Model fitting and evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import pickle\n",
    "import sklearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "interest = pickle.load(open(\"interest_ready.pickle\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>dtype</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>X1</th>\n",
       "      <td>float64</td>\n",
       "      <td>0.139463</td>\n",
       "      <td>0.043779</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X4</th>\n",
       "      <td>float64</td>\n",
       "      <td>14276.968353</td>\n",
       "      <td>8252.264104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X5</th>\n",
       "      <td>float64</td>\n",
       "      <td>14248.306140</td>\n",
       "      <td>8238.064335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X6</th>\n",
       "      <td>float64</td>\n",
       "      <td>14182.000445</td>\n",
       "      <td>8255.138412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X8</th>\n",
       "      <td>float64</td>\n",
       "      <td>2.808658</td>\n",
       "      <td>1.314939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X9</th>\n",
       "      <td>float64</td>\n",
       "      <td>12.052704</td>\n",
       "      <td>6.496930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X11</th>\n",
       "      <td>float64</td>\n",
       "      <td>5.991935</td>\n",
       "      <td>3.534170</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X13</th>\n",
       "      <td>float64</td>\n",
       "      <td>73147.445129</td>\n",
       "      <td>51749.421285</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X15</th>\n",
       "      <td>float64</td>\n",
       "      <td>13.027980</td>\n",
       "      <td>1.167932</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X21</th>\n",
       "      <td>float64</td>\n",
       "      <td>16.997242</td>\n",
       "      <td>7.697223</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X22</th>\n",
       "      <td>float64</td>\n",
       "      <td>0.274286</td>\n",
       "      <td>0.779294</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X23</th>\n",
       "      <td>float64</td>\n",
       "      <td>227.144860</td>\n",
       "      <td>86.317714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X24</th>\n",
       "      <td>float64</td>\n",
       "      <td>0.817911</td>\n",
       "      <td>1.059264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X27</th>\n",
       "      <td>float64</td>\n",
       "      <td>11.118081</td>\n",
       "      <td>4.896081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X28</th>\n",
       "      <td>float64</td>\n",
       "      <td>0.152689</td>\n",
       "      <td>0.505363</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X29</th>\n",
       "      <td>float64</td>\n",
       "      <td>15974.343946</td>\n",
       "      <td>19765.169424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X30</th>\n",
       "      <td>float64</td>\n",
       "      <td>0.613905</td>\n",
       "      <td>0.487126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X31</th>\n",
       "      <td>float64</td>\n",
       "      <td>24.978309</td>\n",
       "      <td>11.471413</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X7_ 36 months</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.730971</td>\n",
       "      <td>0.443456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X7_ 60 months</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.269029</td>\n",
       "      <td>0.443456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X12_MORTGAGE</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.430569</td>\n",
       "      <td>0.495157</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X12_OWN</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.073678</td>\n",
       "      <td>0.261246</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X12_RENT</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.342070</td>\n",
       "      <td>0.474404</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X14_VERIFIED - income</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.374761</td>\n",
       "      <td>0.484062</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X14_VERIFIED - income source</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.307019</td>\n",
       "      <td>0.461258</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X14_not verified</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.318220</td>\n",
       "      <td>0.465786</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X32_f</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.686158</td>\n",
       "      <td>0.464054</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>X32_w</th>\n",
       "      <td>uint8</td>\n",
       "      <td>0.313842</td>\n",
       "      <td>0.464054</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                dtype          mean           std\n",
       "X1                            float64      0.139463      0.043779\n",
       "X4                            float64  14276.968353   8252.264104\n",
       "X5                            float64  14248.306140   8238.064335\n",
       "X6                            float64  14182.000445   8255.138412\n",
       "X8                            float64      2.808658      1.314939\n",
       "X9                            float64     12.052704      6.496930\n",
       "X11                           float64      5.991935      3.534170\n",
       "X13                           float64  73147.445129  51749.421285\n",
       "X15                           float64     13.027980      1.167932\n",
       "X21                           float64     16.997242      7.697223\n",
       "X22                           float64      0.274286      0.779294\n",
       "X23                           float64    227.144860     86.317714\n",
       "X24                           float64      0.817911      1.059264\n",
       "X27                           float64     11.118081      4.896081\n",
       "X28                           float64      0.152689      0.505363\n",
       "X29                           float64  15974.343946  19765.169424\n",
       "X30                           float64      0.613905      0.487126\n",
       "X31                           float64     24.978309     11.471413\n",
       "X7_ 36 months                   uint8      0.730971      0.443456\n",
       "X7_ 60 months                   uint8      0.269029      0.443456\n",
       "X12_MORTGAGE                    uint8      0.430569      0.495157\n",
       "X12_OWN                         uint8      0.073678      0.261246\n",
       "X12_RENT                        uint8      0.342070      0.474404\n",
       "X14_VERIFIED - income           uint8      0.374761      0.484062\n",
       "X14_VERIFIED - income source    uint8      0.307019      0.461258\n",
       "X14_not verified                uint8      0.318220      0.465786\n",
       "X32_f                           uint8      0.686158      0.464054\n",
       "X32_w                           uint8      0.313842      0.464054"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# print summary statistics of raw features\n",
    "df = pd.concat([interest.dtypes, interest.mean(axis=0), interest.std(axis=0)], axis=1)\n",
    "df.columns = [\"dtype\", \"mean\", \"std\"]\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A subset of features have been manually removed in the previous exploratory analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "y = interest.values[:, 0]   # target: first column\n",
    "X = interest.values[:, 1:]  # features: column with index >= 1\n",
    "del interest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Hold-out test set (20%) for final evaluation\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=123)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Forest - Best params:\n",
      "{'estimator__max_samples': 1.0, 'dim_reduction': None, 'estimator__base_estimator__max_depth': 10, 'estimator__base_estimator__min_samples_split': 10}\n",
      "\n",
      "Random Forest - Best model:\n",
      "GridSearchCV(cv=None, error_score='raise',\n",
      "       estimator=Pipeline(steps=[('scaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('dim_reduction', PCA(copy=True, iterated_power='auto', n_components=0.95, random_state=None,\n",
      "  svd_solver='auto', tol=0.0, whiten=False)), ('estimator', BaggingRegressor(base_estimator=DecisionTreeRegressor(criterion='mse'...estimators=50, n_jobs=6, oob_score=False,\n",
      "         random_state=None, verbose=0, warm_start=False))]),\n",
      "       fit_params={}, iid=True, n_jobs=1,\n",
      "       param_grid={'estimator__max_samples': [0.1, 1.0], 'dim_reduction': [PCA(copy=True, iterated_power='auto', n_components=0.95, random_state=None,\n",
      "  svd_solver='auto', tol=0.0, whiten=False), None], 'estimator__base_estimator__max_depth': [10, 20], 'estimator__base_estimator__min_samples_split': [10]},\n",
      "       pre_dispatch='2*n_jobs', refit=True, return_train_score=True,\n",
      "       scoring=None, verbose=0)\n",
      "\n",
      "Random Forest - Score on test set:\n",
      "0.990125540368\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAi0AAAF5CAYAAABTHhoOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKvwPA5RE1FcEu0sAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdf5e6b2c90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.ensemble import BaggingRegressor\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "rf = BaggingRegressor(DecisionTreeRegressor(max_depth=20, min_samples_split=20,\n",
    "                                               min_samples_leaf=1), max_samples=0.1, n_estimators=50, n_jobs=6)\n",
    "\n",
    "pl = Pipeline(steps=[(\"scaler\", StandardScaler()),\n",
    "                     (\"dim_reduction\", PCA(n_components=0.95)),\n",
    "                     (\"estimator\", rf)])\n",
    "\n",
    "model = GridSearchCV(pl, dict(dim_reduction=[PCA(n_components=0.95), None],\n",
    "                              estimator__base_estimator__max_depth=[10, 20],\n",
    "                              estimator__base_estimator__min_samples_split=[10],\n",
    "                              estimator__max_samples=[0.1, 1.0]))\n",
    "model.fit(X_train, y_train)\n",
    "print \"Random Forest - Best params:\"\n",
    "print model.best_params_\n",
    "print\n",
    "print \"Random Forest - Best model:\"\n",
    "print model\n",
    "print \n",
    "print \"Random Forest - Score on test set:\"\n",
    "# R-squared\n",
    "print model.score(X_test, y_test)\n",
    "\n",
    "y_pred = model.predict(X_test)\n",
    "plt.scatter(y_test, y_pred, s=1)\n",
    "plt.xlabel(\"y true\")\n",
    "plt.ylabel(\"y pred\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Bagging SGDRegressor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bagging SGDRegressor - Best params:\n",
      "{'dim_reduction': None, 'estimator__base_estimator__alpha': 0.0001}\n",
      "\n",
      "Bagging SGDRegressor - Best model:\n",
      "GridSearchCV(cv=None, error_score='raise',\n",
      "       estimator=Pipeline(steps=[('scaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('dim_reduction', PCA(copy=True, iterated_power='auto', n_components=0.95, random_state=None,\n",
      "  svd_solver='auto', tol=0.0, whiten=False)), ('estimator', BaggingRegressor(base_estimator=SGDRegressor(alpha=0.0001, average=Fa...estimators=50, n_jobs=6, oob_score=False,\n",
      "         random_state=None, verbose=0, warm_start=False))]),\n",
      "       fit_params={}, iid=True, n_jobs=1,\n",
      "       param_grid={'dim_reduction': [PCA(copy=True, iterated_power='auto', n_components=0.95, random_state=None,\n",
      "  svd_solver='auto', tol=0.0, whiten=False), None], 'estimator__base_estimator__alpha': [0.01, 0.0001]},\n",
      "       pre_dispatch='2*n_jobs', refit=True, return_train_score=True,\n",
      "       scoring=None, verbose=0)\n",
      "\n",
      "Bagging SGDRegressor - Score on test set:\n",
      "0.961596144195\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAi0AAAF5CAYAAABTHhoOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdf508145d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.ensemble import BaggingRegressor\n",
    "from sklearn.linear_model import SGDRegressor\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "bg_sgdr = BaggingRegressor(SGDRegressor(), n_estimators=50,\n",
    "                           max_samples=1.0, n_jobs=6)\n",
    "pl = Pipeline(steps=[(\"scaler\", StandardScaler()),\n",
    "                     (\"dim_reduction\", PCA(n_components=0.95)),\n",
    "                     (\"estimator\", bg_sgdr)])\n",
    "\n",
    "model = GridSearchCV(pl, dict(dim_reduction=[PCA(n_components=0.95), None],\n",
    "                              estimator__base_estimator__alpha=[0.01, 0.0001]))\n",
    "model.fit(X_train, y_train)\n",
    "print \"Bagging SGDRegressor - Best params:\"\n",
    "print model.best_params_\n",
    "print\n",
    "print \"Bagging SGDRegressor - Best model:\"\n",
    "print model\n",
    "print \n",
    "print \"Bagging SGDRegressor - Score on test set:\"\n",
    "# R-squared\n",
    "print model.score(X_test, y_test)\n",
    "\n",
    "y_pred = model.predict(X_test)\n",
    "plt.scatter(y_test, y_pred, s=1)\n",
    "plt.xlabel(\"y true\")\n",
    "plt.ylabel(\"y pred\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "# Bagging Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bagging MLPRegressor - Best params:\n",
      "{'dim_reduction': None, 'estimator__base_estimator__alpha': 0.01}\n",
      "\n",
      "Bagging MLPRegressor - Best model:\n",
      "GridSearchCV(cv=None, error_score='raise',\n",
      "       estimator=Pipeline(steps=[('scaler', StandardScaler(copy=True, with_mean=True, with_std=True)), ('dim_reduction', PCA(copy=True, iterated_power='auto', n_components=None, random_state=None,\n",
      "  svd_solver='auto', tol=0.0, whiten=False)), ('estimator', BaggingRegressor(base_estimator=MLPRegressor(activation='relu', alpha...estimators=20, n_jobs=6, oob_score=False,\n",
      "         random_state=None, verbose=0, warm_start=False))]),\n",
      "       fit_params={}, iid=True, n_jobs=1,\n",
      "       param_grid={'dim_reduction': [PCA(copy=True, iterated_power='auto', n_components=0.95, random_state=None,\n",
      "  svd_solver='auto', tol=0.0, whiten=False), None], 'estimator__base_estimator__alpha': [0.01, 0.0001]},\n",
      "       pre_dispatch='2*n_jobs', refit=True, return_train_score=True,\n",
      "       scoring=None, verbose=0)\n",
      "\n",
      "Bagging MLPRegressor - Score on test set:\n",
      "0.984054033702\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAi0AAAF5CAYAAABTHhoOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n1AgWLZIkqRH+H78ycNTEDce6AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdf52152750>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.ensemble import BaggingRegressor\n",
    "from sklearn.neural_network import MLPRegressor\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "bg_ann = BaggingRegressor(MLPRegressor(), n_estimators=20,\n",
    "                       n_jobs=6, max_samples=1.0)\n",
    "pl = Pipeline(steps=[(\"scaler\", StandardScaler()),\n",
    "                     (\"dim_reduction\", PCA()),\n",
    "                     (\"estimator\", bg_ann)])\n",
    "\n",
    "model = GridSearchCV(pl, dict(dim_reduction=[PCA(n_components=0.95), None],\n",
    "                              estimator__base_estimator__alpha=[0.01, 0.0001]))\n",
    "model.fit(X_train, y_train)\n",
    "print \"Bagging MLPRegressor - Best params:\"\n",
    "print model.best_params_\n",
    "print\n",
    "print \"Bagging MLPRegressor - Best model:\"\n",
    "print model\n",
    "print \n",
    "print \"Bagging MLPRegressor - Score on test set:\"\n",
    "# R-squared\n",
    "print model.score(X_test, y_test)\n",
    "\n",
    "y_pred = model.predict(X_test)\n",
    "plt.scatter(y_test, y_pred, s=1)\n",
    "plt.xlabel(\"y true\")\n",
    "plt.ylabel(\"y pred\")\n",
    "plt.xlim(0, 0.3)\n",
    "plt.ylim(0, 0.3)\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "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": 1
}