  \n",
       "I(source_n_mesh_ex == 0)[T.True]                    3.226e-237  \n",
       "I(sink_n_mesh_ex == 0)[T.True]                       4.778e-04  \n",
       "np.log10(source_n_mesh_ex + 1)                      2.737e-109  \n",
       "np.log10(sink_n_mesh_ex + 1)                        1.515e-101  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  \n",
       "C(gender, levels=GENDERS)[T.F]                       1.590e-03  \n",
       "C(gender, levels=GENDERS)[T.M]                       6.570e-01  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_t_la = merge_results([results_full_la, results_flt_1_la, results_flt_2_la])\n",
    "with pd.option_context(\"display.max_columns\", 20, \"display.max_rows\", 200, \"precision\", 3):\n",
    "    display(df_t_la)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Both combined"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((70, 9), (70, 9))"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_t.shape, df_t_la.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "set(df_t.index.values) == set(df_t_la.index.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"9\" halign=\"left\">First</th>\n",
       "      <th colspan=\"9\" halign=\"left\">Last</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"3\" halign=\"left\">Full</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 1</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 2</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Full</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 1</th>\n",
       "      <th colspan=\"3\" halign=\"left\">Filtered 2</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Intercept</th>\n",
       "      <td>-1.957e+00</td>\n",
       "      <td>0.031</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-4.047</td>\n",
       "      <td>0.289</td>\n",
       "      <td>2.037e-44</td>\n",
       "      <td>-3.542</td>\n",
       "      <td>0.482</td>\n",
       "      <td>1.978e-13</td>\n",
       "      <td>-2.790</td>\n",
       "      <td>2.674e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>8.550e-01</td>\n",
       "      <td>0.488</td>\n",
       "      <td>7.964e-02</td>\n",
       "      <td>-0.491</td>\n",
       "      <td>0.695</td>\n",
       "      <td>4.799e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(sink_prev_ncites == 0)[T.True]</th>\n",
       "      <td>1.265e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.064e-235</td>\n",
       "      <td>0.166</td>\n",
       "      <td>0.007</td>\n",
       "      <td>4.064e-120</td>\n",
       "      <td>0.156</td>\n",
       "      <td>0.011</td>\n",
       "      <td>2.378e-47</td>\n",
       "      <td>0.240</td>\n",
       "      <td>3.101e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>2.968e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.292</td>\n",
       "      <td>0.009</td>\n",
       "      <td>2.798e-249</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(sink_prev_ncites + 1)</th>\n",
       "      <td>-4.433e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.147</td>\n",
       "      <td>0.011</td>\n",
       "      <td>7.272e-39</td>\n",
       "      <td>-0.137</td>\n",
       "      <td>0.016</td>\n",
       "      <td>5.717e-17</td>\n",
       "      <td>-0.079</td>\n",
       "      <td>4.424e-03</td>\n",
       "      <td>4.191e-71</td>\n",
       "      <td>1.991e-01</td>\n",
       "      <td>0.009</td>\n",
       "      <td>6.238e-119</td>\n",
       "      <td>0.215</td>\n",
       "      <td>0.012</td>\n",
       "      <td>2.376e-77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(sink_prev_ncites + 1) ** 2)</th>\n",
       "      <td>-1.698e-01</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.226</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.181</td>\n",
       "      <td>0.006</td>\n",
       "      <td>2.708e-203</td>\n",
       "      <td>-0.169</td>\n",
       "      <td>1.530e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-2.088e-01</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.201</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(auth_prev_papers == 0)[T.True]</th>\n",
       "      <td>-1.421e+00</td>\n",
       "      <td>0.009</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.232</td>\n",
       "      <td>0.018</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.236</td>\n",
       "      <td>0.030</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.232</td>\n",
       "      <td>1.639e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.259e+00</td>\n",
       "      <td>0.039</td>\n",
       "      <td>7.020e-231</td>\n",
       "      <td>-1.236</td>\n",
       "      <td>0.054</td>\n",
       "      <td>3.506e-114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(auth_prev_papers == 1)[T.True]</th>\n",
       "      <td>-2.357e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.159</td>\n",
       "      <td>0.010</td>\n",
       "      <td>2.285e-58</td>\n",
       "      <td>-0.158</td>\n",
       "      <td>0.017</td>\n",
       "      <td>1.823e-21</td>\n",
       "      <td>-0.187</td>\n",
       "      <td>1.058e-02</td>\n",
       "      <td>3.454e-70</td>\n",
       "      <td>-2.089e-01</td>\n",
       "      <td>0.024</td>\n",
       "      <td>5.352e-18</td>\n",
       "      <td>-0.189</td>\n",
       "      <td>0.035</td>\n",
       "      <td>6.216e-08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(auth_prev_papers + 1)</th>\n",
       "      <td>1.610e+00</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>1.862</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>1.911</td>\n",
       "      <td>0.020</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>2.172</td>\n",
       "      <td>7.733e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>2.458e+00</td>\n",
       "      <td>0.016</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>2.468</td>\n",
       "      <td>0.022</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(auth_prev_papers + 1) ** 2)</th>\n",
       "      <td>-9.062e-02</td>\n",
       "      <td>0.002</td>\n",
       "      <td>1.091e-294</td>\n",
       "      <td>-0.127</td>\n",
       "      <td>0.004</td>\n",
       "      <td>9.929e-187</td>\n",
       "      <td>-0.156</td>\n",
       "      <td>0.006</td>\n",
       "      <td>4.687e-130</td>\n",
       "      <td>-0.340</td>\n",
       "      <td>2.154e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-4.035e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.401</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_journal[T.True]</th>\n",
       "      <td>2.661e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>6.473e-129</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.229</td>\n",
       "      <td>1.021e-02</td>\n",
       "      <td>1.707e-111</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_review[T.True]</th>\n",
       "      <td>-7.869e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>5.938e-199</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.249</td>\n",
       "      <td>2.125e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_case_rep[T.True]</th>\n",
       "      <td>-9.404e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.948</td>\n",
       "      <td>5.222e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_let_ed_com[T.True]</th>\n",
       "      <td>-4.295e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.343</td>\n",
       "      <td>9.952e-03</td>\n",
       "      <td>9.331e-261</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_journal[T.True]</th>\n",
       "      <td>3.283e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.814e-181</td>\n",
       "      <td>0.548</td>\n",
       "      <td>0.022</td>\n",
       "      <td>5.741e-140</td>\n",
       "      <td>0.496</td>\n",
       "      <td>0.032</td>\n",
       "      <td>3.460e-55</td>\n",
       "      <td>0.324</td>\n",
       "      <td>9.445e-03</td>\n",
       "      <td>1.501e-257</td>\n",
       "      <td>5.370e-01</td>\n",
       "      <td>0.020</td>\n",
       "      <td>3.782e-163</td>\n",
       "      <td>0.575</td>\n",
       "      <td>0.027</td>\n",
       "      <td>8.207e-101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_review[T.True]</th>\n",
       "      <td>-7.826e-01</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.741</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.639</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.641</td>\n",
       "      <td>2.142e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-6.616e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.641</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_case_rep[T.True]</th>\n",
       "      <td>-6.369e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.882</td>\n",
       "      <td>0.012</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.861</td>\n",
       "      <td>0.017</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.731</td>\n",
       "      <td>4.949e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.038e+00</td>\n",
       "      <td>0.011</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.046</td>\n",
       "      <td>0.014</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_let_ed_com[T.True]</th>\n",
       "      <td>-2.838e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>9.580e-138</td>\n",
       "      <td>-0.421</td>\n",
       "      <td>0.021</td>\n",
       "      <td>1.665e-86</td>\n",
       "      <td>-0.451</td>\n",
       "      <td>0.031</td>\n",
       "      <td>7.459e-47</td>\n",
       "      <td>-0.292</td>\n",
       "      <td>9.374e-03</td>\n",
       "      <td>3.266e-213</td>\n",
       "      <td>-3.867e-01</td>\n",
       "      <td>0.019</td>\n",
       "      <td>9.806e-90</td>\n",
       "      <td>-0.373</td>\n",
       "      <td>0.026</td>\n",
       "      <td>5.465e-46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(year_span &lt; 0)[T.True]</th>\n",
       "      <td>-5.564e-01</td>\n",
       "      <td>0.016</td>\n",
       "      <td>2.011e-275</td>\n",
       "      <td>-0.445</td>\n",
       "      <td>0.028</td>\n",
       "      <td>1.636e-55</td>\n",
       "      <td>-0.327</td>\n",
       "      <td>0.041</td>\n",
       "      <td>2.789e-15</td>\n",
       "      <td>-0.383</td>\n",
       "      <td>1.459e-02</td>\n",
       "      <td>4.270e-152</td>\n",
       "      <td>-3.821e-01</td>\n",
       "      <td>0.032</td>\n",
       "      <td>9.889e-33</td>\n",
       "      <td>-0.394</td>\n",
       "      <td>0.044</td>\n",
       "      <td>2.207e-19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(year_span == 0)[T.True]</th>\n",
       "      <td>5.382e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.446</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.580</td>\n",
       "      <td>0.015</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.735</td>\n",
       "      <td>4.656e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>7.146e-01</td>\n",
       "      <td>0.009</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.748</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.score_log_1(year_span)</th>\n",
       "      <td>5.387e-01</td>\n",
       "      <td>0.015</td>\n",
       "      <td>2.008e-275</td>\n",
       "      <td>-0.204</td>\n",
       "      <td>0.026</td>\n",
       "      <td>6.462e-15</td>\n",
       "      <td>0.152</td>\n",
       "      <td>0.038</td>\n",
       "      <td>7.348e-05</td>\n",
       "      <td>1.232</td>\n",
       "      <td>1.131e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>8.259e-01</td>\n",
       "      <td>0.022</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.856</td>\n",
       "      <td>0.029</td>\n",
       "      <td>2.733e-190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(mf.score_log_1(year_span) ** 2)</th>\n",
       "      <td>-1.281e+00</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.767</td>\n",
       "      <td>0.016</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.857</td>\n",
       "      <td>0.023</td>\n",
       "      <td>1.099e-302</td>\n",
       "      <td>-1.402</td>\n",
       "      <td>6.820e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.160e+00</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.155</td>\n",
       "      <td>0.017</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>journal_same[T.True]</th>\n",
       "      <td>4.567e-01</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.509</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.501</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.534</td>\n",
       "      <td>2.218e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>6.124e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>0.616</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(jj_sim == 0)[T.True]</th>\n",
       "      <td>-2.805e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.735e-259</td>\n",
       "      <td>-0.172</td>\n",
       "      <td>0.015</td>\n",
       "      <td>4.417e-31</td>\n",
       "      <td>-0.163</td>\n",
       "      <td>0.022</td>\n",
       "      <td>1.448e-13</td>\n",
       "      <td>-0.296</td>\n",
       "      <td>6.332e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.677e-01</td>\n",
       "      <td>0.013</td>\n",
       "      <td>2.733e-40</td>\n",
       "      <td>-0.188</td>\n",
       "      <td>0.017</td>\n",
       "      <td>8.701e-29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(jj_sim + 1)</th>\n",
       "      <td>6.298e-02</td>\n",
       "      <td>0.010</td>\n",
       "      <td>3.421e-10</td>\n",
       "      <td>0.205</td>\n",
       "      <td>0.019</td>\n",
       "      <td>3.512e-28</td>\n",
       "      <td>0.255</td>\n",
       "      <td>0.027</td>\n",
       "      <td>1.016e-20</td>\n",
       "      <td>0.126</td>\n",
       "      <td>8.068e-03</td>\n",
       "      <td>4.277e-55</td>\n",
       "      <td>3.594e-01</td>\n",
       "      <td>0.016</td>\n",
       "      <td>1.723e-107</td>\n",
       "      <td>0.358</td>\n",
       "      <td>0.022</td>\n",
       "      <td>2.322e-61</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(jj_sim + 1) ** 2)</th>\n",
       "      <td>5.949e-03</td>\n",
       "      <td>0.003</td>\n",
       "      <td>4.459e-02</td>\n",
       "      <td>-0.044</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.126e-15</td>\n",
       "      <td>-0.056</td>\n",
       "      <td>0.008</td>\n",
       "      <td>8.183e-12</td>\n",
       "      <td>-0.050</td>\n",
       "      <td>2.456e-03</td>\n",
       "      <td>1.148e-91</td>\n",
       "      <td>-1.456e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>3.399e-185</td>\n",
       "      <td>-0.143</td>\n",
       "      <td>0.007</td>\n",
       "      <td>5.665e-103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_ncites == 1)[T.True]</th>\n",
       "      <td>-2.997e-01</td>\n",
       "      <td>0.022</td>\n",
       "      <td>1.212e-41</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.405</td>\n",
       "      <td>2.079e-02</td>\n",
       "      <td>2.477e-84</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(source_ncites)</th>\n",
       "      <td>-1.647e+00</td>\n",
       "      <td>0.034</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-3.223</td>\n",
       "      <td>0.562</td>\n",
       "      <td>1.002e-08</td>\n",
       "      <td>-7.395</td>\n",
       "      <td>0.816</td>\n",
       "      <td>1.230e-19</td>\n",
       "      <td>-2.221</td>\n",
       "      <td>3.101e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.776e+01</td>\n",
       "      <td>1.055</td>\n",
       "      <td>1.635e-63</td>\n",
       "      <td>-17.484</td>\n",
       "      <td>1.420</td>\n",
       "      <td>7.628e-35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(source_ncites) ** 2)</th>\n",
       "      <td>8.198e-01</td>\n",
       "      <td>0.023</td>\n",
       "      <td>4.438e-269</td>\n",
       "      <td>2.579</td>\n",
       "      <td>0.379</td>\n",
       "      <td>9.953e-12</td>\n",
       "      <td>5.400</td>\n",
       "      <td>0.550</td>\n",
       "      <td>1.023e-22</td>\n",
       "      <td>1.463</td>\n",
       "      <td>2.148e-02</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>1.354e+01</td>\n",
       "      <td>0.772</td>\n",
       "      <td>6.040e-69</td>\n",
       "      <td>13.420</td>\n",
       "      <td>1.039</td>\n",
       "      <td>3.616e-38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(source_ncites) ** 3)</th>\n",
       "      <td>-2.294e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.771</td>\n",
       "      <td>0.084</td>\n",
       "      <td>3.470e-20</td>\n",
       "      <td>-1.386</td>\n",
       "      <td>0.122</td>\n",
       "      <td>5.992e-30</td>\n",
       "      <td>-0.387</td>\n",
       "      <td>4.793e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-3.438e+00</td>\n",
       "      <td>0.186</td>\n",
       "      <td>4.052e-76</td>\n",
       "      <td>-3.422</td>\n",
       "      <td>0.251</td>\n",
       "      <td>2.177e-42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.UNKNOWN]</th>\n",
       "      <td>-4.468e-02</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.435e-12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.158</td>\n",
       "      <td>5.211e-03</td>\n",
       "      <td>3.526e-203</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.UK]</th>\n",
       "      <td>-5.242e-03</td>\n",
       "      <td>0.003</td>\n",
       "      <td>1.058e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.090</td>\n",
       "      <td>2.373e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.JAPAN]</th>\n",
       "      <td>2.669e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.092</td>\n",
       "      <td>6.048e-03</td>\n",
       "      <td>9.138e-53</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.GERMANY]</th>\n",
       "      <td>2.382e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>9.888e-09</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.169</td>\n",
       "      <td>3.061e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.FRANCE]</th>\n",
       "      <td>-4.425e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>3.363e-17</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.116</td>\n",
       "      <td>3.820e-03</td>\n",
       "      <td>1.689e-203</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.ITALY]</th>\n",
       "      <td>-1.010e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>3.885e-54</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.298</td>\n",
       "      <td>4.877e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.CANADA]</th>\n",
       "      <td>-3.874e-03</td>\n",
       "      <td>0.004</td>\n",
       "      <td>3.608e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.069</td>\n",
       "      <td>3.035e-03</td>\n",
       "      <td>3.086e-113</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.CHINA]</th>\n",
       "      <td>-4.120e-01</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.432</td>\n",
       "      <td>7.300e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.AUSTRALIA]</th>\n",
       "      <td>8.788e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.664e-63</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.099</td>\n",
       "      <td>4.043e-03</td>\n",
       "      <td>4.252e-132</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.SPAIN]</th>\n",
       "      <td>-1.223e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.850e-83</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.184</td>\n",
       "      <td>4.889e-03</td>\n",
       "      <td>6.632e-310</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.NETHERLANDS]</th>\n",
       "      <td>6.808e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>1.152e-21</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.179</td>\n",
       "      <td>5.204e-03</td>\n",
       "      <td>5.058e-260</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.SWEDEN]</th>\n",
       "      <td>1.074e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>4.490e-58</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.083</td>\n",
       "      <td>4.978e-03</td>\n",
       "      <td>6.425e-63</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.INDIA]</th>\n",
       "      <td>-4.363e-02</td>\n",
       "      <td>0.010</td>\n",
       "      <td>3.026e-05</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.206</td>\n",
       "      <td>8.064e-03</td>\n",
       "      <td>1.078e-143</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(source_country, levels=TOP_15_COUNTRIES)[T.OTHER]</th>\n",
       "      <td>-6.440e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>8.308e-128</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.213</td>\n",
       "      <td>2.054e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.nan_to_num(source_V_novelty) + 1)</th>\n",
       "      <td>-1.058e-01</td>\n",
       "      <td>0.001</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.096</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.112</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.108e-214</td>\n",
       "      <td>-0.139</td>\n",
       "      <td>9.809e-04</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.345e-01</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.157</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.nan_to_num(sink_V_novelty) + 1)</th>\n",
       "      <td>9.484e-02</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.753e-35</td>\n",
       "      <td>0.200</td>\n",
       "      <td>0.014</td>\n",
       "      <td>2.878e-49</td>\n",
       "      <td>0.158</td>\n",
       "      <td>0.021</td>\n",
       "      <td>2.289e-14</td>\n",
       "      <td>0.048</td>\n",
       "      <td>5.268e-03</td>\n",
       "      <td>5.457e-20</td>\n",
       "      <td>1.570e-01</td>\n",
       "      <td>0.010</td>\n",
       "      <td>1.388e-52</td>\n",
       "      <td>0.199</td>\n",
       "      <td>0.014</td>\n",
       "      <td>3.391e-45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(np.nan_to_num(sink_V_novelty) + 1) ** 2)</th>\n",
       "      <td>-2.774e-02</td>\n",
       "      <td>0.001</td>\n",
       "      <td>1.412e-100</td>\n",
       "      <td>-0.044</td>\n",
       "      <td>0.002</td>\n",
       "      <td>7.102e-82</td>\n",
       "      <td>-0.037</td>\n",
       "      <td>0.004</td>\n",
       "      <td>6.418e-26</td>\n",
       "      <td>-0.022</td>\n",
       "      <td>9.206e-04</td>\n",
       "      <td>1.213e-127</td>\n",
       "      <td>-4.152e-02</td>\n",
       "      <td>0.002</td>\n",
       "      <td>3.386e-118</td>\n",
       "      <td>-0.049</td>\n",
       "      <td>0.002</td>\n",
       "      <td>9.311e-89</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>source_is_eng[T.True]</th>\n",
       "      <td>6.494e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.701</td>\n",
       "      <td>6.660e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sink_is_eng[T.True]</th>\n",
       "      <td>-1.846e-01</td>\n",
       "      <td>0.010</td>\n",
       "      <td>6.033e-81</td>\n",
       "      <td>2.075</td>\n",
       "      <td>0.084</td>\n",
       "      <td>2.546e-133</td>\n",
       "      <td>3.625</td>\n",
       "      <td>0.268</td>\n",
       "      <td>1.149e-41</td>\n",
       "      <td>-0.031</td>\n",
       "      <td>8.947e-03</td>\n",
       "      <td>5.789e-04</td>\n",
       "      <td>2.656e+00</td>\n",
       "      <td>0.102</td>\n",
       "      <td>4.895e-149</td>\n",
       "      <td>3.959</td>\n",
       "      <td>0.268</td>\n",
       "      <td>1.711e-49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_n_authors &gt; 20)[T.True]</th>\n",
       "      <td>1.722e-04</td>\n",
       "      <td>0.015</td>\n",
       "      <td>9.906e-01</td>\n",
       "      <td>-0.037</td>\n",
       "      <td>0.025</td>\n",
       "      <td>1.343e-01</td>\n",
       "      <td>-0.047</td>\n",
       "      <td>0.034</td>\n",
       "      <td>1.627e-01</td>\n",
       "      <td>-0.050</td>\n",
       "      <td>1.263e-02</td>\n",
       "      <td>7.036e-05</td>\n",
       "      <td>2.323e-02</td>\n",
       "      <td>0.024</td>\n",
       "      <td>3.258e-01</td>\n",
       "      <td>-0.037</td>\n",
       "      <td>0.032</td>\n",
       "      <td>2.452e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(np.clip(source_n_authors, 0, 20))</th>\n",
       "      <td>8.239e-02</td>\n",
       "      <td>0.018</td>\n",
       "      <td>3.324e-06</td>\n",
       "      <td>0.242</td>\n",
       "      <td>0.030</td>\n",
       "      <td>1.357e-15</td>\n",
       "      <td>0.303</td>\n",
       "      <td>0.045</td>\n",
       "      <td>1.063e-11</td>\n",
       "      <td>-0.444</td>\n",
       "      <td>1.368e-02</td>\n",
       "      <td>1.041e-230</td>\n",
       "      <td>-1.366e-01</td>\n",
       "      <td>0.026</td>\n",
       "      <td>1.286e-07</td>\n",
       "      <td>-0.183</td>\n",
       "      <td>0.035</td>\n",
       "      <td>1.756e-07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)</th>\n",
       "      <td>-2.563e-01</td>\n",
       "      <td>0.013</td>\n",
       "      <td>1.194e-85</td>\n",
       "      <td>-0.336</td>\n",
       "      <td>0.023</td>\n",
       "      <td>3.234e-50</td>\n",
       "      <td>-0.353</td>\n",
       "      <td>0.033</td>\n",
       "      <td>2.710e-26</td>\n",
       "      <td>0.066</td>\n",
       "      <td>1.004e-02</td>\n",
       "      <td>3.481e-11</td>\n",
       "      <td>-1.128e-01</td>\n",
       "      <td>0.019</td>\n",
       "      <td>4.115e-09</td>\n",
       "      <td>-0.087</td>\n",
       "      <td>0.026</td>\n",
       "      <td>8.039e-04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[0]</th>\n",
       "      <td>-1.426e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.087e-04</td>\n",
       "      <td>-0.033</td>\n",
       "      <td>0.006</td>\n",
       "      <td>2.974e-08</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.062</td>\n",
       "      <td>2.555e-03</td>\n",
       "      <td>6.084e-130</td>\n",
       "      <td>5.838e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>5.079e-36</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[1]</th>\n",
       "      <td>5.313e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>2.200e-36</td>\n",
       "      <td>-0.038</td>\n",
       "      <td>0.007</td>\n",
       "      <td>1.021e-07</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.082</td>\n",
       "      <td>3.304e-03</td>\n",
       "      <td>1.645e-134</td>\n",
       "      <td>-2.033e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.542e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[2]</th>\n",
       "      <td>4.562e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>5.136e-29</td>\n",
       "      <td>0.180</td>\n",
       "      <td>0.006</td>\n",
       "      <td>9.166e-200</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.034</td>\n",
       "      <td>3.646e-03</td>\n",
       "      <td>4.623e-21</td>\n",
       "      <td>1.815e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.474e-188</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[3]</th>\n",
       "      <td>-1.021e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.235e-60</td>\n",
       "      <td>-0.147</td>\n",
       "      <td>0.009</td>\n",
       "      <td>7.110e-61</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.048</td>\n",
       "      <td>5.959e-03</td>\n",
       "      <td>6.342e-16</td>\n",
       "      <td>1.517e-01</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.340e-43</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[4]</th>\n",
       "      <td>4.401e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.985e-22</td>\n",
       "      <td>0.116</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.022e-50</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.032</td>\n",
       "      <td>4.138e-03</td>\n",
       "      <td>4.544e-15</td>\n",
       "      <td>1.617e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.828e-83</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[5]</th>\n",
       "      <td>-7.411e-02</td>\n",
       "      <td>0.004</td>\n",
       "      <td>1.704e-63</td>\n",
       "      <td>-0.066</td>\n",
       "      <td>0.008</td>\n",
       "      <td>4.727e-16</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.038</td>\n",
       "      <td>3.142e-03</td>\n",
       "      <td>6.573e-33</td>\n",
       "      <td>-2.034e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.611e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[6]</th>\n",
       "      <td>-5.946e-02</td>\n",
       "      <td>0.006</td>\n",
       "      <td>7.073e-23</td>\n",
       "      <td>-0.092</td>\n",
       "      <td>0.010</td>\n",
       "      <td>7.999e-20</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.026</td>\n",
       "      <td>4.390e-03</td>\n",
       "      <td>5.446e-09</td>\n",
       "      <td>5.540e-02</td>\n",
       "      <td>0.008</td>\n",
       "      <td>3.340e-11</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[7]</th>\n",
       "      <td>-6.379e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>5.197e-34</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.007</td>\n",
       "      <td>4.308e-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.020</td>\n",
       "      <td>4.486e-03</td>\n",
       "      <td>6.955e-06</td>\n",
       "      <td>7.448e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.266e-30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[8]</th>\n",
       "      <td>1.355e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.099e-172</td>\n",
       "      <td>0.030</td>\n",
       "      <td>0.012</td>\n",
       "      <td>8.899e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.070</td>\n",
       "      <td>3.496e-03</td>\n",
       "      <td>2.459e-89</td>\n",
       "      <td>9.378e-02</td>\n",
       "      <td>0.009</td>\n",
       "      <td>1.986e-24</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[9]</th>\n",
       "      <td>-5.473e-02</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.299e-20</td>\n",
       "      <td>-0.028</td>\n",
       "      <td>0.009</td>\n",
       "      <td>3.004e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.085</td>\n",
       "      <td>5.591e-03</td>\n",
       "      <td>8.268e-52</td>\n",
       "      <td>-3.558e-02</td>\n",
       "      <td>0.009</td>\n",
       "      <td>1.047e-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[10]</th>\n",
       "      <td>3.040e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>3.702e-06</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.014</td>\n",
       "      <td>6.575e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.004</td>\n",
       "      <td>4.629e-03</td>\n",
       "      <td>4.259e-01</td>\n",
       "      <td>-2.958e-02</td>\n",
       "      <td>0.011</td>\n",
       "      <td>6.362e-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[11]</th>\n",
       "      <td>-1.869e-01</td>\n",
       "      <td>0.007</td>\n",
       "      <td>2.492e-146</td>\n",
       "      <td>-0.005</td>\n",
       "      <td>0.012</td>\n",
       "      <td>6.557e-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.076</td>\n",
       "      <td>6.431e-03</td>\n",
       "      <td>1.624e-32</td>\n",
       "      <td>9.743e-02</td>\n",
       "      <td>0.016</td>\n",
       "      <td>2.776e-09</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[12]</th>\n",
       "      <td>1.261e-02</td>\n",
       "      <td>0.018</td>\n",
       "      <td>4.875e-01</td>\n",
       "      <td>-0.062</td>\n",
       "      <td>0.026</td>\n",
       "      <td>1.588e-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.046</td>\n",
       "      <td>1.422e-02</td>\n",
       "      <td>1.126e-03</td>\n",
       "      <td>-1.005e-01</td>\n",
       "      <td>0.023</td>\n",
       "      <td>7.966e-06</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mf.MC(eth1, eth2, weights=eth_weight, levels=TOP_15_ETHNICITIES)[13]</th>\n",
       "      <td>-1.807e-01</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.110</td>\n",
       "      <td>0.008</td>\n",
       "      <td>4.316e-46</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.057</td>\n",
       "      <td>3.398e-03</td>\n",
       "      <td>5.873e-63</td>\n",
       "      <td>6.383e-02</td>\n",
       "      <td>0.007</td>\n",
       "      <td>9.465e-22</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(source_n_mesh_ex == 0)[T.True]</th>\n",
       "      <td>-6.666e-01</td>\n",
       "      <td>0.013</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-0.511</td>\n",
       "      <td>0.024</td>\n",
       "      <td>5.006e-100</td>\n",
       "      <td>-0.628</td>\n",
       "      <td>0.036</td>\n",
       "      <td>1.575e-66</td>\n",
       "      <td>-0.892</td>\n",
       "      <td>9.581e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-7.059e-01</td>\n",
       "      <td>0.020</td>\n",
       "      <td>6.641e-269</td>\n",
       "      <td>-0.908</td>\n",
       "      <td>0.028</td>\n",
       "      <td>3.226e-237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I(sink_n_mesh_ex == 0)[T.True]</th>\n",
       "      <td>-9.292e-02</td>\n",
       "      <td>0.015</td>\n",
       "      <td>1.355e-09</td>\n",
       "      <td>0.019</td>\n",
       "      <td>0.028</td>\n",
       "      <td>4.991e-01</td>\n",
       "      <td>-0.156</td>\n",
       "      <td>0.044</td>\n",
       "      <td>3.500e-04</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.108e-02</td>\n",
       "      <td>6.763e-01</td>\n",
       "      <td>-9.296e-02</td>\n",
       "      <td>0.022</td>\n",
       "      <td>2.792e-05</td>\n",
       "      <td>-0.106</td>\n",
       "      <td>0.030</td>\n",
       "      <td>4.778e-04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(source_n_mesh_ex + 1)</th>\n",
       "      <td>-1.704e-01</td>\n",
       "      <td>0.005</td>\n",
       "      <td>2.015e-212</td>\n",
       "      <td>-0.107</td>\n",
       "      <td>0.009</td>\n",
       "      <td>2.730e-30</td>\n",
       "      <td>-0.150</td>\n",
       "      <td>0.014</td>\n",
       "      <td>2.657e-26</td>\n",
       "      <td>-0.251</td>\n",
       "      <td>4.138e-03</td>\n",
       "      <td>0.000e+00</td>\n",
       "      <td>-1.537e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>7.027e-84</td>\n",
       "      <td>-0.239</td>\n",
       "      <td>0.011</td>\n",
       "      <td>2.737e-109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>np.log10(sink_n_mesh_ex + 1)</th>\n",
       "      <td>-1.681e-01</td>\n",
       "      <td>0.006</td>\n",
       "      <td>3.170e-202</td>\n",
       "      <td>-0.143</td>\n",
       "      <td>0.010</td>\n",
       "      <td>1.055e-48</td>\n",
       "      <td>-0.256</td>\n",
       "      <td>0.015</td>\n",
       "      <td>8.514e-68</td>\n",
       "      <td>-0.092</td>\n",
       "      <td>4.139e-03</td>\n",
       "      <td>1.767e-108</td>\n",
       "      <td>-1.789e-01</td>\n",
       "      <td>0.008</td>\n",
       "      <td>1.370e-112</td>\n",
       "      <td>-0.230</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.515e-101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(gender, levels=GENDERS)[T.F]</th>\n",
       "      <td>-2.852e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>1.196e-17</td>\n",
       "      <td>-0.052</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.845e-19</td>\n",
       "      <td>-0.033</td>\n",
       "      <td>0.012</td>\n",
       "      <td>6.140e-03</td>\n",
       "      <td>0.026</td>\n",
       "      <td>2.934e-03</td>\n",
       "      <td>7.435e-19</td>\n",
       "      <td>3.812e-04</td>\n",
       "      <td>0.006</td>\n",
       "      <td>9.488e-01</td>\n",
       "      <td>0.034</td>\n",
       "      <td>0.011</td>\n",
       "      <td>1.590e-03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(gender, levels=GENDERS)[T.M]</th>\n",
       "      <td>-3.870e-02</td>\n",
       "      <td>0.003</td>\n",
       "      <td>2.422e-36</td>\n",
       "      <td>-0.030</td>\n",
       "      <td>0.005</td>\n",
       "      <td>1.872e-08</td>\n",
       "      <td>-0.006</td>\n",
       "      <td>0.012</td>\n",
       "      <td>6.056e-01</td>\n",
       "      <td>-0.002</td>\n",
       "      <td>2.597e-03</td>\n",
       "      <td>3.487e-01</td>\n",
       "      <td>-1.142e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>3.326e-02</td>\n",
       "      <td>0.005</td>\n",
       "      <td>0.010</td>\n",
       "      <td>6.570e-01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                        First           \\\n",
       "                                                         Full            \n",
       "                                                        Coef. Std.Err.   \n",
       "Intercept                                          -1.957e+00    0.031   \n",
       "I(sink_prev_ncites == 0)[T.True]                    1.265e-01    0.004   \n",
       "np.log10(sink_prev_ncites + 1)                     -4.433e-01    0.006   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)             -1.698e-01    0.002   \n",
       "I(auth_prev_papers == 0)[T.True]                   -1.421e+00    0.009   \n",
       "I(auth_prev_papers == 1)[T.True]                   -2.357e-01    0.005   \n",
       "np.log10(auth_prev_papers + 1)                      1.610e+00    0.007   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)             -9.062e-02    0.002   \n",
       "source_is_journal[T.True]                           2.661e-01    0.011   \n",
       "source_is_review[T.True]                           -7.869e-02    0.003   \n",
       "source_is_case_rep[T.True]                         -9.404e-01    0.007   \n",
       "source_is_let_ed_com[T.True]                       -4.295e-01    0.011   \n",
       "sink_is_journal[T.True]                             3.283e-01    0.011   \n",
       "sink_is_review[T.True]                             -7.826e-01    0.003   \n",
       "sink_is_case_rep[T.True]                           -6.369e-01    0.006   \n",
       "sink_is_let_ed_com[T.True]                         -2.838e-01    0.011   \n",
       "I(year_span < 0)[T.True]                           -5.564e-01    0.016   \n",
       "I(year_span == 0)[T.True]                           5.382e-01    0.006   \n",
       "mf.score_log_1(year_span)                           5.387e-01    0.015   \n",
       "I(mf.score_log_1(year_span) ** 2)                  -1.281e+00    0.010   \n",
       "journal_same[T.True]                                4.567e-01    0.003   \n",
       "I(jj_sim == 0)[T.True]                             -2.805e-01    0.008   \n",
       "np.log10(jj_sim + 1)                                6.298e-02    0.010   \n",
       "I(np.log10(jj_sim + 1) ** 2)                        5.949e-03    0.003   \n",
       "I(source_ncites == 1)[T.True]                      -2.997e-01    0.022   \n",
       "np.log10(source_ncites)                            -1.647e+00    0.034   \n",
       "I(np.log10(source_ncites) ** 2)                     8.198e-01    0.023   \n",
       "I(np.log10(source_ncites) ** 3)                    -2.294e-01    0.005   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN... -4.468e-02    0.006   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]   -5.242e-03    0.003   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...  2.669e-01    0.006   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...  2.382e-02    0.004   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR... -4.425e-02    0.005   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT... -1.010e-01    0.007   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA... -3.874e-03    0.004   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH... -4.120e-01    0.010   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...  8.788e-02    0.005   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP... -1.223e-01    0.006   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...  6.808e-02    0.007   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...  1.074e-01    0.007   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN... -4.363e-02    0.010   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT... -6.440e-02    0.003   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)      -1.058e-01    0.001   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)         9.484e-02    0.008   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *... -2.774e-02    0.001   \n",
       "source_is_eng[T.True]                               6.494e-01    0.007   \n",
       "sink_is_eng[T.True]                                -1.846e-01    0.010   \n",
       "I(source_n_authors > 20)[T.True]                    1.722e-04    0.015   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))          8.239e-02    0.018   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2) -2.563e-01    0.013   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.426e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  5.313e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.562e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.021e-01    0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.401e-02    0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -7.411e-02    0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -5.946e-02    0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -6.379e-02    0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.355e-01    0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -5.473e-02    0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.040e-02    0.007   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.869e-01    0.007   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.261e-02    0.018   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.807e-01    0.004   \n",
       "I(source_n_mesh_ex == 0)[T.True]                   -6.666e-01    0.013   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                     -9.292e-02    0.015   \n",
       "np.log10(source_n_mesh_ex + 1)                     -1.704e-01    0.005   \n",
       "np.log10(sink_n_mesh_ex + 1)                       -1.681e-01    0.006   \n",
       "C(gender, levels=GENDERS)[T.F]                     -2.852e-02    0.003   \n",
       "C(gender, levels=GENDERS)[T.M]                     -3.870e-02    0.003   \n",
       "\n",
       "                                                                           \\\n",
       "                                                               Filtered 1   \n",
       "                                                         P>|z|      Coef.   \n",
       "Intercept                                            0.000e+00     -4.047   \n",
       "I(sink_prev_ncites == 0)[T.True]                    1.064e-235      0.166   \n",
       "np.log10(sink_prev_ncites + 1)                       0.000e+00     -0.147   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)               0.000e+00     -0.226   \n",
       "I(auth_prev_papers == 0)[T.True]                     0.000e+00     -1.232   \n",
       "I(auth_prev_papers == 1)[T.True]                     0.000e+00     -0.159   \n",
       "np.log10(auth_prev_papers + 1)                       0.000e+00      1.862   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)              1.091e-294     -0.127   \n",
       "source_is_journal[T.True]                           6.473e-129        NaN   \n",
       "source_is_review[T.True]                            5.938e-199        NaN   \n",
       "source_is_case_rep[T.True]                           0.000e+00        NaN   \n",
       "source_is_let_ed_com[T.True]                         0.000e+00        NaN   \n",
       "sink_is_journal[T.True]                             1.814e-181      0.548   \n",
       "sink_is_review[T.True]                               0.000e+00     -0.741   \n",
       "sink_is_case_rep[T.True]                             0.000e+00     -0.882   \n",
       "sink_is_let_ed_com[T.True]                          9.580e-138     -0.421   \n",
       "I(year_span < 0)[T.True]                            2.011e-275     -0.445   \n",
       "I(year_span == 0)[T.True]                            0.000e+00      0.446   \n",
       "mf.score_log_1(year_span)                           2.008e-275     -0.204   \n",
       "I(mf.score_log_1(year_span) ** 2)                    0.000e+00     -0.767   \n",
       "journal_same[T.True]                                 0.000e+00      0.509   \n",
       "I(jj_sim == 0)[T.True]                              1.735e-259     -0.172   \n",
       "np.log10(jj_sim + 1)                                 3.421e-10      0.205   \n",
       "I(np.log10(jj_sim + 1) ** 2)                         4.459e-02     -0.044   \n",
       "I(source_ncites == 1)[T.True]                        1.212e-41        NaN   \n",
       "np.log10(source_ncites)                              0.000e+00     -3.223   \n",
       "I(np.log10(source_ncites) ** 2)                     4.438e-269      2.579   \n",
       "I(np.log10(source_ncites) ** 3)                      0.000e+00     -0.771   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...   3.435e-12        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]     1.058e-01        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...   0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...   9.888e-09        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...   3.363e-17        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...   3.885e-54        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...   3.608e-01        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...   0.000e+00        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...   1.664e-63        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...   1.850e-83        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...   1.152e-21        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...   4.490e-58        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...   3.026e-05        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...  8.308e-128        NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)        0.000e+00     -0.096   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          1.753e-35      0.200   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...  1.412e-100     -0.044   \n",
       "source_is_eng[T.True]                                0.000e+00        NaN   \n",
       "sink_is_eng[T.True]                                  6.033e-81      2.075   \n",
       "I(source_n_authors > 20)[T.True]                     9.906e-01     -0.037   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))           3.324e-06      0.242   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   1.194e-85     -0.336   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.087e-04     -0.033   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   2.200e-36     -0.038   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   5.136e-29      0.180   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.235e-60     -0.147   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.985e-22      0.116   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.704e-63     -0.066   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   7.073e-23     -0.092   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   5.197e-34      0.013   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.099e-172      0.030   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.299e-20     -0.028   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   3.702e-06      0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  2.492e-146     -0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   4.875e-01     -0.062   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   0.000e+00     -0.110   \n",
       "I(source_n_mesh_ex == 0)[T.True]                     0.000e+00     -0.511   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                       1.355e-09      0.019   \n",
       "np.log10(source_n_mesh_ex + 1)                      2.015e-212     -0.107   \n",
       "np.log10(sink_n_mesh_ex + 1)                        3.170e-202     -0.143   \n",
       "C(gender, levels=GENDERS)[T.F]                       1.196e-17     -0.052   \n",
       "C(gender, levels=GENDERS)[T.M]                       2.422e-36     -0.030   \n",
       "\n",
       "                                                                         \\\n",
       "                                                                          \n",
       "                                                   Std.Err.       P>|z|   \n",
       "Intercept                                             0.289   2.037e-44   \n",
       "I(sink_prev_ncites == 0)[T.True]                      0.007  4.064e-120   \n",
       "np.log10(sink_prev_ncites + 1)                        0.011   7.272e-39   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                0.004   0.000e+00   \n",
       "I(auth_prev_papers == 0)[T.True]                      0.018   0.000e+00   \n",
       "I(auth_prev_papers == 1)[T.True]                      0.010   2.285e-58   \n",
       "np.log10(auth_prev_papers + 1)                        0.013   0.000e+00   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                0.004  9.929e-187   \n",
       "source_is_journal[T.True]                               NaN         NaN   \n",
       "source_is_review[T.True]                                NaN         NaN   \n",
       "source_is_case_rep[T.True]                              NaN         NaN   \n",
       "source_is_let_ed_com[T.True]                            NaN         NaN   \n",
       "sink_is_journal[T.True]                               0.022  5.741e-140   \n",
       "sink_is_review[T.True]                                0.005   0.000e+00   \n",
       "sink_is_case_rep[T.True]                              0.012   0.000e+00   \n",
       "sink_is_let_ed_com[T.True]                            0.021   1.665e-86   \n",
       "I(year_span < 0)[T.True]                              0.028   1.636e-55   \n",
       "I(year_span == 0)[T.True]                             0.010   0.000e+00   \n",
       "mf.score_log_1(year_span)                             0.026   6.462e-15   \n",
       "I(mf.score_log_1(year_span) ** 2)                     0.016   0.000e+00   \n",
       "journal_same[T.True]                                  0.005   0.000e+00   \n",
       "I(jj_sim == 0)[T.True]                                0.015   4.417e-31   \n",
       "np.log10(jj_sim + 1)                                  0.019   3.512e-28   \n",
       "I(np.log10(jj_sim + 1) ** 2)                          0.006   3.126e-15   \n",
       "I(source_ncites == 1)[T.True]                           NaN         NaN   \n",
       "np.log10(source_ncites)                               0.562   1.002e-08   \n",
       "I(np.log10(source_ncites) ** 2)                       0.379   9.953e-12   \n",
       "I(np.log10(source_ncites) ** 3)                       0.084   3.470e-20   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]        NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...      NaN         NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...      NaN         NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)         0.002   0.000e+00   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)           0.014   2.878e-49   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...    0.002   7.102e-82   \n",
       "source_is_eng[T.True]                                   NaN         NaN   \n",
       "sink_is_eng[T.True]                                   0.084  2.546e-133   \n",
       "I(source_n_authors > 20)[T.True]                      0.025   1.343e-01   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))            0.030   1.357e-15   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)    0.023   3.234e-50   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.006   2.974e-08   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   1.021e-07   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.006  9.166e-200   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.009   7.110e-61   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   1.022e-50   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   4.727e-16   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.010   7.999e-20   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.007   4.308e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.012   8.899e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.009   3.004e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.014   6.575e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.012   6.557e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.026   1.588e-02   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...    0.008   4.316e-46   \n",
       "I(source_n_mesh_ex == 0)[T.True]                      0.024  5.006e-100   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                        0.028   4.991e-01   \n",
       "np.log10(source_n_mesh_ex + 1)                        0.009   2.730e-30   \n",
       "np.log10(sink_n_mesh_ex + 1)                          0.010   1.055e-48   \n",
       "C(gender, levels=GENDERS)[T.F]                        0.006   1.845e-19   \n",
       "C(gender, levels=GENDERS)[T.M]                        0.005   1.872e-08   \n",
       "\n",
       "                                                                        \\\n",
       "                                                   Filtered 2            \n",
       "                                                        Coef. Std.Err.   \n",
       "Intercept                                              -3.542    0.482   \n",
       "I(sink_prev_ncites == 0)[T.True]                        0.156    0.011   \n",
       "np.log10(sink_prev_ncites + 1)                         -0.137    0.016   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                 -0.181    0.006   \n",
       "I(auth_prev_papers == 0)[T.True]                       -1.236    0.030   \n",
       "I(auth_prev_papers == 1)[T.True]                       -0.158    0.017   \n",
       "np.log10(auth_prev_papers + 1)                          1.911    0.020   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                 -0.156    0.006   \n",
       "source_is_journal[T.True]                                 NaN      NaN   \n",
       "source_is_review[T.True]                                  NaN      NaN   \n",
       "source_is_case_rep[T.True]                                NaN      NaN   \n",
       "source_is_let_ed_com[T.True]                              NaN      NaN   \n",
       "sink_is_journal[T.True]                                 0.496    0.032   \n",
       "sink_is_review[T.True]                                 -0.639    0.007   \n",
       "sink_is_case_rep[T.True]                               -0.861    0.017   \n",
       "sink_is_let_ed_com[T.True]                             -0.451    0.031   \n",
       "I(year_span < 0)[T.True]                               -0.327    0.041   \n",
       "I(year_span == 0)[T.True]                               0.580    0.015   \n",
       "mf.score_log_1(year_span)                               0.152    0.038   \n",
       "I(mf.score_log_1(year_span) ** 2)                      -0.857    0.023   \n",
       "journal_same[T.True]                                    0.501    0.007   \n",
       "I(jj_sim == 0)[T.True]                                 -0.163    0.022   \n",
       "np.log10(jj_sim + 1)                                    0.255    0.027   \n",
       "I(np.log10(jj_sim + 1) ** 2)                           -0.056    0.008   \n",
       "I(source_ncites == 1)[T.True]                             NaN      NaN   \n",
       "np.log10(source_ncites)                                -7.395    0.816   \n",
       "I(np.log10(source_ncites) ** 2)                         5.400    0.550   \n",
       "I(np.log10(source_ncites) ** 3)                        -1.386    0.122   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]          NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...        NaN      NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)          -0.112    0.004   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)             0.158    0.021   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...     -0.037    0.004   \n",
       "source_is_eng[T.True]                                     NaN      NaN   \n",
       "sink_is_eng[T.True]                                     3.625    0.268   \n",
       "I(source_n_authors > 20)[T.True]                       -0.047    0.034   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))              0.303    0.045   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)     -0.353    0.033   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...        NaN      NaN   \n",
       "I(source_n_mesh_ex == 0)[T.True]                       -0.628    0.036   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                         -0.156    0.044   \n",
       "np.log10(source_n_mesh_ex + 1)                         -0.150    0.014   \n",
       "np.log10(sink_n_mesh_ex + 1)                           -0.256    0.015   \n",
       "C(gender, levels=GENDERS)[T.F]                         -0.033    0.012   \n",
       "C(gender, levels=GENDERS)[T.M]                         -0.006    0.012   \n",
       "\n",
       "                                                                 Last  \\\n",
       "                                                                 Full   \n",
       "                                                         P>|z|  Coef.   \n",
       "Intercept                                            1.978e-13 -2.790   \n",
       "I(sink_prev_ncites == 0)[T.True]                     2.378e-47  0.240   \n",
       "np.log10(sink_prev_ncites + 1)                       5.717e-17 -0.079   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)              2.708e-203 -0.169   \n",
       "I(auth_prev_papers == 0)[T.True]                     0.000e+00 -1.232   \n",
       "I(auth_prev_papers == 1)[T.True]                     1.823e-21 -0.187   \n",
       "np.log10(auth_prev_papers + 1)                       0.000e+00  2.172   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)              4.687e-130 -0.340   \n",
       "source_is_journal[T.True]                                  NaN  0.229   \n",
       "source_is_review[T.True]                                   NaN -0.249   \n",
       "source_is_case_rep[T.True]                                 NaN -0.948   \n",
       "source_is_let_ed_com[T.True]                               NaN -0.343   \n",
       "sink_is_journal[T.True]                              3.460e-55  0.324   \n",
       "sink_is_review[T.True]                               0.000e+00 -0.641   \n",
       "sink_is_case_rep[T.True]                             0.000e+00 -0.731   \n",
       "sink_is_let_ed_com[T.True]                           7.459e-47 -0.292   \n",
       "I(year_span < 0)[T.True]                             2.789e-15 -0.383   \n",
       "I(year_span == 0)[T.True]                            0.000e+00  0.735   \n",
       "mf.score_log_1(year_span)                            7.348e-05  1.232   \n",
       "I(mf.score_log_1(year_span) ** 2)                   1.099e-302 -1.402   \n",
       "journal_same[T.True]                                 0.000e+00  0.534   \n",
       "I(jj_sim == 0)[T.True]                               1.448e-13 -0.296   \n",
       "np.log10(jj_sim + 1)                                 1.016e-20  0.126   \n",
       "I(np.log10(jj_sim + 1) ** 2)                         8.183e-12 -0.050   \n",
       "I(source_ncites == 1)[T.True]                              NaN -0.405   \n",
       "np.log10(source_ncites)                              1.230e-19 -2.221   \n",
       "I(np.log10(source_ncites) ** 2)                      1.023e-22  1.463   \n",
       "I(np.log10(source_ncites) ** 3)                      5.992e-30 -0.387   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...         NaN -0.158   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]           NaN -0.090   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...         NaN -0.092   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...         NaN -0.169   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...         NaN -0.116   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...         NaN -0.298   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...         NaN -0.069   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...         NaN -0.432   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...         NaN -0.099   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...         NaN -0.184   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...         NaN -0.179   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...         NaN -0.083   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...         NaN -0.206   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...         NaN -0.213   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)       1.108e-214 -0.139   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          2.289e-14  0.048   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...   6.418e-26 -0.022   \n",
       "source_is_eng[T.True]                                      NaN  0.701   \n",
       "sink_is_eng[T.True]                                  1.149e-41 -0.031   \n",
       "I(source_n_authors > 20)[T.True]                     1.627e-01 -0.050   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))           1.063e-11 -0.444   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   2.710e-26  0.066   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  0.062   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  0.082   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  0.034   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN -0.048   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  0.032   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN -0.038   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  0.026   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  0.020   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN  0.070   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN -0.085   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN -0.004   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN -0.076   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN -0.046   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...         NaN -0.057   \n",
       "I(source_n_mesh_ex == 0)[T.True]                     1.575e-66 -0.892   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                       3.500e-04  0.005   \n",
       "np.log10(source_n_mesh_ex + 1)                       2.657e-26 -0.251   \n",
       "np.log10(sink_n_mesh_ex + 1)                         8.514e-68 -0.092   \n",
       "C(gender, levels=GENDERS)[T.F]                       6.140e-03  0.026   \n",
       "C(gender, levels=GENDERS)[T.M]                       6.056e-01 -0.002   \n",
       "\n",
       "                                                                           \\\n",
       "                                                                            \n",
       "                                                     Std.Err.       P>|z|   \n",
       "Intercept                                           2.674e-02   0.000e+00   \n",
       "I(sink_prev_ncites == 0)[T.True]                    3.101e-03   0.000e+00   \n",
       "np.log10(sink_prev_ncites + 1)                      4.424e-03   4.191e-71   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)              1.530e-03   0.000e+00   \n",
       "I(auth_prev_papers == 0)[T.True]                    1.639e-02   0.000e+00   \n",
       "I(auth_prev_papers == 1)[T.True]                    1.058e-02   3.454e-70   \n",
       "np.log10(auth_prev_papers + 1)                      7.733e-03   0.000e+00   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)              2.154e-03   0.000e+00   \n",
       "source_is_journal[T.True]                           1.021e-02  1.707e-111   \n",
       "source_is_review[T.True]                            2.125e-03   0.000e+00   \n",
       "source_is_case_rep[T.True]                          5.222e-03   0.000e+00   \n",
       "source_is_let_ed_com[T.True]                        9.952e-03  9.331e-261   \n",
       "sink_is_journal[T.True]                             9.445e-03  1.501e-257   \n",
       "sink_is_review[T.True]                              2.142e-03   0.000e+00   \n",
       "sink_is_case_rep[T.True]                            4.949e-03   0.000e+00   \n",
       "sink_is_let_ed_com[T.True]                          9.374e-03  3.266e-213   \n",
       "I(year_span < 0)[T.True]                            1.459e-02  4.270e-152   \n",
       "I(year_span == 0)[T.True]                           4.656e-03   0.000e+00   \n",
       "mf.score_log_1(year_span)                           1.131e-02   0.000e+00   \n",
       "I(mf.score_log_1(year_span) ** 2)                   6.820e-03   0.000e+00   \n",
       "journal_same[T.True]                                2.218e-03   0.000e+00   \n",
       "I(jj_sim == 0)[T.True]                              6.332e-03   0.000e+00   \n",
       "np.log10(jj_sim + 1)                                8.068e-03   4.277e-55   \n",
       "I(np.log10(jj_sim + 1) ** 2)                        2.456e-03   1.148e-91   \n",
       "I(source_ncites == 1)[T.True]                       2.079e-02   2.477e-84   \n",
       "np.log10(source_ncites)                             3.101e-02   0.000e+00   \n",
       "I(np.log10(source_ncites) ** 2)                     2.148e-02   0.000e+00   \n",
       "I(np.log10(source_ncites) ** 3)                     4.793e-03   0.000e+00   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...  5.211e-03  3.526e-203   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]    2.373e-03   0.000e+00   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...  6.048e-03   9.138e-53   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...  3.061e-03   0.000e+00   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...  3.820e-03  1.689e-203   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...  4.877e-03   0.000e+00   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...  3.035e-03  3.086e-113   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...  7.300e-03   0.000e+00   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...  4.043e-03  4.252e-132   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...  4.889e-03  6.632e-310   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...  5.204e-03  5.058e-260   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...  4.978e-03   6.425e-63   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...  8.064e-03  1.078e-143   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...  2.054e-03   0.000e+00   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)       9.809e-04   0.000e+00   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)         5.268e-03   5.457e-20   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...  9.206e-04  1.213e-127   \n",
       "source_is_eng[T.True]                               6.660e-03   0.000e+00   \n",
       "sink_is_eng[T.True]                                 8.947e-03   5.789e-04   \n",
       "I(source_n_authors > 20)[T.True]                    1.263e-02   7.036e-05   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))          1.368e-02  1.041e-230   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)  1.004e-02   3.481e-11   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  2.555e-03  6.084e-130   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.304e-03  1.645e-134   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.646e-03   4.623e-21   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  5.959e-03   6.342e-16   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.138e-03   4.544e-15   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.142e-03   6.573e-33   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.390e-03   5.446e-09   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.486e-03   6.955e-06   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.496e-03   2.459e-89   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  5.591e-03   8.268e-52   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  4.629e-03   4.259e-01   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  6.431e-03   1.624e-32   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.422e-02   1.126e-03   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.398e-03   5.873e-63   \n",
       "I(source_n_mesh_ex == 0)[T.True]                    9.581e-03   0.000e+00   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                      1.108e-02   6.763e-01   \n",
       "np.log10(source_n_mesh_ex + 1)                      4.138e-03   0.000e+00   \n",
       "np.log10(sink_n_mesh_ex + 1)                        4.139e-03  1.767e-108   \n",
       "C(gender, levels=GENDERS)[T.F]                      2.934e-03   7.435e-19   \n",
       "C(gender, levels=GENDERS)[T.M]                      2.597e-03   3.487e-01   \n",
       "\n",
       "                                                                        \\\n",
       "                                                   Filtered 1            \n",
       "                                                        Coef. Std.Err.   \n",
       "Intercept                                           8.550e-01    0.488   \n",
       "I(sink_prev_ncites == 0)[T.True]                    2.968e-01    0.006   \n",
       "np.log10(sink_prev_ncites + 1)                      1.991e-01    0.009   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)             -2.088e-01    0.003   \n",
       "I(auth_prev_papers == 0)[T.True]                   -1.259e+00    0.039   \n",
       "I(auth_prev_papers == 1)[T.True]                   -2.089e-01    0.024   \n",
       "np.log10(auth_prev_papers + 1)                      2.458e+00    0.016   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)             -4.035e-01    0.004   \n",
       "source_is_journal[T.True]                                 NaN      NaN   \n",
       "source_is_review[T.True]                                  NaN      NaN   \n",
       "source_is_case_rep[T.True]                                NaN      NaN   \n",
       "source_is_let_ed_com[T.True]                              NaN      NaN   \n",
       "sink_is_journal[T.True]                             5.370e-01    0.020   \n",
       "sink_is_review[T.True]                             -6.616e-01    0.004   \n",
       "sink_is_case_rep[T.True]                           -1.038e+00    0.011   \n",
       "sink_is_let_ed_com[T.True]                         -3.867e-01    0.019   \n",
       "I(year_span < 0)[T.True]                           -3.821e-01    0.032   \n",
       "I(year_span == 0)[T.True]                           7.146e-01    0.009   \n",
       "mf.score_log_1(year_span)                           8.259e-01    0.022   \n",
       "I(mf.score_log_1(year_span) ** 2)                  -1.160e+00    0.013   \n",
       "journal_same[T.True]                                6.124e-01    0.004   \n",
       "I(jj_sim == 0)[T.True]                             -1.677e-01    0.013   \n",
       "np.log10(jj_sim + 1)                                3.594e-01    0.016   \n",
       "I(np.log10(jj_sim + 1) ** 2)                       -1.456e-01    0.005   \n",
       "I(source_ncites == 1)[T.True]                             NaN      NaN   \n",
       "np.log10(source_ncites)                            -1.776e+01    1.055   \n",
       "I(np.log10(source_ncites) ** 2)                     1.354e+01    0.772   \n",
       "I(np.log10(source_ncites) ** 3)                    -3.438e+00    0.186   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]          NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...        NaN      NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...        NaN      NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)      -1.345e-01    0.002   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)         1.570e-01    0.010   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *... -4.152e-02    0.002   \n",
       "source_is_eng[T.True]                                     NaN      NaN   \n",
       "sink_is_eng[T.True]                                 2.656e+00    0.102   \n",
       "I(source_n_authors > 20)[T.True]                    2.323e-02    0.024   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))         -1.366e-01    0.026   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2) -1.128e-01    0.019   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  5.838e-02    0.005   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -2.033e-02    0.007   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.815e-01    0.006   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.517e-01    0.011   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  1.617e-01    0.008   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -2.034e-02    0.007   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  5.540e-02    0.008   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  7.448e-02    0.007   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  9.378e-02    0.009   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -3.558e-02    0.009   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -2.958e-02    0.011   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  9.743e-02    0.016   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO... -1.005e-01    0.023   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  6.383e-02    0.007   \n",
       "I(source_n_mesh_ex == 0)[T.True]                   -7.059e-01    0.020   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                     -9.296e-02    0.022   \n",
       "np.log10(source_n_mesh_ex + 1)                     -1.537e-01    0.008   \n",
       "np.log10(sink_n_mesh_ex + 1)                       -1.789e-01    0.008   \n",
       "C(gender, levels=GENDERS)[T.F]                      3.812e-04    0.006   \n",
       "C(gender, levels=GENDERS)[T.M]                     -1.142e-02    0.005   \n",
       "\n",
       "                                                                           \\\n",
       "                                                               Filtered 2   \n",
       "                                                         P>|z|      Coef.   \n",
       "Intercept                                            7.964e-02     -0.491   \n",
       "I(sink_prev_ncites == 0)[T.True]                     0.000e+00      0.292   \n",
       "np.log10(sink_prev_ncites + 1)                      6.238e-119      0.215   \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)               0.000e+00     -0.201   \n",
       "I(auth_prev_papers == 0)[T.True]                    7.020e-231     -1.236   \n",
       "I(auth_prev_papers == 1)[T.True]                     5.352e-18     -0.189   \n",
       "np.log10(auth_prev_papers + 1)                       0.000e+00      2.468   \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)               0.000e+00     -0.401   \n",
       "source_is_journal[T.True]                                  NaN        NaN   \n",
       "source_is_review[T.True]                                   NaN        NaN   \n",
       "source_is_case_rep[T.True]                                 NaN        NaN   \n",
       "source_is_let_ed_com[T.True]                               NaN        NaN   \n",
       "sink_is_journal[T.True]                             3.782e-163      0.575   \n",
       "sink_is_review[T.True]                               0.000e+00     -0.641   \n",
       "sink_is_case_rep[T.True]                             0.000e+00     -1.046   \n",
       "sink_is_let_ed_com[T.True]                           9.806e-90     -0.373   \n",
       "I(year_span < 0)[T.True]                             9.889e-33     -0.394   \n",
       "I(year_span == 0)[T.True]                            0.000e+00      0.748   \n",
       "mf.score_log_1(year_span)                            0.000e+00      0.856   \n",
       "I(mf.score_log_1(year_span) ** 2)                    0.000e+00     -1.155   \n",
       "journal_same[T.True]                                 0.000e+00      0.616   \n",
       "I(jj_sim == 0)[T.True]                               2.733e-40     -0.188   \n",
       "np.log10(jj_sim + 1)                                1.723e-107      0.358   \n",
       "I(np.log10(jj_sim + 1) ** 2)                        3.399e-185     -0.143   \n",
       "I(source_ncites == 1)[T.True]                              NaN        NaN   \n",
       "np.log10(source_ncites)                              1.635e-63    -17.484   \n",
       "I(np.log10(source_ncites) ** 2)                      6.040e-69     13.420   \n",
       "I(np.log10(source_ncites) ** 3)                      4.052e-76     -3.422   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]           NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...         NaN        NaN   \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...         NaN        NaN   \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)        0.000e+00     -0.157   \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)          1.388e-52      0.199   \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...  3.386e-118     -0.049   \n",
       "source_is_eng[T.True]                                      NaN        NaN   \n",
       "sink_is_eng[T.True]                                 4.895e-149      3.959   \n",
       "I(source_n_authors > 20)[T.True]                     3.258e-01     -0.037   \n",
       "np.log10(np.clip(source_n_authors, 0, 20))           1.286e-07     -0.183   \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)   4.115e-09     -0.087   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   5.079e-36        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   2.542e-03        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...  3.474e-188        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.340e-43        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.828e-83        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   2.611e-03        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   3.340e-11        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   2.266e-30        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.986e-24        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   1.047e-04        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   6.362e-03        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   2.776e-09        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   7.966e-06        NaN   \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...   9.465e-22        NaN   \n",
       "I(source_n_mesh_ex == 0)[T.True]                    6.641e-269     -0.908   \n",
       "I(sink_n_mesh_ex == 0)[T.True]                       2.792e-05     -0.106   \n",
       "np.log10(source_n_mesh_ex + 1)                       7.027e-84     -0.239   \n",
       "np.log10(sink_n_mesh_ex + 1)                        1.370e-112     -0.230   \n",
       "C(gender, levels=GENDERS)[T.F]                       9.488e-01      0.034   \n",
       "C(gender, levels=GENDERS)[T.M]                       3.326e-02      0.005   \n",
       "\n",
       "                                                                         \n",
       "                                                                         \n",
       "                                                   Std.Err.       P>|z|  \n",
       "Intercept                                             0.695   4.799e-01  \n",
       "I(sink_prev_ncites == 0)[T.True]                      0.009  2.798e-249  \n",
       "np.log10(sink_prev_ncites + 1)                        0.012   2.376e-77  \n",
       "I(np.log10(sink_prev_ncites + 1) ** 2)                0.004   0.000e+00  \n",
       "I(auth_prev_papers == 0)[T.True]                      0.054  3.506e-114  \n",
       "I(auth_prev_papers == 1)[T.True]                      0.035   6.216e-08  \n",
       "np.log10(auth_prev_papers + 1)                        0.022   0.000e+00  \n",
       "I(np.log10(auth_prev_papers + 1) ** 2)                0.006   0.000e+00  \n",
       "source_is_journal[T.True]                               NaN         NaN  \n",
       "source_is_review[T.True]                                NaN         NaN  \n",
       "source_is_case_rep[T.True]                              NaN         NaN  \n",
       "source_is_let_ed_com[T.True]                            NaN         NaN  \n",
       "sink_is_journal[T.True]                               0.027  8.207e-101  \n",
       "sink_is_review[T.True]                                0.006   0.000e+00  \n",
       "sink_is_case_rep[T.True]                              0.014   0.000e+00  \n",
       "sink_is_let_ed_com[T.True]                            0.026   5.465e-46  \n",
       "I(year_span < 0)[T.True]                              0.044   2.207e-19  \n",
       "I(year_span == 0)[T.True]                             0.013   0.000e+00  \n",
       "mf.score_log_1(year_span)                             0.029  2.733e-190  \n",
       "I(mf.score_log_1(year_span) ** 2)                     0.017   0.000e+00  \n",
       "journal_same[T.True]                                  0.005   0.000e+00  \n",
       "I(jj_sim == 0)[T.True]                                0.017   8.701e-29  \n",
       "np.log10(jj_sim + 1)                                  0.022   2.322e-61  \n",
       "I(np.log10(jj_sim + 1) ** 2)                          0.007  5.665e-103  \n",
       "I(source_ncites == 1)[T.True]                           NaN         NaN  \n",
       "np.log10(source_ncites)                               1.420   7.628e-35  \n",
       "I(np.log10(source_ncites) ** 2)                       1.039   3.616e-38  \n",
       "I(np.log10(source_ncites) ** 3)                       0.251   2.177e-42  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UN...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.UK]        NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.JA...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.GE...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.FR...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IT...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CA...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.CH...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.AU...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SP...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.NE...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.SW...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.IN...      NaN         NaN  \n",
       "C(source_country, levels=TOP_15_COUNTRIES)[T.OT...      NaN         NaN  \n",
       "np.log10(np.nan_to_num(source_V_novelty) + 1)         0.003   0.000e+00  \n",
       "np.log10(np.nan_to_num(sink_V_novelty) + 1)           0.014   3.391e-45  \n",
       "I(np.log10(np.nan_to_num(sink_V_novelty) + 1) *...    0.002   9.311e-89  \n",
       "source_is_eng[T.True]                                   NaN         NaN  \n",
       "sink_is_eng[T.True]                                   0.268   1.711e-49  \n",
       "I(source_n_authors > 20)[T.True]                      0.032   2.452e-01  \n",
       "np.log10(np.clip(source_n_authors, 0, 20))            0.035   1.756e-07  \n",
       "I(np.log10(np.clip(source_n_authors, 0, 20)) ** 2)    0.026   8.039e-04  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "mf.MC(eth1, eth2, weights=eth_weight, levels=TO...      NaN         NaN  \n",
       "I(source_n_mesh_ex == 0)[T.True]                      0.028  3.226e-237  \n",
       "I(sink_n_mesh_ex == 0)[T.True]                        0.030   4.778e-04  \n",
       "np.log10(source_n_mesh_ex + 1)                        0.011  2.737e-109  \n",
       "np.log10(sink_n_mesh_ex + 1)                          0.011  1.515e-101  \n",
       "C(gender, levels=GENDERS)[T.F]                        0.011   1.590e-03  \n",
       "C(gender, levels=GENDERS)[T.M]                        0.010   6.570e-01  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_t_both = merge_results([results_full, results_flt_1, results_flt_2,\n",
    "                           results_full_la, results_flt_1_la, results_flt_2_la],\n",
    "                         keys=[(k1, k2) for k1 in (\"First\", \"Last\") for k2 in (\"Full\", \"Filtered 1\", \"Filtered 2\")])\n",
    "with pd.option_context(\"display.max_columns\", 20, \"display.max_rows\", 200, \"precision\", 3):\n",
    "    display(df_t_both)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [default]",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}