comparison_to_regression.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from __future__ import division\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from random_forest import RandomForest\n",
    "from gui import Gui\n",
    "import config\n",
    "import helpers\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "import copy\n",
    "import altair as alt\n",
    "#import statsmodels.api as sm\n",
    "import sklearn\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "\n",
    "from reducer import Reducer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "red_ruleset = []\n",
    "new_ruleset = []\n",
    "\n",
    "feature_names = config.feature_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "groundTruth = helpers.create_separate_files_status(\"data.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, x_test, y_train, y_test = train_test_split(helpers.create_dataframe(\"dataWoName.csv\"), groundTruth, test_size=0.33,\n",
    "                                                        random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='warn',\n",
       "          n_jobs=None, penalty='l2', random_state=None, solver='saga',\n",
       "          tol=0.0001, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log_res = LogisticRegression(solver = 'saga')\n",
    "log_res.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.00341324])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log_res.intercept_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 1.35228815e-02, -1.09826527e-03, -1.20926654e-02,\n",
       "         7.09394890e-05,  5.75885895e-07,  4.09226004e-05,\n",
       "         4.15938942e-05,  1.22779111e-04,  3.32524685e-04,\n",
       "         3.31086320e-03,  1.71468313e-04,  2.05862829e-04,\n",
       "         2.85993722e-04,  5.14402501e-04,  3.87500215e-04,\n",
       "         2.18551352e-02,  2.44532556e-03,  2.98973637e-03,\n",
       "        -1.31568235e-03,  1.80545866e-03,  1.26902386e-02,\n",
       "         2.19125194e-03]])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log_res.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7384615384615385"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log_res.score(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "def perform_several_logistic_regressions(nr_iterations):\n",
    "    res = []\n",
    "    for i in range(nr_iterations):\n",
    "        x_train, x_test, y_train, y_test = train_test_split(helpers.create_dataframe(\"dataWoName.csv\"), groundTruth, test_size=0.33,\n",
    "                                                        random_state=i)\n",
    "        log_res = LogisticRegression(penalty='l1', solver = 'saga')\n",
    "        log_res.fit(x_train, y_train)\n",
    "        score = log_res.score(x_test, y_test)\n",
    "        res.append(score)\n",
    "        \n",
    "    return res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "scores = perform_several_logistic_regressions(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7fb853c22c50>]"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7481538461538462"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.042671443273059294"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.std(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import svm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/tamara/anaconda3/envs/cdss_python2/lib/python2.7/site-packages/sklearn/svm/base.py:196: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
      "  \"avoid this warning.\", FutureWarning)\n",
      "/home/tamara/anaconda3/envs/cdss_python2/lib/python2.7/site-packages/sklearn/svm/base.py:931: ConvergenceWarning: Liblinear failed to converge, increase the number of iterations.\n",
      "  \"the number of iterations.\", ConvergenceWarning)\n"
     ]
    }
   ],
   "source": [
    "C = 1.0  # SVM regularization parameter\n",
    "svc = svm.SVC(kernel='linear', C=C).fit(x_train, y_train)\n",
    "rbf_svc = svm.SVC(kernel='rbf', gamma=0.7, C=C).fit(x_train, y_train)\n",
    "poly_svc = svm.SVC(kernel='poly', degree=3, C=C).fit(x_train, y_train)\n",
    "lin_svc = svm.LinearSVC(C=C).fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8461538461538461"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svc.score(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7384615384615385"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rbf_svc.score(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8153846153846154"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "poly_svc.score(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.26153846153846155"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lin_svc.score(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "C = 0.1  # SVM regularization parameter\n",
    "svc = svm.SVC(kernel='linear', C=C).fit(x_train, y_train)\n",
    "rbf_svc = svm.SVC(kernel='rbf', gamma=0.7, C=C).fit(x_train, y_train)\n",
    "poly_svc = svm.SVC(kernel='poly', degree=3, C=C).fit(x_train, y_train)\n",
    "lin_svc = svm.LinearSVC(C=C).fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8461538461538461\n",
      "0.7384615384615385\n",
      "0.8153846153846154\n",
      "0.7230769230769231\n"
     ]
    }
   ],
   "source": [
    "print(svc.score(x_test, y_test))\n",
    "print(rbf_svc.score(x_test, y_test))\n",
    "print(poly_svc.score(x_test, y_test))\n",
    "print(lin_svc.score(x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "def perform_several_svm(nr_iterations):\n",
    "    df = pd.DataFrame(columns=[\"svm\", \"score\"])\n",
    "    C = 0.1\n",
    "    for i in range(nr_iterations):\n",
    "        x_train, x_test, y_train, y_test = train_test_split(helpers.create_dataframe(\"dataWoName.csv\"), groundTruth, test_size=0.33,\n",
    "                                                        random_state=i)\n",
    "        svc = svm.SVC(kernel='linear', C=C).fit(x_train, y_train)\n",
    "        rbf_svc = svm.SVC(kernel='rbf', gamma=0.7, C=C).fit(x_train, y_train)\n",
    "        poly_svc = svm.SVC(kernel='poly', degree=3, C=C).fit(x_train, y_train)\n",
    "        lin_svc = svm.LinearSVC(C=C).fit(x_train, y_train)\n",
    "        \n",
    "        new_row = {'svm':'svc', 'score':svc.score(x_test, y_test)}\n",
    "        df = df.append(new_row, ignore_index=True)\n",
    "        \n",
    "        new_row = {'svm':'rbf_svc', 'score':rbf_svc.score(x_test, y_test)}\n",
    "        df = df.append(new_row, ignore_index=True)\n",
    "        \n",
    "        new_row = {'svm':'poly_svc', 'score':poly_svc.score(x_test, y_test)}\n",
    "        df = df.append(new_row, ignore_index=True)\n",
    "\n",
    "        new_row = {'svm':'lin_svc', 'score':lin_svc.score(x_test, y_test)}\n",
    "        df = df.append(new_row, ignore_index=True)\n",
    "\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-44-054f35758365>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mdf_svms\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mperform_several_svm\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mdf_svms\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto_csv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"/home/tamara/Documents/PECLIDES/df_svms.csv\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-34-4671721bd7e4>\u001b[0m in \u001b[0;36mperform_several_svm\u001b[0;34m(nr_iterations)\u001b[0m\n\u001b[1;32m      7\u001b[0m         \u001b[0msvc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msvm\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSVC\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkernel\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'linear'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mC\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mC\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m         \u001b[0mrbf_svc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msvm\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSVC\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkernel\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'rbf'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mgamma\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0.7\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mC\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mC\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 9\u001b[0;31m         \u001b[0mpoly_svc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msvm\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSVC\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkernel\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'poly'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdegree\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mC\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mC\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     10\u001b[0m         \u001b[0mlin_svc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msvm\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mLinearSVC\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mC\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mC\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/tamara/anaconda3/envs/cdss_python2/lib/python2.7/site-packages/sklearn/svm/base.pyc\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, sample_weight)\u001b[0m\n\u001b[1;32m    210\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    211\u001b[0m         \u001b[0mseed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrnd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0miinfo\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'i'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmax\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 212\u001b[0;31m         \u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msolver_type\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkernel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrandom_seed\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mseed\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    213\u001b[0m         \u001b[0;31m# see comment on the other call to np.iinfo in this file\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    214\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/tamara/anaconda3/envs/cdss_python2/lib/python2.7/site-packages/sklearn/svm/base.pyc\u001b[0m in \u001b[0;36m_dense_fit\u001b[0;34m(self, X, y, sample_weight, solver_type, kernel, random_seed)\u001b[0m\n\u001b[1;32m    269\u001b[0m                 \u001b[0mcache_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcache_size\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcoef0\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcoef0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    270\u001b[0m                 \u001b[0mgamma\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_gamma\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepsilon\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mepsilon\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 271\u001b[0;31m                 max_iter=self.max_iter, random_seed=random_seed)\n\u001b[0m\u001b[1;32m    272\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    273\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_warn_from_fit_status\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "df_svms = perform_several_svm(100)\n",
    "df_svms.to_csv(\"/home/tamara/Documents/PECLIDES/df_svms.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_svc = np.mean(df_svms[df_svms.svm == \"svc\"].score)\n",
    "mean_rbf_svc = np.mean(df_svms[df_svms.svm == \"rbf_svc\"].score)\n",
    "mean_poly_svc = np.mean(df_svms[df_svms.svm == \"poly_svc\"].score)\n",
    "mean_lin_svc = np.mean(df_svms[df_svms.svm == \"lin_svc\"].score)\n",
    "\n",
    "std_svc = np.std(df_svms[df_svms.svm == \"svc\"].score)\n",
    "std_rbf_svc = np.std(df_svms[df_svms.svm == \"rbf_svc\"].score)\n",
    "std_poly_svc = np.std(df_svms[df_svms.svm == \"poly_svc\"].score)\n",
    "std_lin_svc = np.std(df_svms[df_svms.svm == \"lin_svc\"].score)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(10, 5))\n",
    "ax.bar([0,1,2,3,4,5,6], [mean_gini, mean_entropy, np.mean(scores), mean_svc, mean_rbf_svc, mean_poly_svc, mean_lin_svc], yerr=[std_gini, std_entropy, np.std(scores), std_svc, std_rbf_svc, std_poly_svc, std_lin_svc], align='center', alpha=0.5, ecolor='black', capsize=10)\n",
    "ax.set_ylabel('accuracy on test set', fontsize=15)\n",
    "ax.set_xticks([0,1,2,3,4,5,6])\n",
    "ax.set_xticklabels([\"rf gini\", \"rf entropy\", \"log reg\", \"svc\", \"rbf svc\", \"poly svc\", \"lin svc\"], fontsize=15)\n",
    "ax.set_title('Scores of different models', fontsize=20)\n",
    "ax.yaxis.grid(True)\n",
    "\n",
    "# Save the figure and show\n",
    "plt.tight_layout()\n",
    "plt.savefig(\"/home/tamara/Documents/PECLIDES/pics/scores_different_models_100_1.svg\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsgAAAFgCAYAAACmDI9oAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi40LCBodHRwOi8vbWF0cGxvdGxpYi5vcmcv7US4rQAAIABJREFUeJzt3XmYJEWZ+PHvCwqoCIjAcDOosILggQOsrD8ZldMLT24FUdBVkJX1wBWhwVsW2RVxFRABFTkVEEYOwVa8cEBF5XRAjuG+QWEGgff3R2RDUlPdnT1dXVXT/f08Tz3dmRGZGZWRWfVWZGRkZCaSJEmSisV6XQBJkiSpnxggS5IkSTUGyJIkSVKNAbIkSZJUY4AsSZIk1RggS5IkSTUGyJLUQEQsExFfi4gbIuKxiMiIePk41ndctY7ptXnTq3nHtcm/TkT8KCJur/LcX0tbOSKOj4i5EfF4lb7cwpZNnTdS3Y5xPTOr9Qx0pmSS2nlGrwsgqbMiYnFgD2BXYEPgucB9wO3A74CzMvOs3pVwkfUV4APA2cB3gccp+3TCVXV6BvCiattzgXm1LMcBWwE/AOYA2ZK+SKmCv4OA12bmYG9LI2kqMkCWJpEqkDob2Aa4HziHEkwtAbwE2Bl4MWCAPHZvAq7NzDdP4DZuAdYDHmiZvzawPnB0Zu5VT4iIJYAtgZ9m5i4TWDZJmjIMkKXJZSdKcHw5sHlmPi3QiohnA5v2omCTwKrALyZyA5n5T+DqYbYNcGubtJUp3eXapUmSFoJ9kKXJZbPq73GtwTFAZj6cmT9rt2BE7BARF0bEvRExr+pr+4OImNGSb8mI2D8i/hwRD0fEgxFxcURs32adT/a7jIh1I+LkiLgzIp6IiJm1fMtHxBcj4qqIeCQiHqjKslWbdS4RER+JiN9HxH1VGW6IiDMjYoumOyoiVomII6tlH42IuyLihxHxypZ8gxGRQACbV+8nI2Kw4Xa2qPbPP6p9e0ZEvHiYvAv0U622/fNq8qDa9gci4gbgxiptt1racS3r3SkifhYR91d1e1VEHBARS7YpQ1bveeWIOCYibqn6Ne9ey/PsiPhURPyxel9/j4jfRMRObdY3s1bel0fEOVU5Ho6In0fEZi35b6B0rwD4We095Yg7esFtzYiIc6tj6b6IOD0i1qjyvSAiTqrq/JFq37xsmHU2Ok5q+Z8bEV+N0h98XkRcHRH7McL37Vj25wjreEFEHBURc6r3dG+Uc/SbEfH8puuRVNiCLE0u91R/1226QEQE8B1gN+Bu4IfAXcDqwGuBa4BLq7xLAOcBm1NaOo8Eng28Ezg5Il6emf/VZjMvBC4BrgW+DzwLeLBa51rAIDAduBg4F3gOpUvDuRHxgcw8urau4ygt5X8BTgAeobSwvprSev7TBu95beCX1XIXUfrurgG8C3hjRLwjM8+ubW+QErTdWE0D3NBgO+8ETgYerf7eVpXzN8CfRlu+cjBl3+xGCZQHq/mDlG4004F9KVcNzqjS/lgrw7HAeyldbU6vlvlX4LPA6yNiy8x8rGWbywO/Bf5OOR6eAO6o1rccZZ+9Avg9cCwl+NsaODEiXpKZB7R5HzOAT1Tv/RhgTeAdwIXVcXNNle9/gLdSjrHjabCf29gY+CRlfx1N6Yv/dmCDiNiOUvdXU46ftaq0CyLiBZn596GVjPE4ofrBcWG1/cspx/pywGeq97OAcezP+jpWAWYDywCzKPW8FKVrzruBr/PUZ4OkJjLTly9fk+RF+ZJ9lBLQfJfyxb/WKMvsRbmp63fAsi1piwOr1KY/VeWdBTyjNn8lSiCTwGa1+dOreQl8YZjtD1bl3bFl/nKUQO8RYFo1b9kq76XA4m3W9fyG++m8qkyfbpm/GfAYJZhYuiUtgcEx1MXS1Xr+CcxoSTu8tl+mt9lfx7Xkn1nNH2iznbbLVGm7V2k/BJ7VkjZQpe3b5n0mJXh8Rpt1Hlelf6Jl/lKUHzdPAC9vU/YEdm9Z5gPV/G8MU7aZYzz+69vapSXt29X8e9vU+2eG2RdjOk6A/6rynw4sVpu/drXddnW7sPtzoDZvn3blr9Ke01r3vnz5Gv1lFwtpEsnMP1BGr7ij+ns6cENE3BNliLB2N5jtU/39QLZ0y8jMxzPzttqsPShfxPtlrdUxM++ktEgCvL/NNu6gtIQ+TXVZe3Pg9Mw8qWXb91NabZeitDRSbTuA+ZTAgZZlRm0li4jVKSM+3EQZmaK+/K8prYTLU35cjMd21XpOzMxLW9IGWPBGvImwLyWQ2yMzH2lJ+ywlwGt3Y9+jwMeypWW5ulS/K3BpZrbuu3mUVtug3Aza6leZeVzLvGOr8m3S6N0098vM/H7LvOOrvw8AX2pJO6H6++SwfQt5nLyXclx+IjOfqOX/G/C11kKOc3+201rHZOY/2tS9pFHYxUKaZDLzlIj4EaV7xKsprcqvply2fmtEnEBpycuIeA6wAXBHFVwPKyKeSxlm7JbMbHcj2UXV31e0Sbs8M+e3mf+q6u+y0X5c1xWrv+tV7+3BiPgx8GbgjxFxOqVbxiWZ+fBI5a8ZKt/FWW6Ka3URJWh5BU8FTgtjo+rvz1sTMvOBiPgjw1x274QoN2S+jNJt5j9KT5oFzKfaty1uqH70tNqYclVhuHF4n1n9bbfO1h8JZOY/I+IO4HntCjcOC2yLp25i/GNmPt6Sdkv1d/XavDEdJ7Xz4+bMvK5N/kGe6ls9ZDz7s+4s4AvAkRGxNaXl+1fAlZk5at9tSQsyQJYmoeoL/fzqNTT82zsoLXbvAX5E6a869DCJW9qsptWy1d/bhkkfmt/uARXDjRc8dPPQltVrOEvX/t+B0rK2M0+1Ss+LiNMorZ53jLAeGN/7GIuh7QxXnokeQ/l5lNbHFVkwMBvNaPW1cfUaztJt5t3fZh6UFuTFG5arqXat848Nl5aZj1U/IJ5Zmz3W42Rh6ns8+/NJmXljRGxCuTKxDU+1at8cEf+dmQu0XksamV0spCmg6ipxCqXvK8Drqr9DQctqDVYzFFisPEz6Ki35nlaEUda5b2bGCK/3PrmizEcycyAz16Xc6LUr5UaqXYHTJvh9jMXQ8tOGSR9u+50ytP0/jLJv2zUtj1Zfh4+yztd2/u103ViPk4Wp747tz8y8KjN3oATdM4D9Kd/x/xsR7xtteUlPZ4AsTS0PVX8DSv9EymgQ0yKiXdeIJ2XmQ8B1wGoRsU6bLENf4r8fQ3l+W/39f2NYpl6mm6u+pltTniD36gZDWg11JXl1RLS7irYw76OdoeUX6EYREctS6+86EbKMxnAF8JKIWL5Dq/0dpY/tQtXXGAx1geh0y/JYjOk4qc6POZTz44Vt8s9sM6/j+zMzH8vMyzLzy5TRXqB0r5I0BgbI0iQSZbzbLSNigXM7IlYG9qwm6w+8GLr8+q0qcKsvs1g1hNSQYynB9aFVt42hfCtQRgIYytNIdfPaxcDbI2KPYd7ThhGxUvX/ihGxYZtsz6Fchn6McoPZSNucC1xAGf3hP1q2tSml68Z9lG4o43FmtZ6do2Usacql8GUXWKLzvkp5iuKx1XBiTxMRz4uIjRZcrL2qX/L3gRkR8Zn6MVBb5wur4dHGY+hmyzXHuZ6FtpDHyXco36tfrp+D1f74SJttdGR/RsQrW8/dylBrdtP++ZIq9kGWJpdNKSMX3B4RvwT+Vs1fG3gjZfzhM3l6V4RjKC1Y7wb+GhFnUsZBXpXSFeNYSkAH8N/AtpQRGi6PiFmUcZDfRRnq7SuZ+csxlnlnyg1P346Ij1DGS76fcsPUSyk3Eb4KuJPSFeQPEfFnyjjCN1PGfn0T5RL216qWvNF8kHIT06FRHkZyKU+Nb/sE8N6G6xlWZv49IvaijH98cUTUx0HegPIj5TXj2UaDMhxbPdDiQ8B1EXEeZVSG5SnHxGsoQd0Hx7DavYF1gEOAd1fH2R2U42U9Sl/anXjq2FsYP6PUwxcjYgNKIEpmfm4c61wYYz1ODqO01r4D+H21v5cDtqfU91vabKMT+/PdwAeqZa+j7K8XUm5mnU8ZW1rSGBggS5PLYcBfgS0oweXWlGHS7qHcRX8iZdixJ/uYVv+/p/oy34vyZb4kJZi7mHKH/FDeRyNiS2A/SmC7D6XV9nLgPzLzB2MtcGbOrYK4fSiBxS6US+u3A1cCRwB/rrLfQLnhbCblEvcKlPFlr6H0uXzaUHEjbPP6qlX3AOAN1foepIw7+/nMnD3W9zHMdk6LiG2qMm9PCVZ+QQn492eCA+SqDB+OiJ9Qgr0tKAHbvZRA+VDge2Nc34MRsTnlWNmZUmdLUYK6vwIfpbS8jqfMV0XEbsDHKMH9UlVSVwPksR4nmTk/ytMcByg3k+5LOWY/R2lpXiBA7tD+/AHlnN0MeCXlh/AtlPPhsMz8y9jeuaRwBBhJkiTpKfZBliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqpkUo1issMIKOX369F4XQ5IkSX3ssssuuzszVxwt36QIkKdPn86ll17a62JIkiSpj0XEjU3y2cVCkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgB5ETUwMEBEdOw1MDDQ67ckSZLUFwyQF1EDAwNk5oivzTffnM0333zUfJlpgNwh/nCRJGnR94xeF0CaTAYGBkYNamfOnAnA4ODghJdHkiSNnS3IkiRJUo0BsiRJklRjgCxJkiTVGCBLkiRJNQbIkiRJUo0BsiRJklRjgCxJkiTVGCBLkiRJNT4oZBwOv+DaXhdhRHPvewTo/3J+dMt1e10ESZKkJ9mCLEmSJNXYgqxJp99bzG3Zl54yMDDAwQcf3LH1HXTQQaM+7l2SRmOALEnqmYGBgVED2pkzZwIwODg44eWRJLCLhSRJkvQ0BsiSJElSjQGyJEmSVGOALEmSJNUYIEuSJEk1BsiSJElSjQGyJEmSVOM4yIuoc084gvO/9/VGeffb6l9GzbPVrnuzzXv2GW+xJEmSFnkGyIuobd6zjwGtpI7o96c6+vRJSd1mgCx1kC37kiQt+gyQpQ6yZV+SpEWfN+lJkiQtAgYGBoiIjr0GBgZ6/Zb6li3IkiRJi4CBgYFRg9qZM2cCMDg4OOHlmcxsQZYkSZJqDJAlSZKkGgNkSZIkqcYAWZIkSaoxQJYkSZJqDJAlSZKkGod5kyT1jE+flNSPDJAlST3j0ycl9aOud7GIiG0i4pqImBMR+7dJXzMifhYRf4iIP0XEG7pdRkmSJE1dXQ2QI2Jx4EhgW2B9YKeIWL8l2wHAKZn5CmBH4BvdLKMkSZKmtm63IG8CzMnM6zPzUeAkYLuWPAksU/2/LHBrF8snSZKkKa7bfZBXA26uTc8FNm3JMwCcHxH7AM8Btmi3oojYC9gLYNq0aT155vhq8+Z3fZuT0eBgZ38DWS+d0el6Uf/ynOkMzxn1g/vvvx+gJ3HRZNKPN+ntBByXmYdFxKuA70bEBpn5RD1TZh4FHAUwY8aMnDlzZtcLevgF13Z9m5PR9jPX7ej6rJfO6HS9qH95znSG54z6wXLLLQdAL+KiyaTbXSxuAdaoTa9ezat7H3AKQGb+BlgKWKErpZMkSdKU1+0AeTawTkSsHRFLUG7CO6slz03A6wEiYj1KgHxXV0spaVIZGBggIjr2GhgY6PVbkiaU54ymuq52scjMxyJib+A8YHHg2My8IiIOAS7NzLOA/wSOjoiPUm7Y2z0zs5vllDS5DAwMjPoFPXQ50n57kufMaPq5W9Lc+x4B+ruMAB/dsr+7JHW9D3JmzgJmtcw7sPb/lcC/dbtckiRJEvTgQSGSJElSPzNAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSppusPCpE0NfX7U518+pT6Tb8fi54zmsxsQZYkSZJqDJAlSZKkGgNkSZIkqcYAWZIkSaoxQJYkSZJqGo1iEREHAsdk5q1t0lYB9szMQzpdOEmSJBXnnnAE53/v643y7rfVv4yaZ6td92ab9+wz3mJNSk2HeTsIOBdYIEAGVq3SDZAlSZImyDbv2ceAtkuadrEIIIdJWx24rzPFkSRJknpr2BbkiNgN2K2aTOD/IuLBlmxLARsC509M8SRJkqTuGqmLxcPAPdX/ATwA3NuS51HgJ8A3Ol80SZIkqfuGDZAz81TgVICI+A7w2cy8vlsFkyRJveHNYJrqGt2kl5nvBYiIoPQ5XgO4PDP/MYFlk6SO8MteGhtvBtNU13QUCyLiQ8ABwMqUPskbA7+PiB8Cv8jM/5mYIkrS+PhlL0kai0ajWETEx4GvAkcDr6P0SR4yCOzQ8ZJJkiRJPdC0BfnDwIGZ+ZWIWLwl7Rpg3c4WS5IkSeqNpuMgrwxcNkzaE5Th3iRJkqRFXtMAeQ6w+TBprwGu7ExxJEmSpN5q2sXif4BvRMSjwGnVvJUi4n3AfsCeE1E4SZIkqduaDvN2TEQ8DzgQOLiaPYvyMJGBzDxxgsonSZIkdVXjYd4y89CI+CbwKmAFylP1fpOZD0xU4SRJkqRuaxwgA2TmQ8D5E1QWSZIkqeeajoP8jqq/8dD02hHx64i4PyJOj4jlJq6IkiRJUvc0HcXiAGCZ2vQRlG4WXwI2Aj7f4XJJkiRJPdG0i8ULgD8DRMSywFbA2zLznIi4iRIof3hiiihJkiR1T9MWZICs/m4OPA78tJqeC6zYyUJJkiRJvdI0QL4c2CUingO8H/hZZs6v0tYE7pyIwkmSJEnd1rSLxX8BPwZ2A/4ObFlLeytwSYfLJUmSJPVE0weF/DIi1gTWBa7LzPtrycdSHkUtSZIkLfLG8qCQh4DL2syf1dESSZIkST00lpv0JEmSpEnPAFmSJEmqMUCWJEmSagyQJUmSpJpGAXJEXBQRLx4mbd2IuKizxZIkSZJ6o2kL8kxgmWHSlgFe05HSSJIkST22MI+aflJELAG8Dri9YyWSJEmSemjYcZAj4iDgwGoygd9GxHDZD+1wuSRJkqSeGOlBIbOAu4EAvgYcBtzQkudR4OrMvHhCSidJkiR12bABcmbOBmYDRMRDwDmZeXe3CiZJkiT1QtNHTX8fWLw+IyK2AtYHfpGZv+90wSRJkqReaBognww8AOwBEBEfAf4HmA8sHhFvz8yzJ6aIkiRJUvc0HcXiXyl9kod8HDgsM58FHAN8utMFkyRJknqhaYD8fKqh3CJiQ2BV4JtV2qmUrhaSJEnSIq9pgHwHML36fxvgxsy8rpp+FvBE0w1GxDYRcU1EzImI/YfJs31EXBkRV0TEiU3XLUmSJI1X0z7IpwJfjoiXAe8Fvl5LewXw1yYriYjFgSOBLYG5wOyIOCszr6zlWQf4FPBvmXlfRKzUsIySJEnSuDUNkPcHHgQ2Bv4P+GIt7ZWUm/ia2ASYk5nXA0TEScB2wJW1PHsCR2bmfQCZeWfDdUuSJEnj1ihAzszHgEOGSXv7GLa3GnBzbXousGlLnnUBIuJXlKHlBjLz3NYVRcRewF4A06ZNY3BwcAzF6IzV5s3v+jYno8HBWzu6PuulM6yX/tTpegHrplM8Z/qT50x/moh66aSmLcgARMS2wAxgDeBzmXlTRLyG0ircqXf6DGAdYCawOvCLiNgwM++vZ8rMo4CjAGbMmJEzZ87s0OabO/yCa7u+zclo+5nrdnR91ktnWC/9qdP1AtZNp3jO9CfPmf40EfXSSY0C5IiYBpxF6U5xA7A2ZRSLmyh9kucB/95gVbdQgushq1fz6uYCl2TmP4G/RcS1lIB5dpOySpIkSePRdBSLI4ClgRdXr6il/RR4fcP1zAbWiYi1I2IJYEdK4F13BqX1mIhYgdLl4vqG65ckSZLGpWkXi22A3TJzTjUSRd1cSt/iUWXmYxGxN3AepX/xsZl5RUQcAlyamWdVaVtFxJXA48DHM/OehuWUJEmSxmUsfZAfG2b+CsAjTVeSmbN4+lP5yMwDa/8nsF/1kiRJkrqqaReLi4GPtLQeZ/V3D+CijpZKkiRJ6pGmLcifBH4J/AX4ESU43jMiXgJsCPzrxBRPkiRJ6q5GLciZ+RfK8G6XArtT+ga/nWoc48x0vBNJkiRNCo37IGfmHODdE1gWSZIkqecatSBHxEUR8eJh0taNCPsgS5IkaVJoepPeTGCZYdKWAV7TkdJIkiRJPdY0QIanRq14UvWwj9cBt3esRJIkSVIPDdsHOSIOAobGJ07gtxExXPZDO1wuSZIkqSdGuklvFnA35bHSXwMOA25oyfMocHVmXjwhpZMkSZK6bNgAOTNnA7MBIuIh4Gwf+SxJkqTJrtEwb5l5/EQXRJIkSeoHY7lJT5IkSZr0DJAlSZKkGgNkSZIkqcYAWZIkSappdJPekIhYElgNWKo1LTOv7FShJEmSpF5pFCBHxKrAUcC27ZIpDxJZvIPlkiRJknqiaQvyMcBGwH7AlZQHhEiSJEmTTtMA+d+APTPzlIksjCRJktRrTW/SuxN4ZCILIkmSJPWDpgHygcAnI2KZiSyMJEmS1GtNu1i8HVgTuDEiZgP3t6RnZu7Q0ZJJkiRJPdA0QF4BuK76/5nAihNTHEmSJKm3GgXImfnaiS6IJEmS1A8W6kl6EfHMThdEkiRJ6geNA+SI2CwifhIRDwHzIuKhiJgVEa+awPJJkiRJXdX0SXpbAucA1wCHAncA04B3AoMR8cbM/OmElVKSJEnqkqY36X0eOAt4V2Zmbf4hEXE68AXAAFmSJEmLvKZdLDYEjm4JjoccVaVLkiRJi7ymAfL9wAuHSXshC46LLEmSJC2SmgbIpwJfjIhdI2IpgIhYKiJ2pXSvOGWiCihJkiR1U9M+yJ8Eng8cDxwfEX8Hlq7SflClS5IkSYu8pg8KeQTYJSI+C2wMrALcBszOzKsnsHySJElSVzVtQQagCoYNiCVJkjRpLdST9CRJkqTJygBZkiRJqjFAliRJkmoMkCVJkqSaRgFyRLw5IgymJUmSNOk1DXrPAOZGxJcjYr2JLJAkSZLUS00D5BcCRwPbA3+JiN9ExJ4RsczEFU2SJEnqvkYBcmbekJkHZebawJbAHOBw4LaI+G5EvHYiCylJkiR1y5j7FWfmRZn5bmBd4DJgF+CnEXF9RHw0Isb08BFJkiSpn4w5QI6IzSPiOOAaYAPgSGAr4DTgYOCEThZQkiRJ6qZGrb0RsRawW/WaDgwCewE/zMz5VbYLI+I3wPc6X0xJkiSpO5p2h7geuBU4Djg2M/82TL4rgN91oFySJElSTzQNkN8EnJeZT4yUKTOvBbxhT5IkSYuspn2QfwlMa5cQEatExNKdK5IkSZLUO01bkL8NPADs2SZtAFgW2LFDZZIkSZJ6pmkL8muAc4ZJm1WlS5IkSYu8pgHyssDDw6TNA57XmeJIkiRJvdU0QP4r8MZh0t4AXNd0gxGxTURcExFzImL/EfK9IyIyImY0XbckSZI0Xk37IB8BfDMiHqUM9XYbsAplXOQPA//eZCURsTjlwSJbAnOB2RFxVmZe2ZLvucC+wCUNyydJkiR1RKMAOTOPjohpwKeA/WpJ84ADMvPohtvbBJiTmdcDRMRJwHbAlS35Pgt8Gfh4w/VKkiRJHdH4UdOZ+TlgVUpXi/dUf1fNzC+NYXurATfXpudW854UERsBa2TmcDcFSpIkSROmaRcLADLzAeDcCSoLEbEY8FVg9wZ596I87ppp06YxODg4UcUa1mrz5o+eSaMaHLy1o+uzXjrDeulPna4XsG46xXOmP3nO9KeJqJdOGlOAHBGvBtYFlmpNy8xvNFjFLcAatenVq3lDngtsAAxGBMDKwFkR8ZbMvLRle0cBRwHMmDEjZ86c2fyNdMjhF1zb9W1ORtvPXLej67NeOsN66U+drhewbjrFc6Y/ec70p4mol05qFCBX/Y8vBNYHEogqKWvZmgTIs4F1ImJtSmC8I7DzUGLVQr1CbbuDwMdag2NJkiRpojTtg3wY5Ul6a1CC402B6cBnKEPANfoZkJmPAXsD5wFXAadk5hURcUhEvGVsRZckSZI6r2kXi80pw67dVk1HZt4EfKHqN/wNYOsmK8rMWZSn79XnHThM3pkNyydJkiR1RNMW5OWAuzLzCeBBYKVa2q+BzTpdMEmSJKkXmgbIf6M8GATgCmCXWtqbgXs7WShJkiSpV5p2sZgFbAWcAnwOODMi5gL/BNYEPjkxxZMkSZK6q+mT9Pav/f+TiNgMeBvwLOCCzPzJBJVPkiRJ6qpRA+SIWBL4GHB2Zl4OUA275tBrkiRJmnRG7YOcmfOBT1Nu1JMkSZImtaY36V0CbDSRBZEkSZL6QdOb9D4BnBgR/6TcsHcHT3+KHpn5cIfLJkmSJHVd0wD5kurv14D/HSbP4uMvjiRJktRbTQPkPWhpMZYkSZImo6bDvB03weWQJEmS+kLTm/QkSZKkKaFRC3JE3MUoXSwyc6WOlEiSJEnqoaZ9kI9kwQD5ecDrgWWAYztZKEmSJKlXmvZBHmg3PyICOAX4ZwfLJEmSJPXMuPogZ2YCxwB7d6Y4kiRJUm914ia9FwBLdGA9kiRJUs81vUnvQ21mLwGsB+wCnNrJQkmSJEm90vQmva+3mTcfmAt8Azi4YyWSJEmSeqjpTXqOlyxJkqQpwcBXkiRJqmkUIEfE5yPiW8OkfTMiPtvZYkmSJEm90bQFeSfg4mHSLgZ27kxxJEmSpN5qGiCvCtwyTNqtVbokSZK0yGsaIN8ObDRM2kbAXZ0pjiRJktRbTQPkU4ADI+KN9ZkR8QbgM8BJnS6YJEmS1AtNx0E+EHg58OOIuAe4DVgFWB44nxIkS5IkSYu8puMgzwO2ioitgdcCzwfuAS7MzAsmsHySJElSVzVtQQYgM88DzpugskiSJEk913Qc5B0j4uPDpH0sIrbvbLEkSZKk3mh6k97+wLxh0h4GPtWZ4kiSJEm91TRAXgf4yzBpV1XpkiRJ0iKvaYD8MLD6MGlrAPM7UxxJkiSpt5oGyD8FPhMRK9VnRsSKwKcpQ71JkiRJi7ymo1h8EvgtcF1EnMtT4yBvDdwPfGJiiidJkiR1V6MW5My8CXgZ8HWs2fhIAAASXElEQVRKl4ptq79HABtl5s0TVkJJkiSpixqPg5yZd+FoFZIkSZrkmvZBliRJkqaExi3IEbEDsCewLrBUa3pmrrTAQpIkSdIipumT9HYGjgfmUIZ7Ows4u1r+QUrfZEmSJGmR17SLxceBzwIfrqa/kZl7AGsDd1PGSZYkSZIWeWN5kt6vMvNx4HFgGYDMfAj4MrD3xBRPkiRJ6q6mAfKDwJLV/7cA69XSAnh+JwslSZIk9UrTm/RmAy8FzqP0Pz4wIh4DHgUOpDxERJIkSVrkNQ2QvwisVf1/YPX//1FaoGcDH+h80SRJkqTuaxQgZ+ZvqVqJM/N+YLuIWBJYMjMfnMDySZIkSV3VeBzkVpk5H5jfwbJIkiRJPeeT9CRJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqSargfIEbFNRFwTEXMiYv826ftFxJUR8aeIuDAi1mq3HkmSJGkidDVAjojFgSOBbYH1gZ0iYv2WbH8AZmTmS4HTgK90s4ySJEma2rrdgrwJMCczr8/MR4GTgO3qGTLzZ5n5cDX5W2D1LpdRkiRJU9hCPyhkIa0G3FybngtsOkL+9wE/aZcQEXsBewFMmzaNwcHBDhWxudXm+ZyUThgcvLWj67NeOsN66U+drhewbjrFc6Y/ec70p4mol07qdoDcWETsCswANm+XnplHAUcBzJgxI2fOnNm9wlUOv+Darm9zMtp+5rodXZ/10hnWS3/qdL2AddMpnjP9yXOmP01EvXRStwPkW4A1atOrV/OeJiK2AD4NbF490lqSJEnqim73QZ4NrBMRa0fEEsCOwFn1DBHxCuBbwFsy884ul0+SJElTXFcD5Mx8DNgbOA+4CjglM6+IiEMi4i1VtkOBpYFTI+KPEXHWMKuTJEmSOq7rfZAzcxYwq2XegbX/t+h2mSRJkqQhPklPkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSpxgBZkiRJqjFAliRJkmoMkCVJkqQaA2RJkiSppusBckRsExHXRMSciNi/TfqSEXFylX5JREzvdhklSZI0dXU1QI6IxYEjgW2B9YGdImL9lmzvA+7LzBcBhwNf7mYZJUmSNLV1uwV5E2BOZl6fmY8CJwHbteTZDji++v804PUREV0soyRJkqawyMzubSzincA2mfn+avrdwKaZuXctz1+qPHOr6euqPHe3rGsvYK9q8l+Aa7rwFhZFKwB3j5pL3Wa99CfrpX9ZN/3JeulP1svw1srMFUfL9IxulGQiZOZRwFG9Lke/i4hLM3NGr8uhp7Ne+pP10r+sm/5kvfQn62X8ut3F4hZgjdr06tW8tnki4hnAssA9XSmdJEmSprxuB8izgXUiYu2IWALYETirJc9ZwG7V/+8ELspu9gORJEnSlNbVLhaZ+VhE7A2cBywOHJuZV0TEIcClmXkW8G3guxExB7iXEkRr4dkNpT9ZL/3Jeulf1k1/sl76k/UyTl29SU+SJEnqdz5JT5IkSaoxQJYkSZJqDJAXERGxXURcFRGPRsQN41zXcRFx6UIsNxgRp41n2/2mk/u14fY2iYiBid7OZLSwx62kBUXE7hGREbH0KPkWi4gjI+KOKv9Al4qoNibj93C/WmTHQZ5Kqkd0nwD8BNgT+Mc4V/lZ4FkLsdyHgH+Oc9t9YwL2axObAAcBA13YliSN19spn/3vA64E5va2OFJ3GCD3sSqAWxxYCVgGODEzfzne9WbmdQu53JXj3XY/mKj92knV49WXzMx5vS6Lioh4JvBEZj7e67JIC6v2+dfUi4H7MvPYCSqS1JfsYtFHhi4hR8RbI+IKYB7wSeDmKsuZo13iioiXRsSvI2JeRFwREW+o1nlc63Zq00OX2jaMiAsi4h8RcXVEvL1l3YvkpZ0O7dflI+Ko6jLjvGofb9qSJyNi34j4QkTcFRF3Vpcml6zSdweOqOXNiBispgci4u6IeHVEzK7K+K4qbe2IOCMiHoyIhyLixxHxojbb3i8i/jci7o2I+yPiiGq88aHyz6vKUF8uIuL6iDh8oXZuj0XEyyPiwoh4OCLui4jvR8S0ljxrRsRPIuKRiPhbdbyfNrTvR1j3YJVvryiPvJ8HrFqlbRAR51T18VBEnBoRK7csP+q5ONVFxEsi4tzqmP1HlO5OH67Oh9sjYrGW/G+sjvUX1ebtGRF/rvbzHVWdLdv9d9Ofhvn8q392rRcRF1fnx7UR8bbasoOUK47Pq31mTR9mO23rskqb0vXZUgdXV+/tlxGxfku+Z0fE16p9NS8iZkfEViOsd/1q/81smb90RPw9IvYdYdm3RMRlVV3dFxGXRMTmVdpgRJzaZplDI+KmiIhq+lkR8ZWIuDEi5lefr18c4+7pWwbI/Wc68BXgi8C2wHGUS1wAHwNeBRzTbsGIeDZljOlnATsBnwMOB9ZsuO0TKQ9qeRvwV+CkiFh9Id5DP5rOwu/XJYGfAlsAHwfeCtwF/LQ1KAL+kxJE7QocCnwAGPqQOgc4rPr/VdXrQ7Vlnw0cX5VjG+B31bYvBNajdAPZHVgb+HlELN9m26sDu1Dqfi/g8wCZeS/wo2r5upnV+ha51qGIWBEYpOy3nYF9gM2BC+KpHwZBOabXA/YA9gM+wtMDhJH8G/DvlB9UbwYeqL7MfwUsRann3YGXAD+ufXGM91ycKn4MPE7Zj2+h/IB8LnAyMI1Sn3U7AJdl5hyAiDgA+Bbwc8p5+e/AA8CI/WqnoOk8/fPvb7W0k4EzKZ+HfwZOjYiXVWkfojyb4AGe+sy6bZhtDFeXQ9uY6vW5FvBVyg+OnSlPCT4vIpaq5TkaeC/lc/ttlEaccyLi1e1WWF3V/S0Lfq6/C3gm8L12y0XEC4HTgIson2u7AGcDQ98pJwNviIjn1JYJYHvglMzMavpMSh0dCbyB0n1whVH2w6IjM331yYsStCXw8pb506v5bxpl+Q8DjwKr1eZtUi17XMt2Lq1N717l2aM27/nAY8AHa/MGgdN6vZ96sF/fV+3XdWrzngFcBxxam5fAL1qWPQP4bW1673LaLbCNgWr57Vrmf7CqhxfU5q1eledTLdu+GlisNu/TwMPA8tX0FsATLes6oX4s9POrzXH7JeB+YJnavE2rfbFTNf3GanrjWp7VKH3pB0fZ3iDwCDCtZf53gWuAJWrz1qEEB2+sphudi1P5RfkiTWDDYdIvB75Zm16SEix9rJperjq+v9rr99LPrxE+/3av5v9Xbd5i1efISbV5A8Dd46nLqV6ftTrYrDZvLWrfsZQf8U8Au7XUx1+A82rzBql9DwPvB/4OLF2b9wtG+K6mPKX4nhHSV6zKtmNt3quq9zCjmt66mn5Lr/fvRL1sQe4/t2TmHxdy2Y0pv8ZvGZqRmb8D7mi4/Pm15e4B7qQEY5PBePbrFsBlwN8i4hkRMdR3/+fAjJa857dMX0nzfZiUGwbrNgF+n5nXP5kpcy6lBbO1VeHMzHyiNv1DSgvmBtX0hcCNVI9yj4jnAu8AvtOwfP1mE+D8zHxwaEZmXgLcwFP7ZmPg9sycXctzC6U+m7gsM1vPny0orfFP1I6Hv1XbHToexnsuTgX3UlrIvhkRO0TESi3pJwPvqJ1v21JaJE+ppl9FOb4X1eO3m0b6/PvR0D/V58eZlHNrLEarS7A+78zMXw9NZOaNlM+hoX29MRDAqbU8T1TTbVuQKydXf4e65L2wyj/SfvwzsGxEHB8RW9Vbiqvt3kVpXd6hNnsH4LrMHOqe+Trg3ixPQJ6UDJD7z3i+QFemXPpv1W5eO/e3TD9KuYw8GYxnv64A/Cul1bH+ei+wRkve8ezD+zLz0ZZ5q9C+7Hfw1OWwIXcOM70KVM3W5UNzt9rlssUpXWsWRU32zXjPiXbrX4HS5aL1eHgBTx0P493upFd9+W8F3E7p4nN71Rf2FVWWkyn7+nXV9A7AbzLzpmr6+dXf4S756ykjff61+9xYZSwrb1CXYH227ueheUP7ehXg75n5cEueO4BnV93tFpCZD1F+ZLy3mrU7pR7OHa4gmXkNsB3lM2sWcHdEnFh1WxtyErBtRCxT9R1/F08F41Dqa7LWFWCA3I/G8+zv2ymXRlq1mzfVjGe/3gtcSvmF3/p62wjLjVW7Mt5GGW2j1bSqXHWt+Yam6x9i36EEca+lfJCekZn3jbmk/aHJvhnvOdGuTu6l9JNsdzx8rkPbnRIy8+rMfAfl8voWlB+T50TEYllG27kU2KHq0/1mnv4FfU/1d0zB3BQ10udfu8+NMQc+I9VllT7V67PdZ1V9X98GLF3tm7ppwMOZOX+EdR8DvDoi1gHeA5yQo4y2k5nnZOb/owS676PU2RG1LD+itGhvR2mRXpUF62uy1hVggDzZzAZeGRGrDc2IiE0oJ5gW3oXAi4CbMvPSltefx7iuRwFabswYySWUOl17aEZVv5sBrUPTbddyl/jbKX1o/zI0IzNvpnQDOZjRL8P1u0uArauuIgBExMaUvuVD+2Y2sHJ1HgzlWQ145Ti2eyHlprzL2hwPN9S267nYUGb+MzMvotzEtAolyILSivW26vUsapefgd9Qju/duljUyag+asVilIDodwu7shHqEqZ2fa4UEZsNTUTEmsBGPLWvZ1N+yLyzlieq6RGHIa26blxDab1fk9LnuZHMfCAzT6QExOvX5t9H+a7YoXpdlZl/qi16IbB8RLyp6bYWNY6DPLl8BzgAODsiDqZ8AB1Muaz7xEgLakQnUG6WG4yI/waup/zq3oTSv3UsQ6RdXf3dNyIuAh6sLncN5zjK5fyfRMSBlBvBDgLuprRi1j2Xcgf60ZQA7jPAkVlGsKj7NuWLaS5wwRjK3m++SrmD+ryI+DLlTvcvUfrXnV7lmUW5OeiUiPgU5Qv4IMply4U9JwYoX2rnRMSxlLpYDdiScgPeIJ6Lo4qIlwL/TWmVuh54HuVYv7x2zJ5CGQ3mUMoNsE+2bGbm/RHxWeDz1aglsyg3fr0ROLje/1sjen9EPEr5If1+SmPATmNZQcO6hKldn3cD36tG6niE8nlwJ1Uwm5lXRcQPgK9XP/qvo4xc9GLK59xovk3Zr7/JzKtHyhgRH6D0+T4XuJVyk/G7KN91dSdTgu4HgK+3pF1AGannxIg4BPg95QfRazLzAw3K2/dsQZ5Eqr5L21BOvpMpX+SfoPSLfXD4JTWSLA/reC3lA+Fgyq/q/6V8qIy1peViyofYvpQW0NYgt3Xb8ymXvq6mfAAeD9wEzGwT+B5GuUz3A+DAKv9/tVnt2ZQ7lI9vualvkVLdSPJayriuP6AMNXQxsOVQX+6q3/V2lP33HUq9/R/l5smFOicy81pKn/SHgaMoN1YeDMwH5lR5PBdHdzvlh8qnKfvwG8BVlCHCgCevePya8sV7UusKMvOLlOBhC8rNZd+itFg+NMFln0x2pLTongG8DNghM/8wxnWMWpcw5evzRsqQogOU9/4QsHU+/WFQe1I+4w+kvP+1KKMsNXmQ1RnV3yZDdv6J0t3rq5TvswMoQ8x9siXfmZTvihVoqa/qs/VtlM/A/6DU++coPwQmhSjvUZNVdWn+WmCvzFyUL6drBBGRwD6Z2forv13eN1CC5HWzGn90Kony0IHrga9n5kFd3K7nojQFRXk40AaZ2TrqUSe38SHKWNer1kf20cKzi8UkU11GvpXya3VN4FOUy7qnj7ScJr+IWJXS6v0lYNZUCY4j4oOUbg1/pbSa7Ee5dDuhD0fxXJQ00aI82XBdytXC4wyOO8cAefJJSh/LVSmXfC+mDMTuSaO9KJfSfk956txUMfRo8bUo58fvgC2qcUgnkueipIk2QHky388p952oQ+xiIUmSJNV4k54kSZJUY4AsSZIk1RggS5IkSTUGyJIkSVKNAbIkSZJU8/8BRMpsBOpu++4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(10, 5))\n",
    "ax.bar([0,1,2,3,4,5], [mean_gini, mean_entropy, np.mean(scores), mean_svc, mean_rbf_svc, mean_poly_svc], yerr=[std_gini, std_entropy, np.std(scores), std_svc, std_rbf_svc, std_poly_svc], align='center', alpha=0.5, ecolor='black', capsize=10)\n",
    "ax.set_ylabel('accuracy on test set', fontsize=15)\n",
    "ax.set_xticks([0,1,2,3,4,5])\n",
    "ax.set_xticklabels([\"rf gini\", \"rf entropy\", \"log reg\", \"svc\", \"rbf svc\", \"poly svc\", \"lin svc\"], fontsize=15)\n",
    "ax.set_title('Scores of different models', fontsize=20)\n",
    "ax.yaxis.grid(True)\n",
    "\n",
    "# Save the figure and show\n",
    "plt.tight_layout()\n",
    "plt.savefig(\"/home/tamara/Documents/PECLIDES/pics/scores_different_models_100_3.svg\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_svms.to_csv(\"/home/tamara/Documents/PECLIDES/df_svms.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "red_ruleset = []\n",
    "new_ruleset = []\n",
    "\n",
    "feature_names = config.feature_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "groundTruth = helpers.create_separate_files_status(\"data.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, x_test, y_train, y_test = train_test_split(helpers.create_dataframe(\"dataWoName.csv\"), groundTruth, test_size=0.33, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test_rf_criterions(nr_iterations):\n",
    "    df = pd.DataFrame(columns=[\"criterion\", \"score\"])\n",
    "    for i in range(nr_iterations):\n",
    "        x_train, x_test, y_train, y_test = train_test_split(helpers.create_dataframe(\"dataWoName.csv\"), groundTruth, test_size=0.33,\n",
    "                                                        random_state=i)\n",
    "        rf = RandomForest(x_test, y_test, x_train, y_train, \"entropy\")\n",
    "        score = rf.score\n",
    "        \n",
    "        new_row = {'criterion':'entropy', 'score':score}\n",
    "        df = df.append(new_row, ignore_index=True)\n",
    "        \n",
    "        x_train, x_test, y_train, y_test = train_test_split(helpers.create_dataframe(\"dataWoName.csv\"), groundTruth, test_size=0.33,\n",
    "                                                        random_state=i)\n",
    "        rf = RandomForest(x_test, y_test, x_train, y_train, \"gini\")\n",
    "        score = rf.score\n",
    "        \n",
    "        new_row = {'criterion':'gini', 'score':score}\n",
    "        df = df.append(new_row, ignore_index=True)\n",
    "\n",
    "        \n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_random_forests = test_rf_criterions(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_gini = np.mean(df_random_forests[df_random_forests.criterion == \"gini\"].score)\n",
    "mean_entropy = np.mean(df_random_forests[df_random_forests.criterion == \"entropy\"].score)\n",
    "std_gini = np.std(df_random_forests[df_random_forests.criterion == \"gini\"].score)\n",
    "std_entropy = np.std(df_random_forests[df_random_forests.criterion == \"entropy\"].score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:cdss_python2] *",
   "language": "python",
   "name": "conda-env-cdss_python2-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": "2.7.17"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}