05_example_prediction_Age_sexe_internaute.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "<h1>Scikit-Learn</h1>\n",
    "\n",
    "\n",
    "Cet exercice simule sur une mission pour une compagnie gérant un site web.<br>\n",
    "On nous demande de prévoir l'âge et le sexe d'un internaute en regardant les pages qu'il visite. <br>\n",
    "Vous avez donc deux fichiers dont chaque ligne représente un internaute :<br>\n",
    "1) X : une colonne par page web, et le nombre de visites de l'internaute<br>\n",
    "2) Y : des informations sur l'internaute : âge / sexe / âge bucketisé \n",
    "\n",
    "On demande de prévoir la tranche d'âge et le sexe en s'appuyant sur les informations de pages vues. <br>\n",
    "Le notebook ci-après est un début de réponse à cette question."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# -1) Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Applications/anaconda/lib/python2.7/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.\n",
      "  warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import pandas as pd\n",
    "import os\n",
    "import sys\n",
    "import numpy as np\n",
    "from datetime import datetime\n",
    "from collections import defaultdict\n",
    "import itertools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import scipy.stats as stats\n",
    "import pylab\n",
    "import matplotlib.pyplot as plot"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 0) Useful functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def print_time():\n",
    "    \"\"\"\n",
    "    Affiche et retourne l'heure de la machine.\n",
    "    Utilisé pour mesurer le temps d'execution du classifieur.\n",
    "    \"\"\"\n",
    "    import time\n",
    "    import datetime\n",
    "    t = datetime.datetime.now().time()\n",
    "    #print t.isoformat()[:8]\n",
    "    return time.time()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def draw_confusion_matrix(y_test, predictions, title):\n",
    "    \"\"\"\n",
    "    Créé et affiche la matrice de confusion des données de Y_test et de celles sur lesquelles on a fait la prédiction.\n",
    "    Permet d'évaluer la qualité de la classification souhaitée.\n",
    "    Plus les valeurs diagonales sont élevées, plus le classifieur prédit correctement.\n",
    "    \"\"\"\n",
    "    from sklearn.metrics import confusion_matrix,accuracy_score\n",
    "    import matplotlib.pyplot as plt\n",
    "    cm    = confusion_matrix(y_test, predictions)\n",
    "    accur = accuracy_score  (y_test, predictions)\n",
    "    \n",
    "    plt.matshow  (cm)\n",
    "    plt.ylabel   ('True Label')\n",
    "    plt.xlabel   ('Predicted Label')\n",
    "    plt.suptitle ('Confusion matrix (Accuracy of %.2f) for [%s]'%(accur,title))\n",
    "    plt.colorbar ()\n",
    "    plt.show     ()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def draw_roc_curve(fitted_c, x_test, y_test, title):\n",
    "    \"\"\"\n",
    "    Créé et affiche la courbe ROC des données de test X et Y selon un classifieur (SVM, KNN, RF).\n",
    "    Permet de représenter l'évolution de la vitesse d'apprentissage du classifieur.\n",
    "    Taux de vrais-positifs en fonction du taux de faux-négatif.\n",
    "    Plus l'aire sous la courbe est grande, plus le classifieur est idéal.\n",
    "    \"\"\"\n",
    "    from sklearn.metrics import roc_curve, auc\n",
    "    import matplotlib.pyplot as plt\n",
    "    c                                        = fitted_c\n",
    "    probas                                   = c.predict_proba(x_test)\n",
    "    false_positive_rate, recall_, thresholds = roc_curve(y_test, probas[:,1])\n",
    "    roc_auc                                  = auc(false_positive_rate, recall_)\n",
    "    \n",
    "    plt.title  ('ROC %.2f %s'%(roc_auc, title))\n",
    "    plt.legend (loc=\"lower right\")\n",
    "    plt.plot   ([0,1],[0,1], \"r--\")\n",
    "    plt.plot   (false_positive_rate, recall_, 'b', label='AUC = %.2f'%roc_auc)\n",
    "    plt.xlim   ([0.0,1.1])\n",
    "    plt.ylim   ([0.0,1.1])\n",
    "    plt.ylabel ('Recall')\n",
    "    plt.xlabel ('Fall-out')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1) Chargement des données"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X = pd.read_csv(\"base_x.csv\", index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "Y = pd.read_csv(\"base_y.csv\", index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Fait correspondre l'index des données X avec celles de Y\n",
    "assert(sum(Y.index == X.index) == Y.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1) First checks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "         Number of rows and columns : 39045 rows, 7 columns \n",
      "    Number of categorical variables : 1 variable, and 10 modalities\n",
      "       Number of values by modality :\n",
      "                                       (14, 18]   151\n",
      "                                       (18, 20]   357\n",
      "                                       (20, 25]  1734\n",
      "                                       (25, 30]  4070\n",
      "                                       (30, 35]  4150\n",
      "                                       (35, 40]  4702\n",
      "                                       (40, 45]  4591\n",
      "                                       (45, 65] 16136\n",
      "                                       (65, 70]  1743\n",
      "                                      (70, 150]  1411\n",
      "                  Moyenne des âges : 45.5\n",
      "                         Ecart type : 13.6\n",
      "           Frontière des Quartiles : 18.0 , 35.0 , 45.0 , 55.0 , 95.0\n",
      "          Ecartement inter quartile : 17.0 , 10.0 , 10.0 , 40.0\n",
      "             Frontière des Deciles : 18.0 , 28.0 , 33.0 , 37.0 , 41.0 , 45.0 , 49.0 , 53.0 , 58.0 , 64.0\n",
      "            Ecartement inter decile : 10.0 , 5.0 , 4.0 , 4.0 , 4.0 , 4.0 , 4.0 , 5.0 , 6.0\n"
     ]
    }
   ],
   "source": [
    "# Statistiques descriptives\n",
    "# Nombre de row, de variable, nb de modalité de la variable âge\n",
    "# Moyenne, écart type, quartiles, déciles\n",
    "\n",
    "for data in [Y]:\n",
    "    title = \"Number of rows and columns\"\n",
    "    print \"% 35s : %s rows, %s columns \"%(title, data.shape[0], data.shape[1])\n",
    "    title = \"Number of categorical variables\"\n",
    "    print \"% 35s : 1 variable, and 10 modalities\"%title\n",
    "    title = \"Number of values by modality\" \n",
    "    print \"% 35s :\"% (title)\n",
    "    t = data.bucket.value_counts()\n",
    "    t = t.sort_index()\n",
    "    for x,y in t.iteritems():\n",
    "        print \"% 37s % 9s % 5s\"%(\"\", x,y)\n",
    "\n",
    "    title = \"Moyenne des âges\"\n",
    "    print \"% 35s : %.1f\"%(title, data.age.mean())\n",
    "    title = \"Ecart type\"\n",
    "    print \"% 35s : %.1f\"%(title,data.age.std())\n",
    "\n",
    "    quartiles = {}\n",
    "    for i in range(5):\n",
    "        quartiles[i] = np.percentile(data.age, i*100/4)\n",
    "\n",
    "    title = \"Frontière des Quartiles\"\n",
    "    print \"% 35s : %s\"%(title, \" , \".join([\"%s\"%y for x,y in quartiles.iteritems()]))\n",
    "\n",
    "    title = \"Ecartement inter quartile\"\n",
    "    print \"% 35s : %s\"%(title, \" , \".join([\"%s\"%(quartiles[i]-quartiles[i-1]) for i in range(1,5)]))\n",
    "\n",
    "\n",
    "    decile = {}\n",
    "    for i in range(10):\n",
    "        decile[i] = np.percentile(data.age, i*100/10)\n",
    "    title = \"Frontière des Deciles\"\n",
    "    print \"% 35s : %s\"%(title, \" , \".join([\"%s\"%y for x,y in decile.iteritems()]))\n",
    "\n",
    "    title = \"Ecartement inter decile\"\n",
    "    print \"% 35s : %s\"%(title, \" , \".join([\"%s\"%(decile[i]-decile[i-1]) for i in range(1,10)]))\n",
    "\n",
    "# REMARQUES :\n",
    "# La répartition de la population dans les catégories d'âge n'est pas homogène\n",
    "# Certaines classes d'âge sont sous-représentées (jeunes, vétérans)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## L'âge est-il normal ?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((array([-4.13494008, -3.92652537, -3.81279465, ...,  3.81279465,\n",
       "          3.92652537,  4.13494008]),\n",
       "  array([ 18.,  18.,  18., ...,  94.,  95.,  95.])),\n",
       " (13.475733117621862, 45.471046228710456, 0.99294913534192819))"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEZCAYAAAB4hzlwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKCmIiEiMkoKIiMT8P/+n9+viw41NAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1566013d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Répartition des ages selon les quantiles (création d'intervalles contenant le même nombre de données)\n",
    "stats.probplot(Y.age, dist=\"norm\", plot=pylab)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2) Binarization des âges"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Mise en place de nouveaux intervalles d'âge pour obtenir une meilleur répartition de la population dans les classes d'âge\n",
    "bins = np.array([13,19,24,29,34,39,44,49,54,59,100])\n",
    "labels = pd.cut(data.age, bins)\n",
    "Y['bucketed_age'] = labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "dataSerie = pd.Series(Y.bucketed_age)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Applications/anaconda/lib/python2.7/site-packages/ipykernel/__main__.py:1: FutureWarning: sort is deprecated, use sort_values(inplace=True) for INPLACE sorting\n",
      "  if __name__ == '__main__':\n"
     ]
    }
   ],
   "source": [
    "dataSerie.sort()\n",
    "x = dataSerie.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "x = x.sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1565b0c10>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAEuCAYAAACAv9lxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xu0HGWd7vHvEwIEJFyFbEkgoExCQEWiZrwO2xvIuAQ8\nc4zxcA8qA3hgLl4Sz5lJHGeU6KwRb6CzRiEggvGCoMYQMWkdz4iAgCgJJCqEJJDNPYA4mJjf+aPe\nnV3p7Esnuy/vrv181tor1W9X1/t0dfUv1W9VdykiMDOz6hrT6QBmZtZaLvRmZhXnQm9mVnEu9GZm\nFedCb2ZWcS70ZmYV50KfGUnzJF2Vpg+R9JQk7cRy5kr69+YnzJek5ZJmD3L/ZZL+z04s90xJ/zm8\ndMPX6hySFks6vXT7nyU9IunBtC0+vTPbYgP9Pi3psGYvt9kkbZH0wgHuy2IbGcjYTgdoJkk14KXA\nhIjY1OE4wxEAEbEW2HuomSUdB3w1Ig7ZuoCIT7QuXv4knQm8JyJe39sWEecNY5G5fOGkZTki4i97\npyUdAvwdcEhEPJaaxw+3D0nLgasi4iulfoe93DYZat3nso1spzJ79JImA68DtgAndTgOAJJ2aVdX\nZLyRdYjXyfBMBh4tFflKa/C92vRPM+1SmUIPnAH8DLgCOKt8h6T9JX1X0kZJP5f0sfLHLElHSloq\n6TFJKyW9c6BO0vDAx9NyNkq6TtK+6b7J6ePdbElrgB+l9ldJ+n+SnpB0R9oD713eYZJqaVk3As8v\n3de7vDHp9n6SviJpfcr6bUl7AouBg9NH4KckdZWHgNJjT5L0a0mPS1om6cjSffdJ+ntJv0wZr5G0\nWyMrXdILJf1I0qOSHpb0VUlDfgqpe35nSXogPadzJb0iZXlc0udK89c/p23WT6n9SOAy4NVpnTye\n2i+X9E+D5Jkk6VvpeTwi6bMDzHdJyrtR0q2SXle675WpbaOkhyT9a2rfXdJVaT09kbafA3PKkbbt\n2ZLeBCyl2KaeStvckNtiat9XxXvt4dT+XUkHp/v+GXg98Pm03M+m9q1DIpL2lnRlevx9Kg21KQ2P\nSPpU2jZ+K+mtg7ye90maI+nulOXLvdu1pOMkrZX0IUkPAV9J7e+VtDqtn+9IekHdYt+W+n1Y0icH\n6XvAmpK2wy+oGCp7Oj2nCZI+nZ7XCknHDLTsnRIRlfgDVgPnAtOBPwIHlu67FvgasDswDXgA+Em6\nb890+wyK/7GPAR4Gjhygn+XA2rScPYBvUnwUhWIvaAvFfzZ7pP4OBh4FTkjzvCndPiDd/i/gU8Cu\nFG+Cp4ArS8v7EzAm3f4+cA3FcM4uwOtT+3HAA3U555WWMwV4BnhjetwH0/oam+6/D7gZmADsC6wA\n3jfA8z8EeByYlG6/KD2nscABQA34twZfs971dSmwG/Bm4A/At9OyDgZ6Ss9z63MaYP0sB2an6TN7\nX+PS/JcD/zRAljHAncC/AuNSntf0tyzgf6X1NAb4W+AhYLfS63lqaduakabfB1yftgkBxwJ75ZSj\nbv1ts031s64H2hb3B96Rlv884OvAdXXvn9l1z/lPwAvT9JXAdSnzZOBe4OzS838OmJ2y/zWwfpDt\n6z7grrQd7Qv8tPf1T89vE/Bxivfe7hTvj0coasCuwGeBH5eWt4Vi520fYFLKtt32Rv815RFSTaHY\nDh8GXpZe3x8BvwNOTfN/DFjW1PrYiqLb7j+KIZvngP3S7RXARaU3zh+BI0rzf6z0oswsv5ip7YvA\nPwzQ13Lg46Xb01LfKr0ZJpfu/xCwsG4ZS4DTKYrmH4E9SvddTT+FHngBsBnYu59MQxX6/wtcW7pP\nwDrgL0pviHeX7l8AXLqTr8XJwC8anLf3+XWV2h4F3lm6/U3gwvrnVL9+Sq/Nzhb6V1H8pzKmn/u2\nW1bd/Y8DL0nTtZTzgLp5zqYoNC8ZYp10LAcNFvrBtsV+lvky4LH++ii1bQFemJb9HDC1dN/7SEUv\nPf9Vpfv2SJkOGqDv+4D3lm6fCKwuPb//BnYt3f8fwMWl28+jeH8eWsr5ltL95wE/rH9tGKKmpO3w\nS6X73g/cXbr9YuDxnXn/DfRXlaGbM4ClEfFEun0NxYoHOJBij2Ndaf61penJwKvSR6bHJT1BsafU\nNUh/5cevofjf//mltnJfk4GZdct/LcWb5WDgiYj4Q93y+jOJ4sV/apBcAzm4vNwotqa1wMTSPD2l\n6WeBvRpZsKSDVAz1rJP0JPBVtl0XjXi4NP2Huix/aDTLjih9bH5K0rsp/tNdExFbGnjsB9LH6yfS\n67k3fc/5HGAqcE8aFnlbar8KuBG4Nq2ri9X/uHC7cywYIMdgBtwWJe0h6UuS7k/bw4+BfaWGztZ5\nPsUnwwdKbWvYdjvd0DuR3jdi8O2j/F5cQ/Fe6PVIbHvSRv375PfAY3X9D7a8XgPVlAmleeq38ZZu\n8yP+rBtJ4yj+Bx2Txtqg+Di0r6SXAHdT7H1MAn6T7j+ktIi1QC0iTtiBbsuPn0zxv/6jwKGpLeqW\nf2VEnNtP9kOB/STtUSr2h1LsOdRbC+wvae9+3mDRz/xlD1LsJdQ/h3X9zLujPk6R9+iI2CjpZOBz\nQzxmZ/2e4mNxr/rx07JB10mUzjCB4jgKcKikMYMVWUmvpxj6ekNErEhtj5MO1EXEbyne1Ej6K+Cb\nkvZPr+/HgI+l1/0HFB/9L6/rYm0HctzTT47BDLYt/j3wZ8ArI+KRNNZ8O30Hxwd7XR6lGE6ZnDKR\nptfvQLZ69e/VB0u367M8mOYBQNLzKIYQy++TQ4CVafrQuuX12pma0lJV2KN/B0Uhn0YxFnZMmv4p\ncEZ6s1wHzE97G0dSfALo9T1giqTTJI2VtKuKg4FHMrDT0sGWPYGPAt9Ie8mw/ZH5rwJvl3S8pDGS\nxqUDQQdHxAPAbcBHU7+vA95e9/jeN+4GijflpSoOeI1Nb3Yo9gYO0MAHQRdRHER6Q3rcByg+tv5s\nkOfYqPEU4/9PS5pIUXx2xI6cyXAn8BcqzuneB5gzyLw9wCRJuza47FsoxrgvlrSnioOWr+lnvr0o\nitFjknaT9I+UTjuUdKqk3r3qjRTFZIukbkkvVnEw85m0jP4KeSdy/KmhNTTwtrhraVscT7FH+pSk\n/YH5dcvooRim2U56ry4C/kXSXirOpPtbik8hO+sCSRNTlo9QHK8byDXA2ZJeKml3ip2Ym6M4zbnX\nB9NzPgS4aIDlDVRTpu5A7qae4VOFQn8G8JWIWB8RD/f+AZ8HTk0b9PspDsY8BCykODD7HEBEPAMc\nD8yi+N/5QeBiik8FA7kqLefBNN9Fpfu22UuIiHUU49YfoTggswb4AH3r/lSKcdnHgH9Iy91mEaXp\n0yn+U7uH4g1zUerjXoqN9Hfpo+I2w04RsQo4La2TR4C3AW+PiM39ZR6M+r7ENSk1fRR4OfAk8F3g\nW40ua4C+B7wdETdRHNy7C7g19TfQY5dRfJrbIOlhhpCKzNsp9kYfoNgrm9nPrDemv1UUY8DPsu1Q\n3luBuyU9BXwaeFdEPEcxFPhNiqJ7N8VY9XYFrEM5vtrbfb8rpxSvNF3eFjfQ9x64hOJT16MUB4QX\n1y3jM8A7VZyNckk/y70wPZffAT+h+H7IYJ82hsr8NYoziH5DcQLCvwy4oIgfUbwHv03xKeJwirpQ\n7ut64BcUn1K+Szpbp245A9WU3YfIus1idmDeIalvR3SAGaQpFG+uoPhf5oUUK+Oq1D4ZuB+YGREb\n02PmUhwZ30xxUHRpap9OcUbKOGBxRPxNM59MoyRdTPGlqrN34rHLqfvCh5nlR9J9wDkRsazTWTpt\nyD36iFgVEcdGxHSKPbffUwyFzAFuioipFHtPcwEkHUWxBzKN4ij3paUDMZdRrPgpFB9t2jKGJWlq\nGq9H0gyKA1XfbkffZmadtqNDN28GfpvGrE6mb5hhIXBKmj6J4lS+zRFxP8XHpRlpOGF8RNya5ruy\n9JhWGw98W9IzFEMcn4qI+o/9jWrqRyozaxm/V5MdPevmXRRjXlAMffRAcXBG0kGpfSLbHuRbn9o2\ns+3R63Vse9pSy0TEbRRjns1Y1hubsRwza62I6Peg72jU8B59OnvhJOAbqWmog2hmZpaBHdmjP5Hi\nG4+Ppts9kiZERE8aluk9s2E92567Oim1DdS+HUn+T8PMbCdExHanZu7IGP27Kca3e91A34+HnUlx\n2lFv+6x0bu/hwBHALenc242SZqSDs2eUHtNf2J3+mzdvXtO+OjzSc+SQIZccOWTIJUcOGXLJkUOG\nZuUYSEN79OmLQW+m+N2JXguARSou9LCGdK5vRKyQtIji92Y2AedHX4IL2Pb0yiWN9G9mZjuvoUIf\nEc9S/GZMue1xiuLf3/yfALa78EVE/AJ4yY7HNDOznVWFb8Zup7u7u9MRgDxy5JAB8siRQwbII0cO\nGSCPHDlkgNbmGPKbsZ0gKXLMZWaWM0nEMA/GmpnZCORCb2ZWcS70ZmYV50JvZtZiXV2HIWlYf11d\nh+10/z4Ya2bWYsV3RIdb0zTol6J6+/HBWDOzUciF3sys4lzozcwqzoXezKziXOjNzCrOhd7MrOJc\n6M3MKs6F3sys4lzozcwqzoXezKziXOjNzCrOhd7MrOJc6M3MKs6F3sys4lzozcwqzoXezKziGir0\nkvaR9A1JKyXdLenPJe0naamkeyXdKGmf0vxzJa1O8x9fap8u6S5JqyRd0oonZGZm22p0j/4zwOKI\nmAYcA9wDzAFuioipwDJgLoCko4CZwDTgROBSFZdXAbgMOCcipgBTJJ3QtGdiZmb9GrLQS9obeH1E\nXA4QEZsjYiNwMrAwzbYQOCVNnwRcm+a7H1gNzJDUBYyPiFvTfFeWHmNmZi3SyB794cCjki6XdLuk\nf5e0JzAhInoAImIDcFCafyKwtvT49altIrCu1L4utZmZWQs1UujHAtOBL0TEdOD3FMM29Vep9dW8\nzcwyNLaBedYBayPitnT7WxSFvkfShIjoScMyD6f71wOHlB4/KbUN1N6v+fPnb53u7u6mu7u7gahm\nZqNHrVajVqsNOZ8iht4Rl/Rj4L0RsUrSPGDPdNfjEbFA0oeB/SJiTjoYezXw5xRDMz8E/iwiQtLN\nwIXArcD3gc9GxJJ++otGcpmZjQTF+SjDrWliqLooiYhQfXsje/RQFOerJe0K/A44G9gFWCRpNrCG\n4kwbImKFpEXACmATcH6pal8AXAGMoziLZ7sib2ZmzdXQHn27eY/ezKqk03v0/masmVnFudCbmVWc\nC72ZWcW50JuZVZwLvZlZxbnQm5lVnAu9mVnFudCbmVWcC72ZWcW50JuZVZwLvZlZxbnQm5lVnAu9\nmVnFudCbmVWcC72ZWcW50JuZVZwLvZlZxbnQm1VQV9dhSBrWX1fXYZ1+GtYkvpSgWQW169J11hhf\nStDMzFrKhd7MrOJc6M3MKs6F3sys4hoq9JLul/RLSXdIuiW17SdpqaR7Jd0oaZ/S/HMlrZa0UtLx\npfbpku6StErSJc1/OmZmVq/RPfotQHdEHBsRM1LbHOCmiJgKLAPmAkg6CpgJTANOBC5VccgZ4DLg\nnIiYAkyRdEKTnoeZmQ2g0UKvfuY9GViYphcCp6Tpk4BrI2JzRNwPrAZmSOoCxkfErWm+K0uPMTOz\nFmm00AfwQ0m3SnpPapsQET0AEbEBOCi1TwTWlh67PrVNBNaV2telNjMza6GxDc732oh4SNKBwFJJ\n97L92f9N/WbF/Pnzt053d3fT3d3dzMWbmY14tVqNWq025Hw7/M1YSfOAZ4D3UIzb96RhmeURMU3S\nHCAiYkGafwkwD1jTO09qnwUcFxHn9dOHvxlrNgz+Zmxesv9mrKQ9Je2Vpp8HHA/8CrgBOCvNdiZw\nfZq+AZglaTdJhwNHALek4Z2Nkmakg7NnlB5jZmYt0sjQzQTgOkmR5r86IpZKug1YJGk2xd76TICI\nWCFpEbAC2AScX9o9vwC4AhgHLI6IJU19NmZmth3/qJlZBXnoJi/ZD92YmdnI5kJvZlZxLvRmZhXn\nQm9mVnEu9FYZvnyeWf981o1Vhs806eN1kRefdWNmZi3lQm9mVnEu9GZmFedCb2Yt4YPj+fDBWKsM\nH4Dsk8O6yCFDLnww1szMWsqF3qyJPFxhOfLQjVVGDkMFOWTIJUcOGXLhoRszM2spF3ozs4pzoTcz\nqzgXejOzinOhNzOrOBd6M7OKc6E3M6s4F3ozs4pzoTczq7iGC72kMZJul3RDur2fpKWS7pV0o6R9\nSvPOlbRa0kpJx5fap0u6S9IqSZc096mYmVl/dmSP/iJgRen2HOCmiJgKLAPmAkg6CpgJTANOBC5V\n8f1fgMuAcyJiCjBF0gnDzG9mZkNoqNBLmgT8JfAfpeaTgYVpeiFwSpo+Cbg2IjZHxP3AamCGpC5g\nfETcmua7svQYG+GG+2Ne/iEvs9YZ2+B8nwY+COxTapsQET0AEbFB0kGpfSLws9J861PbZmBdqX1d\narcK6OlZw3B+tKmnZ7vfYTKzJhmy0Et6G9ATEXdK6h5k1qb+xNz8+fO3Tnd3d9PdPVjXZmajT61W\no1arDTnfkD9TLOnjwGkUe+R7AOOB64BXAN0R0ZOGZZZHxDRJc4CIiAXp8UuAecCa3nlS+yzguIg4\nr58+/TPFI8zwf4bVP83brAy55MghQy6y/5niiPhIRBwaES8EZgHLIuJ04LvAWWm2M4Hr0/QNwCxJ\nu0k6HDgCuCUiNgAbJc1IB2fPKD3GzMxapNEx+v5cDCySNJtib30mQESskLSI4gydTcD5pd3zC4Ar\ngHHA4ohYMoz+zcysAb7ClDWFh27yyZBLjhwy5CL7oRszMxvZXOjNzCrOhd7MrOJc6Ee44X4j1d9K\nNas+H4wd4XI54OWDsflkyCVHDhly4YOxZmbWUi70ZlZpHt700M2Il8vHYw/d5JMhlxw5ZMglh4du\nzMyspVzozcwqzoXezKziXOjNzCrOhd7MrOJc6M3MKs6F3sys4lzozcwqzoXezKziXOjNzCrOhd7M\nrOJc6M3MKs6F3sys4lzozcwqzoXezKzihiz0knaX9HNJd0j6laR5qX0/SUsl3SvpRkn7lB4zV9Jq\nSSslHV9qny7pLkmrJF3SmqdkZmZlQxb6iHgOeENEHAu8DDhR0gxgDnBTREwFlgFzASQdBcwEpgEn\nApeq+NV9gMuAcyJiCjBF0gnNfkJmZrathoZuIuLZNLk7MJbiUiknAwtT+0LglDR9EnBtRGyOiPuB\n1cAMSV3A+Ii4Nc13ZekxZmbWIg0VekljJN0BbAB+mIr1hIjoAYiIDcBBafaJwNrSw9entonAulL7\nutRmZmYtNLaRmSJiC3CspL2B6yQdzfYXQGzqRV7nz5+/dbq7u5vu7u5mLt7MbMSr1WrUarUh59vh\ni4NL+gfgWeA9QHdE9KRhmeURMU3SHCAiYkGafwkwD1jTO09qnwUcFxHn9dOHLw7eoBwufNycHDlk\nGH6OHDLkkiOHDLnkyP7i4JKe33tGjaQ9gLcAK4EbgLPSbGcC16fpG4BZknaTdDhwBHBLGt7ZKGlG\nOjh7RukxZmbWIo0M3bwAWChpDMV/DF+PiMWSbgYWSZpNsbc+EyAiVkhaBKwANgHnl3bPLwCuAMYB\niyNiSVOfjZmZbWeHh27awUM3jcvhY2lzcuSQYfg5csiQS44cMuSSI/uhGzMzG9lc6M3MKs6F3sys\n4lzozcwqzoXezKziXOjNzCrOhd7MrOJc6M3MKs6F3sys4lzozcwqzoXezKziXOjNzCrOhd7MrOJc\n6M3MKs6F3sys4lzozcwqzoXezKziXOjNzCrOhd7MrOJc6M3MKs6F3sys4lzozcwqzoXezKzihiz0\nkiZJWibpbkm/knRhat9P0lJJ90q6UdI+pcfMlbRa0kpJx5fap0u6S9IqSZe05imZmVlZI3v0m4G/\ni4ijgVcDF0g6EpgD3BQRU4FlwFwASUcBM4FpwInApZKUlnUZcE5ETAGmSDqhqc/GzMy2M2Shj4gN\nEXFnmn4GWAlMAk4GFqbZFgKnpOmTgGsjYnNE3A+sBmZI6gLGR8Stab4rS48xM7MW2aExekmHAS8D\nbgYmREQPFP8ZAAel2SYCa0sPW5/aJgLrSu3rUpuZmbXQ2EZnlLQX8E3gooh4RlLUzVJ/e1jmz5+/\ndbq7u5vu7u5mLt7MbMSr1WrUarUh51PE0PVZ0ljge8APIuIzqW0l0B0RPWlYZnlETJM0B4iIWJDm\nWwLMA9b0zpPaZwHHRcR5/fQXjeQyKA5/DHddieGu7+HnyCHD8HPkkCGXHDlkyCVHuzJIIiJU397o\n0M1XgBW9RT65ATgrTZ8JXF9qnyVpN0mHA0cAt6ThnY2SZqSDs2eUHmNmZi0y5B69pNcCPwF+RfFf\nUgAfAW4BFgGHUOytz4yIJ9Nj5gLnAJsohnqWpvaXA1cA44DFEXHRAH16j75BOeytNCdHDhmGnyOH\nDLnkyCFDLjk6vUff0NBNu7nQNy6Hjbg5OXLIMPwcOWTIJUcOGXLJ0elC72/GmplVnAu9mVnFudCb\nmVWcC72ZWcW50JuZVZwLvZlZxbnQm5lVnAu9mVnFudCbmVWcC72ZWcW50JuZVZwLvZlZxbnQm5lV\nnAu9mVnFudAPQ1fXYUja6b+ursM6/RTMbBTw79EPg3+DvZk5csgw/Bw5ZMglRw4Zcsnh36M3M7OW\ncqE3M6s4F3ozs4pzoTczqzgXejOzinOhNzOrOBd6M7OKG7LQS/qypB5Jd5Xa9pO0VNK9km6UtE/p\nvrmSVktaKen4Uvt0SXdJWiXpkuY/FTMz608je/SXAyfUtc0BboqIqcAyYC6ApKOAmcA04ETgUhXf\nFAC4DDgnIqYAUyTVL9PMzFpgyEIfET8FnqhrPhlYmKYXAqek6ZOAayNic0TcD6wGZkjqAsZHxK1p\nvitLjzEzsxba2TH6gyKiByAiNgAHpfaJwNrSfOtT20RgXal9XWozM7MWG9uk5TT9h2nmz5+/dbq7\nu5vu7u5md2FmNqLVajVqtdqQ8zX0o2aSJgPfjYiXptsrge6I6EnDMssjYpqkOUBExII03xJgHrCm\nd57UPgs4LiLOG6A//6hZ2zLkkiOHDMPPkUOGXHLkkCGXHCPlR82U/nrdAJyVps8Eri+1z5K0m6TD\ngSOAW9LwzkZJM9LB2TNKjzEzsxYacuhG0teAbuAASQ9Q7KFfDHxD0myKvfWZABGxQtIiYAWwCTi/\ntGt+AXAFMA5YHBFLmvtUzMysP/49+mHwcEUzc+SQYfg5csiQS44cMuSSY6QM3ZiZ2QjlQm9mVnEu\n9GZmFedCb2ZWcS70ZmYV50JvZlZxLvRmZhXnQm9mVnEu9GZmFedCb2ZWcS70ZmYV50JvZlZxLvRm\nZhXnQm9mVnEu9GZmFedCb2ZWcS70ZmYVNyILfVfXYUga1l9X12GdfhpmZm0xIi8lmMOlwZqTI4cM\nueTIIcPwc+SQIZccOWTIJYcvJWhmZi3lQm9mVnEu9GZmFdf2Qi/prZLukbRK0ofb3b+Z2WjT1kIv\naQzweeAE4Gjg3ZKObH5PteYvcqfUOh2APDJAHjlqnQ6Q1DodgDwyQB45ap0OkNRatuR279HPAFZH\nxJqI2ARcC5zc/G5qzV/kTql1OgB5ZIA8ctQ6HSCpdToAeWSAPHLUOh0gqbVsye0u9BOBtaXb61Kb\nmZm1iA/GmplVXFu/MCXpVcD8iHhruj0HiIhYUDdfft/iMjMbAfr7wlS7C/0uwL3Am4CHgFuAd0fE\nyraFMDMbZca2s7OI+JOk9wNLKYaNvuwib2bWWln+1o2ZmTWPD8aamVVcW4duzMwMJO3fwGxbIuLJ\npvQ30oduJN3QwGyPR8RZVc+RQ4ZccuSQIZccOWTIJUcOGVKO/wYeBLY7Q6Zkl4g4tBn9VWGPfhrw\nnkHuF/CFUZIjhwy55MghQy45csiQS44cMgCsjIhjB5tB0h1N6y0iRvQfMLMZ81QhRw4ZcsmRQ4Zc\ncuSQIZccOWRIfYxrxjyN/o34oRszs5FK0gT6fgZmfUT0tKSfkV7o0y9ingX8FTAJ+BOwCvhiRNQ6\nlwwkrYqIKW3sz+uirz+vi77+vC76+stiXUh6GfBFYB9gfWqeBDwJnB8Rtze1vwoU+suBNcBNwP8E\nngL+E/gwcH1EfK5NOZ6m76KQvQdY9gSepfiZh73bkMHroi+D10VfBq+Lvgy5rIs7gXMj4ud17a8C\nvhQRxzS1w1aPRbVhrOuuuts3p393pzjg0a4cnwWuBCaU2u7zuvC68Lrwuugnx+pB7vtNs/urwhem\nNkl6EYCk6cAfASLiOYZ/2fWGRcSFwGeAayRdmD4itvvjktdFH6+LPl4XfbJYF8APJH1f0rskvSb9\nvUvS94Elze6sCqdXfhBYLuk5iuczC0DSgcD32hkkIn4h6c3A+4EfA+Pa2T9eF2VeF328LvpksS4i\n4kJJJ1JceGnrwVjgCxGxuNn9jfgxegBJAg6IiEc7naWXpBcAx7biRRuiX6+Lvn69Lvr69bro6ze7\nddFqVRi6IQrbvWiS3tLOHJL27v1YGBEP9W7Akl7axhjjKY7k12drZwYkdUnqSjc3A+MkHd3ODMAE\n0qdWSQdK+h8dyLCNiHgIeF0H+t3mPSLp4+3O0I9xFNtFC64bPbDyupB0eNou2ppB0lhJ50r6gaS7\n0t8PJP21pF2b3l8V9ugHIumBaNJXiBvoayZwCfAwsCtwVkTcmu67PSKmj4YMqa9zgTkUZ1YsoDid\n7dcUBe6TEfHl0ZAh5fhsfRNwOsVByd5x68pnSDm+ExGnpOmTKbbVGvAa4BMRccVoyJD6vobiVMqF\nFJdUheL0yjOB/SPiXc3sb8SP0Q/y2xUCDmhjlI8AL4+IhyTNAK6SNDcirmPw37OoWgYoxl+PBvag\nOJXtiIjYIGk/YDnQjiKbQwaAd1CMRS+l7zWYBfyiTf3nkgFgcmn6w8AbI+I+Sc8HfgRcMUoyQPE+\nrf/+wDrgZkmrmt3ZiC/0wOuB04Bn6toFzGhjjl3Sx3Ii4hZJbwC+J+kQ2nc0P4cMAJsi4lngWUm/\njYgNKdNEz9dVAAAE2ElEQVQTat9lInPIAHAU8DHgrcAHIuJBSfMiYuEoywDbboO7RcR9ABHxqKQt\noygDwOOS3gl8KyK2wNYvc70TeKLZnVWh0N8MPBsRP66/Q9K9bczxtKQXRcRvoRiLldQNfIdiz3K0\nZAAISbtGxCbgbb2NksbRvuNCOWQgIp4G/kbSy4Gr0+lzbT02lkOG5BhJT1HshO0u6QVpG90N2GUU\nZYDiE9UC4FJJvYV9X4pPm7Oa3Vmlx+jbSdIxwO8j4jd17btS/EjS1aMhQ+rvUODBiNhc1z4RmBYR\nN42GDP1kEnA+8OqIOK3d/eeSoZ6kfSlek5+NxgySDgCIiMda1ocLvZlZPiS9JSJ+2NRlutCbmeWj\nFWcLVmGM3sxsRGn32YIu9GZm7dfWswUrW+glLaT4+dMvRMSvR3OOHDLkkiOHDLnkyCFDLjk6kKGt\nZwtWdoxe0iuBQ4EZEfHh0Zwjhwy55MghQy45csiQS44cMrRSZQu9mZkVRvyPmknaU9KHJH1Q0jhJ\nZ0m6QdInJe3VxhxjJM1W8RvTv5R0u6Rr0xeWOqYVX6duoE+vi74+vS76+vS66JARv0cvaRGwluI3\nTaYCK4GvAycBXRFxeptydPwSZcrgUm0ph9dFXw6vi74cXhedEm28jFcr/oA7078CNtD3n5eou2xY\ni3N0/BJlZHCpNq8Lrwuvi/z+RvzQTa8oXrHF6d/e2239IS91+BJlkcel2sDroszroo/XxRAkLZR0\nmaQXN3O5VTi98jZJe0XEMxExu7cxbVBPtzFHLpco6/Sl2sDroszroo/XxdA+T3H2z+kUQ1pNMeLH\n6AcjSdHGJ5h+MCqbS5SpQ5dqS317XfT17XXR17fXRQdUZuhmAG9uZ2dR6OglDZXH5Qwhg0saKo/L\nGUJmlzSMDl3OMPW9zXtEnb+kYUcuZ9juM5CqvkfftksJ5pBD+VxKsOM5lM+lBDueQ/lcSrDjOZTP\npQTbegbSiB+jVyaXEswkRy6XEswhRy6XEswhRy6XEswhR06XEjw7Tf9U0s0R8Y+SfgLcCbjQ18nl\nUoI55MjlUoI55MjlUoI55MjlUoI55MjlUoKblK4GV38GUiu2iyoU+lwuJZhDjlwuJZhDjiwuJZhD\njsjkUoKZ5MjlUoJtPQOp0mP0o42KSwk+GxGr69rbfSnBjudQJpcSzCVHqd8sLiWYS45SnrZfSrCd\nZyCN+ELfyCmU7TjNMoccOWTIJUcOGXLJkUOGXHLkkGGAPg8HjgVWRMQ9zV5+FU6vXC7pf6c9p60k\n7SbpjSp+Z/rMUZIjhwy55MghQy45csiQS44cMiDpO6Xpk4FlwNuB6yWd1fT+KrBHPw6YDZwKHA48\nSXFu7C4UR/cvjYg7RkOOHDLkkiOHDLnkGCDDHhQ7ep1eF23NkcPrkXLcERHHpun/Ak4tn/0TEcc0\ntb+RXujL0hjw84E/RMSTozlHDhlyyZFDhlxy5JAhlxydzKDS90kk3RYRryjdt/U/gab1V6VCb2Y2\nEkj6E/B70tk/wOTS2T+3RURTv0FehdMrzcxGlIgY6FTOPYFzm92f9+jNzNqs3Wf/VOGsGzOzkaat\nZ/94j97MrM3affaPC72ZWQe14+wfF3ozs4rzGL2ZWcW50JuZVZwLvZlZxbnQm5lVnAu9mVnF/X/7\nC1bWehWoIgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1140f7990>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "title=\"Age prediction :a  multi-class classification problem\"\n",
    "x.plot(kind=\"bar\", title=title)\n",
    "\n",
    "# Le problème de répartition dans les classes d'âge semble mieux appréhendé, cependant il est encore un peu présent"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4) Grid search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def RandomForest_grid_search(df, Y):\n",
    "    \"\"\"\n",
    "    Réglage des hyper-paramètres du classifieur Random Forest\n",
    "    \"\"\"\n",
    "    import sys, pickle\n",
    "    from sklearn.ensemble          import RandomForestClassifier\n",
    "    from sklearn                   import grid_search\n",
    "    from sklearn.metrics           import make_scorer\n",
    "    from sklearn.preprocessing     import label_binarize\n",
    "    from sklearn.multiclass        import OneVsRestClassifier\n",
    "    #grid search :\n",
    "\n",
    "    parameters = {  \"n_estimators\"         : [ 10, 100, 200 ], \n",
    "                    \"min_samples_leaf\"     : [  3,   1,   7 ],\n",
    "                    \"criterion\"            : [\"gini\"],#[\"gini\", \"entropy\"]         ,\n",
    "                    \"max_features\"         : [\"auto\"]      ,\n",
    "                    \"max_depth\"            : [None]         , \n",
    "                    \"n_jobs\"               : [ -1]             ,  \n",
    "                    \"class_weight\"         : [\"auto\"]          ,\n",
    "                    \"warm_start\"           : [False]           }\n",
    "    parameters = {  \"n_estimators\"         : [ 10], \n",
    "                    \"min_samples_leaf\"     : [  3 ],\n",
    "                    \"criterion\"            : [\"gini\"],#[\"gini\", \"entropy\"]         ,\n",
    "                    \"max_features\"         : [\"auto\"]      ,\n",
    "                    \"max_depth\"            : [None]         , \n",
    "                    \"n_jobs\"               : [ -1]             ,  \n",
    "                    \"class_weight\"         : [\"auto\"]          ,\n",
    "                    \"warm_start\"           : [False]           }\n",
    "    \n",
    "    grid_type = \"RandomForestClassifier\"\n",
    "    \n",
    "    print \"-\"*30, \"\\n %s\"%grid_type; sys.stdout.flush()\n",
    "    \"\"\"\n",
    "        Search \n",
    "    \"\"\"\n",
    "    \n",
    "    grid         = grid_search.GridSearchCV(RandomForestClassifier(), {})#, parameters, verbose=1, scoring='f1_weighted')\n",
    "    \n",
    "    try :grid.fit(df, Y)\n",
    "    except Exception as e : print \"[%s - 2] : %s\"%(grid_type,e) \n",
    "    \n",
    "    print \" best_score_ %s with =  %s \"%( grid.best_score_,  grid.best_estimator_ ) ; sys.stdout.flush()    \n",
    "    return grid.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def svm_grid_search(df, Y):\n",
    "    \"\"\"\n",
    "    Réglage des hyper-paramètres du classifieur Support Vector Machine\n",
    "    Notamment : 'C', 'kernel' et 'gamma'\n",
    "    \"\"\"\n",
    "    import sys\n",
    "    from sklearn                   import grid_search\n",
    "    from sklearn.svm               import SVC\n",
    "    import pickle\n",
    "    parameters = {  \"C\"              : [1.0],#, 0.5, 0.1, 2]                , \n",
    "                    \"kernel\"         : ['rbf'],#, 'poly', 'sigmoid']        ,\n",
    "                    \"degree\"         : [2],#, 3, 4,5,6]                     , \n",
    "                    \"gamma\"          : [0.01],#, 0.1, 0.001, 0.5]           , \n",
    "                    \"coef0\"          : [0.0]                             , \n",
    "                    \"shrinking\"      : [True]                            , \n",
    "                    \"probability\"    : [True]                            , \n",
    "                    \"tol\"            : [0.001]                           , \n",
    "                    \"cache_size\"     : [10]                              , \n",
    "                    \"class_weight\"   : [None]                            , \n",
    "                    \"verbose\"        : [False]                           , \n",
    "                    \"max_iter\"       : [-1]                              , \n",
    "                    \"random_state\"   : [None]                            }\n",
    "   \n",
    "    \n",
    "    \n",
    "    print \"-\"*30, \"\\n svm_grid_search\"; sys.stdout.flush()\n",
    "    \n",
    "    \"\"\"\n",
    "        Search Grid\n",
    "    \"\"\"\n",
    "    grid    = grid_search.GridSearchCV(SVC(), parameters, verbose=0, scoring='f1_weighted')\n",
    "    try :grid.fit(df, Y)\n",
    "    except Exception as e : print \"[svm_grid_search - 2] : %s\"%e ; print \"x_train, y_train : \", x_train, y_train\n",
    "\n",
    "    print \" best_score_ = %.2f with =  %s \"%(grid.best_score_, grid.best_estimator_ ) ; sys.stdout.flush()    \n",
    "    return grid.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def knn_grid_search(df, Y):\n",
    "    \"\"\"\n",
    "    Réglage des hyper-paramètres du classifieur K Nearest Neighbors\n",
    "    Plusieurs hyper-paramètres sont disponibles\n",
    "    L'entrainement du modèle va s'effectuer sur plusieurs paramètres\n",
    "    Le best_score représentera le modèle avec les meilleurs paramètres\n",
    "    \"\"\"\n",
    "    import sys\n",
    "    from sklearn                   import grid_search\n",
    "    from sklearn.neighbors         import KNeighborsClassifier\n",
    "    \n",
    "    import pickle\n",
    "\n",
    "    parameters = {  \"n_neighbors\"    : [2, 3, 5, 8 , 10]                         , \n",
    "                    \"weights\"         : ['uniform', 'distance']                  ,\n",
    "                    \"algorithm\"      : ['auto', 'ball_tree', 'kd_tree', 'brute'] , \n",
    "                    \"leaf_size\"      : [1,3 ]                                    }\n",
    "    \n",
    "    \n",
    "    grid_type = \"knn_grid_search\"\n",
    "    print \"-\"*30, \"\\n %s\"%grid_type; sys.stdout.flush()\n",
    "    \"\"\"\n",
    "        Search Grid\n",
    "    \"\"\"\n",
    "    print parameters;sys.stdout.flush()\n",
    "    grid    = grid_search.GridSearchCV(KNeighborsClassifier(), parameters, verbose=0, scoring='f1_weighted')\n",
    "    try :grid.fit(df, Y)\n",
    "    except Exception as e : print \"Error : [%s - 2] : %s\"%(grid_type,e) ; return -1\n",
    "\n",
    "    \n",
    "    print \" best_score_ = %.2f with =  %s \"%(grid.best_score_, grid.best_estimator_ ) ; sys.stdout.flush()    \n",
    "    return grid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def clf_grid_search(df, Y):\n",
    "    \"\"\"\n",
    "    Réglage des hyper-paramètres du Classifier Gradient Boosting\n",
    "    \"\"\"\n",
    "    import sys\n",
    "    from sklearn                   import grid_search\n",
    "    from sklearn.neighbors         import KNeighborsClassifier\n",
    "    from sklearn import ensemble\n",
    "    from sklearn import datasets\n",
    "    from sklearn.utils import shuffle\n",
    "    from sklearn.metrics import mean_squared_error\n",
    "    from sklearn.datasets import make_hastie_10_2\n",
    "    from sklearn.ensemble import GradientBoostingClassifier\n",
    "    \n",
    "    import pickle\n",
    "\n",
    "    parameters = {'n_estimators': 10                       ,\n",
    "                  'max_depth': None                        ,\n",
    "                  'min_samples_split': 2                   ,\n",
    "                  'learning_rate': 0.01                    ,\n",
    "                  'loss': 'ls'                             }\n",
    "    \n",
    "    grid_type = \"clf_grid_search\"\n",
    "    print \"-\"*30, \"\\n %s\"%grid_type; sys.stdout.flush()\n",
    "    \"\"\"\n",
    "        Search Grid\n",
    "    \"\"\"\n",
    "    print parameters;sys.stdout.flush()\n",
    "    \n",
    "    \n",
    "    grid = grid_search.GridSearchCV(GradientBoostingClassifier(), {})\n",
    "    try :grid.fit(df, Y)\n",
    "    except Exception as e : print \"Error : [%s - 2] : %s\"%(grid_type,e) ; return -1\n",
    "    print \" best_score_ = %.2f with =  %s \"%(grid.best_score_, grid.best_estimator_ ) ; sys.stdout.flush()    \n",
    "    return grid"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 5) Application grid search"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.1) Variable : Age"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.cross_validation import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Split des données originales en données d'entrainement et de test\n",
    "# La classification se fait en fonction de la variable 'age' \n",
    "# 75% de données de d'entrainement\n",
    "# 25% de données de test\n",
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y.bucket_en_entier)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1.1) SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------ \n",
      " svm_grid_search\n",
      " best_score_ = 0.26 with =  SVC(C=1.0, cache_size=10, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape=None, degree=2, gamma=0.01, kernel='rbf',\n",
      "  max_iter=-1, probability=True, random_state=None, shrinking=True,\n",
      "  tol=0.001, verbose=False) \n",
      "\n",
      "Duration time : 562.668259859\n"
     ]
    }
   ],
   "source": [
    "# Temps de calculs : \n",
    "#100 rows : 1.8s / 0.42 pred\n",
    "#200 rows : 6.5s / 0.42 pred\n",
    "#300 rows : 15s  / 0.42 pred\n",
    "#400 rows : 26s  / 0.42 pred\n",
    "#500 rows : 43s  / 0.42 pred\n",
    "#600 rows : 64s  / 0.42 pred\n",
    "#700 rows : 89s  / 0.42 pred\n",
    "\n",
    "nb_lignes = 2000\n",
    "start_time = print_time()\n",
    "\n",
    "# Apprentissage du classifieur SVM sur X_train et Y_train (<nb_lignes> données)\n",
    "svm = svm_grid_search(X_train[:nb_lignes], Y_train[:nb_lignes].get_values())\n",
    "\n",
    "end_time = print_time()\n",
    "print \"\\nDuration time :\", end_time - start_time\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAF+CAYAAAB5z1o2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nlZCDu5lZCf3/pYnW2H7hf/sAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x18b53b00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Matrice de confusion des données X_test sur lesquelles on fait la prédiction et des données Y_test.\n",
    "# La catégorie d'âge [45,65] semble être celle sur laquelle la prédiction se fait le mieux.\n",
    "# Cependant la précision du modèle étant de 0.42, au mieux, n'est pas du tout concevable dans un modèle prédictif.\n",
    "\n",
    "draw_confusion_matrix(Y_test, svm.predict(X_test), \"\") #Y_test possède seulement la variable 'bucket_en_entier'."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1.2) KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y.bucket_en_entier)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------ \n",
      " knn_grid_search\n",
      "{'n_neighbors': [2, 3, 5, 8, 10], 'weights': ['uniform', 'distance'], 'leaf_size': [1, 3], 'algorithm': ['auto', 'ball_tree', 'kd_tree', 'brute']}\n",
      " best_score_ = 0.13 with =  KNeighborsClassifier(algorithm='auto', leaf_size=1, metric='minkowski',\n",
      "           metric_params=None, n_jobs=1, n_neighbors=5, p=2,\n",
      "           weights='uniform') \n",
      "\n",
      "Duration time : 3.62700009346\n"
     ]
    }
   ],
   "source": [
    "#30 rows  : 4.39s / 0.42 acc\n",
    "nb_ligne = 30\n",
    "start_time = print_time()\n",
    "\n",
    "# Apprentissage du classifieur KNN sur X_train et Y_train (<nb_lignes> données)\n",
    "knn = knn_grid_search(X_train[:nb_ligne], Y_train[:nb_ligne].get_values())\n",
    "\n",
    "\n",
    "end_time = print_time()\n",
    "print \"\\nDuration time :\", end_time - start_time\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAF+CAYAAAB5z1o2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x42ef0240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# La  matrice de confusion s'améliore, tout en étant loin de la réalité.\n",
    "draw_confusion_matrix(Y_test, knn.predict(X_test), \"\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1.3) CLF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y.bucket_en_entier)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------ \n",
      " clf_grid_search\n",
      "{'n_estimators': 10, 'min_samples_split': 2, 'learning_rate': 0.01, 'max_depth': None, 'loss': 'ls'}\n",
      " best_score_ = 0.41 with =  GradientBoostingClassifier(init=None, learning_rate=0.1, loss='deviance',\n",
      "              max_depth=3, max_features=None, max_leaf_nodes=None,\n",
      "              min_samples_leaf=1, min_samples_split=2,\n",
      "              min_weight_fraction_leaf=0.0, n_estimators=100,\n",
      "              presort='auto', random_state=None, subsample=1.0, verbose=0,\n",
      "              warm_start=False) \n",
      "\n",
      "Duration time : 30.4290001392\n"
     ]
    }
   ],
   "source": [
    "#30 rows  : 2.53s / 0.42 acc\n",
    "#200 rows  : 2.53s / 0.42 acc\n",
    "nb_ligne = 200\n",
    "start_time = print_time()\n",
    "\n",
    "# Apprentissage du classifieur KNN sur X_train et Y_train (<nb_lignes> données)\n",
    "clf = clf_grid_search(X_train[:nb_ligne], Y_train[:nb_ligne].get_values())\n",
    "\n",
    "\n",
    "end_time = print_time()\n",
    "print \"\\nDuration time :\", end_time - start_time\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAF+CAYAAAB5z1o2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n3M3MKuj/A8A9je6/xSOiAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2e62c2b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# CLF ne semble pas non plus adapté au vue de la précision du modèle\n",
    "draw_confusion_matrix(Y_test, clf.predict(X_test), \"\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.2) Variable : Sex"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# La classification se fait en fonction de la variable 'sex' \n",
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y.Sex)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2.1) Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------ \n",
      " RandomForestClassifier\n",
      " best_score_ 0.675 with =  RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
      "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
      "            min_samples_leaf=1, min_samples_split=2,\n",
      "            min_weight_fraction_leaf=0.0, n_estimators=10, n_jobs=1,\n",
      "            oob_score=False, random_state=None, verbose=0,\n",
      "            warm_start=False) \n",
      "\n",
      "Duration time : 0.964000225067\n"
     ]
    }
   ],
   "source": [
    "#200   rows : 0.26s / 0.68 pred / 0.66 acc\n",
    "#2000  rows : 1.55s / 0.74 pred / 0.68 acc\n",
    "#20000 rows : 15.5s / 0.78 pred / 0.72 acc\n",
    "nb_lignes = 2000\n",
    "start_time = print_time()\n",
    "\n",
    "# Apprentissage du classifieur RF sur X_train et Y_train (<nb_lignes> données)\n",
    "rf = RandomForest_grid_search(X_train[:nb_lignes], Y_train[:nb_lignes])\n",
    "\n",
    "end_time = print_time()\n",
    "print \"\\nDuration time :\", end_time - start_time\n",
    "\n",
    "# REMARQUES :\n",
    "# L'utilisation du classifieur RF semble approprié pour faire une prédiction sur les données que l'on possède.\n",
    "# L'entrainement du classifieur nécessite 15 secondes pour près de 20 000 données semble tout à fait raisonnable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhUAAAGHCAYAAAAHoqCrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAIABJREFUeJzs3XmclXP/x/HXR0qlZAnJVraIROFGskVJdsIUIm5SfpF9\nvS23fSncSiFJNLZut5CiRrZsTSqkLBVatJBM2pvv74/vNTodZ2bOzJwz13XOeT8fj+txmutc13U+\n5+rMXJ/zXT6XOecQERERqaqNwg5AREREsoOSChEREUkJJRUiIiKSEkoqREREJCWUVIiIiEhKKKkQ\nERGRlFBSISIiIimhpEJERERSQkmFiIiIpISSChHJOWZ2hJkVm9nhMeueMbNZKXyN84PX2ClVx6zA\naxeb2aPV/boiSiokcsysW/BHsWRZY2ZzzGyImTUuY79zzew9M1tiZn+a2VQzu8XM6paxz6lmNsrM\nFpnZKjOba2YvmtlRScZ6kpkVmtkKM/vRzG4zsxqVeI/xS17Mtm3N7DUz+yl4nflm9paZHZpMjMEx\nTjSz8Wa2IDg3PwTvs0Oyx8hC8fcocEBxRQ9iZjeY2cmlHD9r74MQk5iVLGuDz9fLZrZngu2HlPJZ\nX2dm7cN4D5J6G4cdgEgpHHALMBuoDRwMXAC0MbN9nHOrSzY0s42AfKAz8D5wK7AcaBv8u7OZtXPO\nLYp9ATMbAnQDJgEPAb8A2wGnAmPNrI1z7pPSAjSzjsCrQAFwGdACuBnYGuhVzvt7DzgnwforgX2B\ncTHr9gDWAY8HMW4R7Pu+mR3vnHu7rBcys6uB+4HxwN34c7MbcAxwFjCmnFhzxUVU7ovWjcDLwGtx\n658F8mM/q1nqYWAiUBP/2b0UOCL4PV0Yt+1K4ELA4tZPSXuUUj2cc1q0RGrBX+jXAa3i1t8TrD8j\nbv0N+G+Y9yY4VidgLfBm3Pqrg30eLCWGrsAB5cT5NVAIbBSz7t/B6+1RifddG1gKvJXEtnWA+cCo\ncrarAfxe2jGBhmH/f1fwHNVN0XGOCD5Lh6fgWEXA02Gfm7iYioFH0/waRwSvc1rc+kuCc3t13Poh\nwB9hnxst6V3U/SGZ5AP8N5xdS1aYWW18gjAd/41xA865N4GhwHFmdlDMPtcD04BrEr2Qc+5559zE\n0gIxs72AvYAnnHOxTeYD8N92z6jQO/NOAuoDz5e3oXNuBbAI2LycTRsCmwETSjnO4tifzayWmd1u\nZt+Z2cqgy+U+M6sVs03JWIHz4/a9MVh/XFkBmdlsMxtpZsea2RdBl87XZnZq3HYlXUSHm9kAM1sA\n/BzzfGMze9rMfgli/crMLkjwetub2f/MbFnQPN8X2IS4b8uJxlSYd3nQlbbCzBYGXU+tgueLgbpA\nyTkpNrOn487TTnHH7BnEujLobnvMzBrEbTM+eM29zOzdoMtqjpkl/LyWca67mNn0IPaJZtY25rkj\ng/j+1nUT7FdsZv+oyOsF/vZ7KrlDSYVkkqbB45KYdYfhuwOGx13cYz2L/yN3Qsw+Wwb7VLbPe398\nF01h7Ern3HxgTvB8RXXFd028muhJM6tvZluZWTMzuxvYGxhbzjEXAiuAE81si7I2NDMDXsd3wbyG\n79J5FegDvFCynXPuGeANoK+ZbR/s2wL4F/Ckc250OTE5fJfOC8AofIK3BnjZzNol2H4AsCdwO3Bv\n8HrbAJ8CRwOPAr2B74DBZtY75j3VxndPHRtsdyf+//9+Eo+piF/3NNAP+BG4Ft9atgLfHQe+G2o1\nvtvtnGAZVNrxzOw24DH8Z+RK4BX8N/sxtuFYHIf/jL4FfBFs+w1wbwXGwRwZxD4M35W4JfCWmTUH\ncM6NxydpXRPs2xX43jn3aZKvFSvR7+lfgs9w7LJZJV5DoirsphItWuIX1nd/HAVsBWwPnA4sAP4E\nGsds2zvY9qQyjrc5vpn25eDn/ytvnyRivCo4xvYJnvsU+KiCx9sC3988vIxt3greR3Gw7QCgVhLH\nvi2ItQh4E99dtH+C7c7BX9wPiVt/cbD/wTHrtgUWA6PxfemTgJlAvSTimRUc7+SYdfWBucDEuM9B\nMX4siMUd4yn8hXnzuPXDgd+ATYKfLw9e67SYbWoD3xLX/YFvnp8Z8/NRwev3Lef9JOz+iPkc7xT8\n3DD4fxsVt13PYLtuMeveDdZ1iVlXE5gHvJTEOS4O9t8vZt2O+KT1lZh1dwXr6sesa4hPlG4p5zVK\nuj+64X9PGwEdgnO7Fmgdt/2QmM9v7FJQ2d9DLdFb1FIhUWX4wYqL8N+mXgaW4ROBeTHb1Q8ei8o4\nVslzm8U9lrVPeeoEj6sSPLcy5vlkdcZfNMrq+rgO/427O/AxUCvYp0zOuduALvgLf3v8t/VC87NW\nYkfpn4H/Nvxt7DdJ/AXO8BfZkmMuwA9GbY9v7t4X6O6cW5bUu4V5zrm/BjY654rwLUr7B60Qfz2F\nb/2Ib0E4Dd+qUiMu1rfxSWSrYLuOwHzn3H9jXmsl8EQSMZ6Ov+jdkeR7Ks8x+P+vh+PWP4n/LHaK\nW7/MOTe85Afn3BrgM2CXJF9vgnNucsz+P+NboDoErVLgz3ltNuyuOxs/FqfcbrjA0/jf03n4xHcz\n4BznXGGCbVcA7fDnomS5KsnXkQyg2R8SVQ7/De47oAH+Qno4/htUrJLEoD6li088/khin/KsCB43\nSfBc7Zjnk9UV/w271K4D59zUkn+b2fP4JGEIcGZ5B3fOvQi8aGb1gH8A5wevOTJmNs3u+G6GRYkO\nAWyzwQrnXjSzc/AXw0HON6cn6/sE674NHpvgu21KzI7dyMy2xicOF+O7DsqKdedSXmtGEjHugk9+\nfk9i22TsHDx+G7vSObfGzGbGPF9iToJjLMHPMkpGaee4Ln6G0kLn3Awz+xz/WRgSbNMF+MQ5NzPJ\n17kd+BCoh585dTalT6Vd55x7N8njSgZSUiFR9rlzbhKAmb2G/8M13MyaOeeWB9t8g/8WvS8wspTj\n7Bs8Tgsepwf7tChjn/LMDx63wzfbx9oO3wWSFDPbEd/PP9A5ty6ZfYIL0UjgOjPbxDmXqMUk0X7L\n8C1A48xsLXAePsn4AD/G6kv8GIr4KX8QM0gyiHtL4AD8BaR5Mq9fSfEJWkkL63P4QbiJTC1lfSYp\n7bOQ6P+mKp4FHjZfA6YOfrxIzwrs/5VzriD490gz2xR4ysw+dM7F/25IllP3h2QE5wdh3oAfX3FZ\nzFMf4qdMdolp0o3XDX/heyNmnyVAXhn7lGcy/o/7AbErzWw7YAf84LpkdQkek21uLlE3iKGyLS4l\ns1u2Cx5/ALZ0zr3rnCtIsHwXt/8A/LfTG4C2ZnZFBV57twTrmgWPs8vZdxG+1alGKXEWuPWzWn4k\n8SyEvxVnSuAHoLGZlTfDJtnBvj8Gj81iV5pZTfzgxh//tkfV7J5gXTP8GIrY1qgX8N08efjP4mrg\npSq87vX41rqbqnAMyVBKKiRjOOfew/cpX2HBFEfnp1Y+iL9I3B2/j5l1wicVo51zn8Xscx/+2/X9\niV7LzLqa2QGJnguOMQ3f4nFxXGLSE/8HekTMseoEMza2KuVwecBPzrmE0z6D5v74dZvj+/x/cnHT\nQuO2q2NmB5fy9PHBY0lXwEvADmb2zwTHqW0xlUnN7Ax8t8t1zrn78RemO80sUbKQSGOLmUIazAA4\nF/jC/b1g0gaCBHMEcLqZ7Z0g1oYxP44KXuv0mOfrAn97jwmMwP+NvLWc7f6k/Km94GfqrMEPLo51\nEX4cwht/26NqDjGzv2YhBS1iJwFjYseoOOd+xY+FOBffDTLaOfdbZV806DYZgZ9mu01520t2UfeH\nRFVpLQgP4Adtns/6wXb3AvsB15rZIfg/aCvwFTW74otUnZ/gOM2BK82X5H4FX62yEXAKcCBQXhns\na/AD394xsxfw3Sm98AMLY/vsD8IPdryNuEF/ZrYPvnvmbwlRjLfMbA6+S2Uhvu/9fHwLQ3njKeoC\nE8zsE/x4jZ/xF8BT8F0urzrnSqoZDguO93hwTj7CD9jbCz+QtD0wKbhQPA6Mc84NCPa9DD+QcyjQ\nppyYwPftP2VmB+Jn9VyIHwfRLW670j4H1+OnTH5qZk/iu7a2BFrjp5mWJBZPBrENC5LE+fiL55/l\nBeicG29mw4DeZrYH/vxthP9cFcS890LgGDPrgx+sOKskgY073mIzuwf4l5mNxne97YmvQPkZFW+p\nKs9XwGgz+w++9eFSfKvKbQm2fRb/O+DwVWGr6gH8Z+kKEtSPkSwW9vQTLVriF0qpqBk8Z/jBm9/y\n92mG5+HrBSzBXzSm4ptg65TxWqfiv6Utws/kmIv/xp5UpUX8N79CfJPyj/g/2DXitimp3vi3KXr4\nZGIdsHcZr3Epvqz3giDGX/D1Iw5NIr4a+EGuI/BTPpfjuw4m4sdObJxg+6uDc7ccP230s+A81gu2\neSU4xzvE7XsiCSopJohpFv6Cegy+G2k5PvE7NdnPQfB8Q3ztidn4GTdz8bM/usdtt0NwvoqCc/gQ\nfhZNoimlPyT4vF0ZxLciOPdvsOFUzT3wSeOy4JhPx8W/U4L/z6+DmOcB/wE2i9vmXWBKgvf8txhL\nOTfrgEfwrWAzgnP8OdC2lO1rAr/iBwuXO0057nN9WinPFwSfk/oxsS+t7N8FLZmxWPCfLSJSLcxX\nrfzSOXdS2LGIFxTemge85py7OOx4JHNpTIWIiJyKb/l5NuxAJLNpTIWISI4yfz+clvhxFJOccx+G\nHJJkOLVUiEh1S3SPDQnHpUB//FiR+EGyIhWmMRUiIiKSEmqpEBERkZTImTEVQeGhDqyffiYiIiLJ\nqY2/L88Y5wumJZQzSQU+oUh1cRkREZFc0hUYXtqTuZRUzAZ47rnn2GuvvUIOJTv06dOHfv36hR1G\nVtE5TS2dz9TTOU2tTDmf33zzDeeccw6Uc2+eXEoqVgLstddetGrVKuxYskKDBg10LlNM5zS1dD5T\nT+c0tTLwfJY5fEADNUVERCQllFSIiIhISiipEBERkZRQUiGVlpeXF3YIWUfnNLV0PlNP5zS1su18\n5kxFTTNrBRQWFhZm2qAYERGRUE2aNInWrVsDtHbOTSptO7VUiIiISEooqRAREZGUUFIhIiIiKaGk\nQkRERFJCSYWIiIikhJIKERERSQklFSIiIpISSipEREQkJZRUiIiISEooqRAREZGUUFIhIiIiKaGk\nQkRERFJCSYWIiIikhJIKERERSYlIJBVm1tbMRprZXDMrNrOTktjnSDMrNLOVZvatmXWrjlhFREQk\nsUgkFcCmwGSgJ+DK29jMmgBvAOOAlsAjwFNmdmz6QhQREZGybBx2AADOudHAaAAzsyR2uRSY6Zy7\nNvh5hpkdBvQB3klPlCIiIlKWqLRUVNTBwNi4dWOAQ0KIRURERMjcpKIRsCBu3QJgMzPbJIR4RERE\nKmzFirAjSK1IdH+IiIhkm7VrYf58+Okn+PnnuMcf1/HTjBX8urIey5dDnTphR5samZpU/AJsG7du\nW+AP59yqsnbs06cPDRo02GBdXl4eeXl5qY1QRESylnPw66+lJAw/+2XuXCguXr9Pgwaw446w06aL\nOeiHMZy+7nt26nEsZoeG90YSyM/PJz8/f4N1S5cuTWpfc67cyRbVysyKgVOccyPL2OZeoKNzrmXM\nuuHA5s6540vZpxVQWFhYSKtWrVIdtoiIZJE//0ycLMQmDbFdF7Vq+YRhxx1hp50SP25mRXDttTBw\nIBx5JAweDLvsEtp7rIhJkybRunVrgNbOuUmlbReJlgoz2xTYDSiZ+bGLmbUEfnPO/Wxm9wCNnXMl\ntSgGAr3M7D7gaaAdcAaQMKEQEREpsWYNzJuXOFEo+fdvv63f3gy22259gtCy5d+Thq23ho3KGqVY\nUADdu8PixfDYY3DppeXskJkikVQABwDv4mtUOOChYP1QoDt+YOaOJRs752abWSegH9AbmANc6JyL\nnxEiIiI5aOlS+OYb+PprmD7dJwslCcP8+Rt2S2y++foE4dBD/54wNG7sWyIq7eGHoU8f3zpRUJAx\nrROVEYmkwjn3HmXMRHHOXZBg3ftA63TGJSIi0fb77zBtml++/nr949y5/nkzaNoUmjSBZs3g2GM3\nTBp23BHq109zkB07Qs2aWds6ESsSSYWIiEhZlixJnDzMm+ef32gj2HVX2Htv6NbNPzZv7hOJ0GdW\nNGvmlxygpEJERCJjyZINk4aSx/nz/fMbbQS77eaThgsu2DB5qF073NhFSYWIiITg11//njhMmwa/\n/OKfr1FjffJw4YUbJg+bqMRhZCmpEBGRtFm8OHHysCCoiVyjBuyxh08YLr7YPzZv7tdlTPJQUABf\nfAFXXRV2JKFTUiEiIlW2aFHi5GHhQv/8xhuvTx569PCPe+8Nu+9exZkVYSqKqTvRrh1ccYXPknKY\nkgoREUmKcz55SDTmYfFiv03NmuuTh5491ycPu+2WwclDIjlSd6KilFSIiMgGnPMtDImSh19/9dvU\nrOnHN+y9Nxx99IbJQ82a4cafVkVxVTGzvO5ERSmpEBHJUc75gZGJui1KKkrWqrU+eTjmmPUDJnfd\nNcuTh0Q+/hjy8tQ6UQYlFSIiWc45PyUzUfKwZInfplYt2HNPnzR06LBh8rCxrhRevXo+w1LrRKn0\nURERyTLO+RLV77zjl48+8pUnwc+o2GsvnzB07Lg+edhlFyUP5WrRAsaMCTuKSNNHSEQkCyxcCGPH\nrk8k5s71CcRhh/mZji1a+ASiadOcn6AgaaSkQkQkA61YAR9+6BOIt9+GKVP8+n33hbPPhvbtfUJR\nt264cUpuUVIhIpIBioth6tT1ScQHH8CqVf6W3MceC1df7QdSNmoUdqQZbNkyeO896NQp7EgylpIK\nEZGImjNnfXfG2LG+RkTdunDEEXDPPb41onlzfydOqaKCAl8P/PffYdYsfz90qTAlFSIiEbFsGYwf\nvz6R+OYbnzC0bg0XXeSTiEMOyaDy1Zlg2TJfd+Lxx33dicGDlVBUgZIKEZGQrFsHhYW+O+Odd3wZ\nhDVrYOedfZfG7bf7wlJbbRV2pFmqpHVi0SLVnUgRJRUiItVo5sz1LREFBb5OxGabwVFHwcMP+2Ri\nt93UpZFW8a0T48ap7kSKKKkQEUmj33/3yUNJIvHDD35K5z/+Ab17+y6Ngw5SjYhq9e23MHy4WifS\nQB9jEZEUWrMGPvlkfRLx2Wd+5sbuu/tKle3b+y/HDRqEHWkOa9UKfvrJNxFJSimpEBGpAudgxoz1\nScS77/rW9S239HfDvvBC36Wx885hRyobUEKRFkoqREQqaNEi3w1fkkj8/LO/uVabNnDjjT6J2H9/\nVa6U3KOkQkSkHCtX+vtnlBSe+uILv37vveGMM3wScfjhsOmm4cYpMebMgR12CDuKnKOkQkQkjnPw\n5ZcbVq9csQK23dYnEFdc4atXNm4cdqTyN0VFfmbHkCHw1Vd+Ko1UGyUVIiLAvHkb3pBrwQKoXdtX\nr/z3v30y0aKFpnpGWkEBdO8OixfDQw9pmmgIlFSISE768094//31hae+/tqv339/OP98n0S0aeMT\nC4m4+LoTBQVKKEKipEJEcsK6dX4sREkSMWECrF7tu93bt4ebb/azNbbeOuxIpUJUFTNSlFSISNaa\nPXt9d8a4cfDbb1Cvnq9e+eCDvjWiWTN1aWSs2bN9Rti2rapiRoSSChHJGn/84etElLRGfPed/9J6\n0EHQq5dPIg4+2E//lCzQpInvwzr4YLVORISSChHJaMXF8N578NRTMGIErFoFu+7qE4h77/WtElts\nEXaUkjaHHhp2BBJDSYWIZKR582DoUH+n6h9+gD32gDvu8HUj1AouEg4lFSKSMdauhVGjfKvEqFFQ\nqxZ07uxLEhx2mMZGZKXiYnVtZBD9T4lI5H3/vS9/vdNOcPLJvpXiscdg/nzfWtG2rRKKrFRQAPvs\n4+8qKhlBLRUiEkkrVsB//+tbJcaPh803h65d/ezB/fcPOzpJq5KqmAMH+roTtWqFHZEkSUmFiETK\n5Mk+kXj+efj9d39Nee45OO00qFMn7Ogk7WKrYqruRMZRUiEioVu6FPLzfTJRWAiNGkGPHv7asvvu\nYUcn1SK+dUJVMTOSkgoRCYVz8OGHfvbGSy/5qaCdOsG//gUdO6qWRE4pKoJ991VVzCygpEJEqtWC\nBfDss75V4ttv/ZfRm2+Gbt1g++3Djk5CUb8+XHMNHHecWicynJIKEUm7detgzBjfKjFyJNSoAaef\n7lu6jzhCX0oF6Nkz7AgkBZRUiEjazJ4NTz/t60jMmeNbuPv29bM4ttwy7OhEJNWUVIhISq1aBf/7\nn2+VGDvW38CrSxe46CJo3Vr1JESymRodRSQlvvoK+vTx4yLOPhuWL/etFPPn+26OAw5QQpGziorg\n8sth1qywI5E0U0uFiFRaURG8+KIfdPnpp7D11nDBBb5A1Z57hh2dREJJ3YlFi+Doo6Fp07AjkjRS\nUiEiFeKcTyCeegpeeMG3SHToAK+8AieeqOKHEoitO3HEEao7kSOUVIhIUhYvhmHDfDIxbZq/D8e1\n18L55/t/i/wltnVCdSdyipIKESlVcTGMG+cTiVdf9etOOQX69YN27fzUUJG/FBfDZZfB44+rdSJH\nRSZ1NLNeZjbLzFaY2SdmdmA523c1s8lm9qeZzTOzwWamSWoiKfDzz3DHHf560L69H4R5330wd66v\nftm+vRIKSWCjjXz/12OPKaHIUZFoqTCzs4CHgIuBz4A+wBgz28M5tzjB9m2AocDlwBvA9sAg4Ang\njOqKWySbrF4Nb7zhWyVGj4a6deGss/xU0IMP1swNSdLDD4cdgYQoKi0VfYBBzrlnnXPTgR7AcqB7\nKdsfDMxyzvV3zv3onJuATyoOqp5wRbLH9Om+QvIOO/gql7/9Bk884aeCDh4MhxyihEJEkhN6UmFm\nNYHWwLiSdc45B4wFDillt4+BHc2sY3CMbYHOwJvpjVYkO/z5JwwdCm3bwl57+XoSXbrA1KnwySe+\ndaJ+/bCjFJFME3pSATQEagAL4tYvABol2iFomTgHeNHMVgPzgSXAZWmMUySjOQcTJ/pbim+3nZ+1\nUbu2nxY6d65vtW7RIuwoJfLGjfM110USiEJSUWFm1hx4BLgNaAV0AJriu0BEJMaSJX7c3P77w4EH\n+nETl18OM2fCO+/4cRO1a4cdpUReUZGfGnrMMX52h0gCURiouRhYB2wbt35b4JdS9rke+Mg51zf4\n+Ssz6wl8YGY3OefiWz3+0qdPHxo0aLDBury8PPLy8ioVvEgUFRfDe+/5QZcjRvi7hJ54Itx9ty9U\npZkbUiHjxvkyqYsXr687IVkrPz+f/Pz8DdYtXbo0qX3ND18Il5l9AnzqnLs8+NmAn4BHnXMPJNj+\nFWC1c65LzLpDgA+B7Z1zf0tGzKwVUFhYWEirVq3S9E5EwjVvnh8rMXgw/PAD7LGHvxacdx40StiZ\nKFKG2KqYRx7pP1iaJpqTJk2aROvWrQFaO+cmlbZdFFoqAPoCz5hZIeunlNYFngEws3uAxs65bsH2\nrwNPmFkPYAzQGOiHT0xKa90QyVrjx/tbio8a5csEdO7sbzd+2GGauSGVpKqYUgmRSCqccy+ZWUPg\nDny3x2Sgg3NuUbBJI2DHmO2Hmlk9oBfwIPA7fvbI9dUauEjIliyBq6/2szf22w/+8x/Iy4PNNw87\nMsl4Q4ZAkyYqYiUVEomkAsA5NwAYUMpzFyRY1x/on+64RKLq1VehZ09/Q68nn/TdHGqVkJQZNMiP\n4FXrhFSAPi0iGWbBAjjzTDjtND+bY9o0X1dCCYWkVN26SiikwiLTUiEiZXPO3yX0iitg4419fYkz\nz1QyISLRoTRUJAP8+CN07AjdukGnTr514qyzlFBIFRQV+SmiIimkpEIkwoqLoX9/2Gcf+PprX7hq\n2DBo2DDsyCSjFRT48qm9e4cdiWQZJRUiETVjBhx+OFx2GZx7rk8qOnUKOyrJaCVVMdu1g6ZN4c47\nw45IsoySCpGIWbMG7rkHWraEhQt9ZcwBA2CzzcKOTDJaSevEsGG+7sS4cZoqKimnpEIkQr74Ag46\nCG6+2d+fY8oU31ohUmnxrRNTp0KvXprZIWmhT5VIBKxcCTfc4KeIFhfDp5/CffdBnTphRyYZ7/rr\nfetE//5qnZC005RSkZB9+KEvXDV7Ntx+u7/VQs2aYUclWePWW33Z1aZNw45EcoBaKkRCUlTkB2G2\nbQtbbQWTJ8NNNymhkBTbZhslFFJt1FIhEoLRo+GSS3yZgEce8V3cuh25iGQ6tVSIVKNff/UFrDp2\nhGbN/DTR3r2VUEgVrVsXdgQigJIKkWrhHLz8MjRvDiNH+htAjhnjbwIpUmklMzvOOy/sSEQAJRUi\naTd/vr/515lnwmGH+RLb55+vEttSRbF1Jw491GeuIiFTUiGSJs7B00/DXnvBxx/DK6/AiBGw3XZh\nRyYZrbS6E8pSJQKUVIikwcyZ0L69nyp6yim+deL008OOSjKeqmJKxCmpEEmhdevg4Yf93/3vvvOz\nPJ55BrbcMuzIJOM98YSqYkrk6RMpkiLTpvkxE1de6VsovvoKOnQIOyrJGiec4G8Co9YJiTAlFSJV\ntHo1/PvfsP/+sGQJfPABPPoo1KsXdmSSVRo39mMp1DohEabiVyJV8PnnvlVi2jR/i4Wbb4batcOO\nSkQkHEp5RSph+XJ/j46DD4aNN4aJE+HOO5VQiEhuU1IhUkHjx0PLlr6L4+674bPPYL/9wo5KMl5B\ngb/5i0gGU1IhkqSlS6FHDzjqKF9rYsoUuO4631IhUmmxdSc+/hhWrgw7IpFK059DkSS8+aa/AdjS\npdC/v08uNF5OqqygALp393eW0wdLsoA+vSJlWLQIunb1s/latPA3AOvZU3/3pYriq2J++aU+WJIV\n1FIhkoBz8MIL/g6ixcXw7LNwzjmqhCwp8Pnn0Lmzb5147DFNE5Wsok+ySJw5c+Ckk6BLFzj6aD9d\n9NxzlVBIimyzDey7r6piSlZSS4VIoLgYnnoKrrkGNt0UXn3V37dDJKV23hlGjgw7CpG0UIosAnz/\nve/evuSEaiOyAAAgAElEQVQS3zI9bZoSChGRilJSITlt7Vp48EE/CPPHH2HsWN9asfnmYUcmIpJ5\nlFRIzvrySzj0UF8Z89JL/c/t2oUdlWS8oiI/ylckBympkJyzahXceiu0agV//gkTJkDfvn4chUiV\nFBT4Zq9//hN++SXsaESqnZIKySmff+6TibvvhhtvhEmT/P07RKokvu7ElCnQqFHYUYlUO83+kJzx\n+utw5pmwzz4+mWjRIuyIJCvEVsVU3QnJcfrkS04YNgxOPRWOPx4+/FAJhaRAfOuE6k6IKKmQ7PfI\nI3DeeXD++fDii7DJJmFHJFlh4UIYMcK3TowbB7vsEnZEIqFT94dkLefgttvgjjt8Qav77lNVTEmh\nXXf185Dr1Ak7EpHIUFIhWam42N+3o39/n0xce23YEUlWUkIhsgElFZJ1Vq9e39Xx5JNw0UVhRyQi\nkhs0pkKyyvLlvrz2iBHw0ktKKKSKvv8+7AhEMoqSCskaS5bAscfC++/Dm2/C6aeHHZFkrJKZHc2a\nwRdfhB2NSMZQ94dkhfnzoUMHmDfPlw046KCwI5KMFVt34tFHoWXLsCMSyRhqqZCMN3MmHHYY/Pab\nb6VQQiGVoroTIlWmlgrJaF9+Ce3bQ/368NFHsPPOYUckGUlVMUVSQkmFZKwJE6BTJ/+lcvRo2Gab\nsCOSjLRwof8gHXywTy5UxEqk0pRUSEYaPRpOOw0OPBBGjoQGDcKOSDLWNtvAxx/DvvuqdUKkiiLz\nG2RmvcxslpmtMLNPzOzAcravZWZ3mdlsM1tpZjPN7PxqCldC9MILcOKJfqbH6NFKKCQF9ttPCYVI\nCkTit8jMzgIeAm4F9gemAGPMrGEZu70MHAVcAOwB5AEz0hyqhOzxx6FLF7+MGKGChiIiURKJpALo\nAwxyzj3rnJsO9ACWA90TbWxmxwFtgeOdc+86535yzn3qnPu4+kKW6uQc3Hkn9OwJl18OQ4bAxuq8\nk2StXRt2BCI5Iek/y2b2BeCS2dY516oCx60JtAbujtnfmdlY4JBSdjsRmAhcZ2bnAn8CI4FbnHMr\nk31tyQzFxXDVVfDwwz6xuPFG3RhMKqCgAC6+GIYP13xjkTSryHe9/6UphoZADWBB3PoFQLNS9tkF\n31KxEjglOMbjwJbAhekJU8KwZo0vtT1sGAwY4Gf6iSSlqMjfSW7gQDjySGhYVm+qiKRC0kmFc+72\ndAZSQRsBxUAX59wyADO7EnjZzHo651aVtmOfPn1oEDeyLy8vj7y8vHTGK5WwYgWcdRa89Zb/knn2\n2WFHJBlDdSdEKi0/P5/8/PwN1i1dujSpfaPQK70YWAdsG7d+W+CXUvaZD8wtSSgC3wAG7AD8UNqL\n9evXj1atku6dkZAsXQonnQQTJ8Lrr8Nxx4UdkWSE+NYJ1Z0QqbBEX7QnTZpE69aty923ImMqlpD8\nmIotkz2uc26NmRUC7fDjIjAzC35+tJTdPgLOMLO6zrnlwbpm+NaLOcm+tkTTwoU+iZg1C955Bw49\nNOyIJCOsXg0HHABz56p1QiQkFWmpuCJtUUBf4JkgufgMPxukLvAMgJndAzR2znULth8O3AwMMbPb\ngK2B+4HBZXV9SPTNnu3Lbi9b5u/j0aJF2BFJxqhVC/71LzjkELVOiISkImMqhqYrCOfcS0FNijvw\n3R6TgQ7OuUXBJo2AHWO2/9PMjgX+A3wO/Aq8CNySrhgl/aZN8wlF7drw4Ye6LkgldO0adgQiOa3K\nYyrMrDZQK3adc+6Pih7HOTcAGFDKcxckWPct0KGiryPR9Nln0LEj7LADjBkDjRqFHZGIiFRUpToc\nzWxTM3vMzBbia0QsiVtEkjZ2LBx9NOy1F7z3nhIKEZFMVdlRTPcDRwOXAquAi/AltucB56UmNMkF\nI0b4G0Qefji8/TZsvnnYEUlkFRVBr14weXLYkYhIKSqbVJwI9HTOjQDWAh845+4EbgTUqSlJefJJ\nOPNMOOMMeO01qFs37IgksgoK/KjdoUPhu+/CjkZESlHZpGJLYGbw7z+CnwE+BA6valCS/e67z1dO\nvvRSXy2zZs2wI5JIKiryH5J27aBpU5g6FTp3DjsqESlFZZOKmUDT4N/TgTODf58I/F7VoCR7Oedr\nE11/Pdx6K/znPyolIKUoaZ0YNszXnRg3TlOCRCKusrM/hgAtgfeAe4HXzewyoCZwZYpikyyzdi1c\ncgk8/TQ88gj07h12RBJJzvkPx2OPqSqmSIapVFLhnOsX8++xZrYn/k6j3zvnpqYqOMkeK1f6EgKv\nvQbPPadyAlIGMz8FSFUxRTJOSu794Zz7EfgxFceS7FNUBKecAhMmwP/+ByecEHZEEnk33RR2BCJS\nCZWtU/Fo0N0Rv/4yM3u46mFJtli82NegmDjRF7VSQiEikr0q2654On6mR7wJwBmVD0eyyc8/Q9u2\n8NNPvqjV4ZoXJCKS1SqbVGwFFCVY/wfQsPLhSLaYMQPatIEVK/x9PPbbL+yIJFIKCvzNXkQkq1Q2\nqfge6JhgfUfW16+QHDVpEhx2GGy2GXz0Eey+e9gRSWTE1p0YNCjsaEQkxSo7ULMv8JiZbQ0UBOva\nAVeR3lukS8SNHw8nnQTNm8OoUbDlluXuIrmioAC6d/cDbUpmdohIVqnslNKnzWwT4CbW3258NnCp\nc+7ZFMUmGea11+Css/zYif/+F+rVCzsiiYRly3zFs8cfV90JkSxX6SmlzrnHgceD1ooVzrllqQtL\nMs3QoXDhhXDaab4A4iabhB2RREJBgf9gLFqkuhMiOaDSv91mtrGZHQOcBliwrrGZ6ftpjunbF84/\n31878vOVUEiMN9+EJk38PTt69VJCIZLlKtVSYWY7A6OBnYBNgHfws0GuC37ukaoAJbqcg5tvhrvv\nhhtugLvu8sUQRf5yzz2w8cZKJkRyRGV/0x8BJgJbACti1r+KH7ApOaAkoXjwQf+ohEL+plYtJRQi\nOaSyYyraAoc651bbhleS2cD2VQ1Kou+///WJxP33w1VXhR2NiIhEQWW/QmwE1EiwfgcSF8WSLDJ9\nuh9D0bkzXH112NFIqIqKYO7csKMQkYiobFLxNhvWo3DBAM3bgVFVjkoia9kyP8Njhx1g8GB1eeS0\nggJo0cKP0BURofJJxVVAGzObBtQGhuO7PnbAD9aULOScr100Z47v/qhfP+yIJBSxVTGbNoUBA8KO\nSEQiorLFr+aYWUvgbGBfoB4wGHg+hbFJxPTrBy+/DCNGwJ57hh2NhCJRVUwNxBSRQFWKX60Fniv5\nOaiw2Qu4FmhU9dAkSsaP90URr73Wd39Ijikq8v/5AweqKqaIlKpCXzHMbBMzu8fMJprZBDM7JVh/\nATAL6AP0S0OcEqK5c9eX377rrrCjkVDcd58vlfrYYzBunBIKEUmooi0VdwCX4ItdtQFeNrMhwMHA\nlcDLzrl1qQ1RwrR6tZ/lUasWvPCCr2MkOej66323h5IJESlDRS8RnYHznHMjzWwfYGpwjJbOOZfy\n6CR0V14JhYXw/vuwzTZhRyOhqVdPd4gTkXJVNKnYASgEcM59ZWargH5KKLLTsGHQv7/vRv/HP8KO\nRkREoq6iw7ZrAKtjfl4L6O6kWWjKFLjkEl/k6uKLw45GqsWaNWFHICIZrqItFQY8E7RQgK9RMdDM\n/ozdyDmn+QEZbMkSP8Njzz19CQIVuMpyy5b5mR3ffQdvv63/cBGptIomFUPjfn4u4VaSsYqL4dxz\nfWIxdizUqRN2RJJWBQW+IuaiRX6Gh3NKKkSk0iqUVDjnLkhXIBINd94Jo0bBm2/6YomSpUpaJx5/\n3Ned0DRREUkBTRCUv7z1Ftx2m186dgw7Gkmb2NYJVcUUkRTSXxIBYNYs6NoVjj8ebr457GgkbZ5/\n3t+zo0kTmDoVevVSQiEiKaO/JsKKFX5g5hZb+GmkusZksRNOgCeeUHeHiKSFuj9ynHO+9XvGDPj4\nY59YSBZr0AD++c+woxCRLKWkIscNGgRDh/oWipYtw45GREQymRq6c9inn0Lv3r5b/Zxzwo5GREQy\nnZKKHLVwIZxxBhxwAPTtG3Y0kjLjxkGPHr5fS0SkmimpyEFr18LZZ/s7kL78sr8DqWS4oiI/OOaY\nY/wAmaKisCMSkRykMRU56Kab/F1Hx42D7bcPOxqpsnHjfN2JxYtVd0JEQqW/PDlmxAi4/36/HHFE\n2NFIlcS2TjRtqroTIhI6tVTkkOnT/V1HzzwT+vQJOxqpkilT4OSTVRVTRCJFf4VyRFGRL3C1444w\neLDuGZXxdtgBDjoIvvxSrRMiEhlqqcgBzkH37jBnDnz2GdSrF3ZEUmVbbQUvvRR2FCIiG4jM1xsz\n62Vms8xshZl9YmYHJrlfGzNbY2aT0h1jpurbF155BZ55BvbcM+xoREQkW0UiqTCzs4CHgFuB/YEp\nwBgza1jOfg2AocDYtAeZocaPh+uu88tpp4UdjYiIZLNIJBVAH2CQc+5Z59x0oAewHOhezn4DgeeB\nT9IcX0aaMwfOOsvP8rjzzrCjkQopKvI11FXESkQySOhJhZnVBFoD40rWOeccvvXhkDL2uwBoCtye\n7hgz0erV0LmzL2yVnw8ba/RM5hg3Dlq0gKuugpkzw45GRCRpoScVQEOgBrAgbv0CoFGiHcxsd+Bu\noKtzrji94WWmK6+ESZN8XYpttgk7GklKoroTu+4adlQiIknLuO+vZrYRvsvjVufcDyWrQwwpcoYN\ng/79YeBAP+tQMoCqYopIFohCUrEYWAdsG7d+W+CXBNvXBw4A9jOz/sG6jQAzs9VAe+fc+NJerE+f\nPjRo0GCDdXl5eeTl5VUu+oiZPBkuvhguuMA/SsQVFcG11/oM8MgjoaAAdtkl7KhEJIfl5+eTn5+/\nwbqlS5cmta+5CAwEM7NPgE+dc5cHPxvwE/Coc+6BuG0N2CvuEL2Ao4DTgdnOuRUJXqMVUFhYWEir\nVq3S8C7Ct2SJv+togwbw0UdQp07YEUm5FiyAf/wDrrlGrRMiElmTJk2idevWAK2dc6WWcIhCSwVA\nX+AZMysEPsPPBqkLPANgZvcAjZ1z3YJBnNNidzazhcBK59w31Rp1hBQXwznn+MRi7FglFBlj223h\n2291q1gRyQqRSCqccy8FNSnuwHd7TAY6OOcWBZs0AnYMK75McOed8NZbMGqUH+MnGUQJhYhkiUgk\nFQDOuQHAgFKeu6CcfW8nh6eWvvUW3HYb3H47HHdc2NGIiEiuUgduhps5E7p0gU6d4Kabwo5GEvrq\nq7AjEBGpFkoqMtjy5XD66f7eUsOGaYxf5JTUnWjRAt59N+xoRETSLjLdH1Ixzvnr1YwZ8MknsPnm\nYUckGygo8LeGXbTI15044oiwIxIRSTt9t81QgwbBs8/CE0/AvvuGHY38paR1ol07aNIEvvwSevVS\nM5KI5AS1VGSgTz6B3r3hssv8NFKJiPjWCdWdEJEco6QiwyxcCGecAQceCA89FHY08pc//vD/Mfvu\nq6qYIpKzlFRkkLVr4eyz/ePLL6u8QaRsthlMmAB77KHWCRHJWUoqMsiNN8L77/t7TzVuHHY08jd7\n7hl2BCIioVJSkSHefhseeMB3eWgigYiIRJHaaTPAypV+AsGRR0KfPmFHk8NWrQo7AhGRSFNSkQEe\neABmz4b+/cEs7GhyVEGB7954662wIxERiSwlFRE3axbcfTdceSU0bx52NDkovu5Es2ZhRyQiElka\nUxFxl18ODRvCLbeEHUkOKqk7sXixbybq0UMzO0REyqCkIsJef90vr7wC9eqFHU0OKSqCa6+FgQP9\nQJZ339X95EVEkqCkIqKWL/dVMzt0gNNOCzuaHLJuHbRp42//qqqYIiIVoqQiou65B+bN81NJNTiz\nGtWoAXfdBXvvraqYIiIVpKQigr77Du6/H667DnbfPexoctCJJ4YdgYhIRlK7bsQ4528U1rgx3HBD\n2NGIiIgkTy0VETNihO/yGDkS6tQJOxoREZHkqaUiQpYtgyuu8K3vaoFPk5K6EwUFYUciIpJ1lFRE\nyB13wK+/wiOPhB1JlioogBYtYNgwmD8/7GhERLKOkoqI+Ppr6NcPbrpJJRFSLrYqZtOm8OWX0LVr\n2FGJiGQdjamIAOf8DcOaNoVrrgk7miyjqpgiItVGSUUEDB8O770HY8bAJpuEHU0Wueoq6NtXVTFF\nRKqJvrKFbOlSf+074wxo3z7saLLMHnv41olx45RQiIhUA7VUhOxf//KzPvr1CzuSLHTJJWFHICKS\nU5RUhGjyZH97iXvvhR12CDsaERGRqlH3R0iKi6FnT9hzT1+bQkREJNOppSIkzzwDH3/sxw/WrBl2\nNBmqoADq1oWDDw47EhERQS0VofjtN3+zsK5d/cQEqaDYuhNPPhl2NCIiElBLRQhuuglWr4YHHww7\nkgwUW3fiscd8ciEiIpGglopq9vnnMGgQ/Pvf0KhR2NFkkPiqmFOn+ophKmQlIhIZaqmoRuvW+cGZ\nLVv6R0lSotYJJRMiIpGjpKIaPfkkTJwIEybAxjrzyfv8c986UVAAu+wSdjQiIlIKfd2rJosWwY03\n+i/chxwSdjQZ5uqrfVVMJRQiIpGm78vV5Lrr/OO994YbR0aqUSPsCEREJAlKKqrBRx/BkCEwcCBs\nvXXY0YiIiKSHuj/SbO1aPyjzwAPhoovCjiaiiorg++/DjkJERKpILRVp1r8/fPklfPaZWvETKpnZ\n0aiRLzFqFnZEIiJSSWqpSKP58+GWW6BHDzjggLCjiZj4uhPDhyuhEBHJcGqpSKOrr4ZNNoG77go7\nkohR3QkRkaykv+Rp8u67/sv3/ffDFluEHU1EqCqmiEhW01/zNFi92l8rDz0UunULO5oIeeIJGDbM\nt06o7oSISNZR90caPPwwzJgBkybpS/gGeveG00+HJk3CjkRERNJAl7wU+/lnuP12+L//8/f4kBg1\nayqhEBHJYkoqUqxPH9hsM59YiIiI5JLIJBVm1svMZpnZCjP7xMwOLGPbU83sbTNbaGZLzWyCmbWv\nzngTGTMGRoyAhx6CBg3CjiYkK1aEHYGIiIQkEkmFmZ0FPATcCuwPTAHGmFnDUnY5HHgb6Ai0At4F\nXjez0DocVq6Eyy6Do46CvLywoghRycyONm1gzZqwoxERkRBEZaBmH2CQc+5ZADPrAXQCugP3x2/s\nnOsTt+omMzsZOBGfkFS7Bx6A2bPhtddysIZTbN2J++5T6VARkRwVekuFmdUEWgPjStY55xwwFkjq\nJuFmZkB94Ld0xFieWbPg7rvhyiuhefMwIgiJ6k6IiEiMKLRUNARqAAvi1i8AmiV5jGuATYGXUhhX\n0i6/HBo29CW5c4aqYoqISJwoJBVVYmZdgFuAk5xzi8vbvk+fPjSIG0WZl5dHXiUHQhQUwOuvwyuv\nQL16lTpE5nntNTjlFDjySH8CVMRKRCRr5Ofnk5+fv8G6pUuXJrWv+Z6G8ATdH8uB051zI2PWPwM0\ncM6dWsa+ZwNPAWc450aX8zqtgMLCwkJatWqVktgBOnWCOXNg8uQcGkuxahW88AKce65aJ0REcsCk\nSZNo3bo1QGvn3KTStgv9iuCcWwMUAu1K1gVjJNoBE0rbz8zygMHA2eUlFOkyfTqMGuVrU+RMQgH+\nLmnduimhEBGRDUSl+6Mv8IyZFQKf4WeD1AWeATCze4DGzrluwc9dgud6A5+b2bbBcVY45/6orqAf\neQS23TZHp5CKiIjEicRXTefcS8DVwB3AF8C+QAfn3KJgk0bAjjG7/BM/uLM/MC9mebi6Yv71Vxg6\nFHr29F/cRUREcl1UWipwzg0ABpTy3AVxPx9VLUGV4cknobgYevQIO5I0KCiAQYPg+edh48h8RERE\nJOIi0VKRadas8bMozzkHttkm7GhSKLbuxMKFkORoXxEREVBSUSkvvwxz5/r6FFmjoABatIBhw3zG\nNG4cbLVV2FGJiEgGUVJRQc5Bv35wzDH+GpzxVBVTRERSRB3mFfTRRzBxIrz5ZtiRpMA330DHjqqK\nKSIiKaErSAX16wfNmsFxx4UdSQrsvLO/rapaJ0REJAXUUlEBs2bB//4H/ftnyfW3bl0YMiTsKERE\nJEtkw6Wx2jz6KGy+OZx3XtiRiIiIRI+SiiT98QcMHgyXXOK/4IuIiMiGlFQkafBgWLHCDz3IGEVF\n8NBDvkqXiIhImimpSMK6db7r46yzYPvtw44mSSV1J269Fb76KuxoREQkByipSMJrr8Hs2f5upJGX\nqO7EvvuGHZWIiOQAzf5IQr9+cNhh4G8lH2EFBdC9u+pOiIhIKJRUlGPiRPjwQxgxIuxIyrBsGVxz\nDQwcCEce6ZOLXXYJOyoREckx+hpbjn79fC/CySeHHUkZnIMPPlh/zw4lFCIiEgK1VJRh7lx46SV4\n4AGoUSPsaMpQvz5MmRLxIEVEJNuppaIMjz0Gder4YQqRp4RCRERCpqSiFH/+CYMGwUUXwWabhR2N\niIhI9CmpKMWzz8LSpdC7d9iRBCZO9GMnREREIkpJRQLFxfDII3DqqdCkScjBLFsGPXvCgQf6u5mJ\niIhElAZqJjB6NMyY4Utzh6qgAC68EBYt8gM8Ij0FRUREcp1aKhLo1883DBx6aEgBlLROtGvnm0qm\nTvU3HVEhKxERiTC1VMT58ksYOxaefx7MQgggvnVCVTFFRCRDKKmI8/DD/qZhnTuH8OKrVkG3brDb\nbipiJSIiGUdJRYyFC30LxW23Qc2aIQSwySa+MuZOO6l1QkREMo6SihiPP+5rSF18cYhBhD7dRERE\npHL0dTiwciUMGOB7H7bcMuxoREREMo+SikB+vu/+uPzyNL/Q8uUqYiUiIllJSUWgf384/nho1iyN\nL1JQAM2b+3KdIiIiWUZJBTB9OhQW+pmcaVFU5KeGtmvn76Petm2aXkhERCQ8GqgJDB8ODRr4loqU\nKyjwtzldvFh1J0REJKvl/NXNOZ9UnH461K6dwgPHVsVs2lRVMUVEJOvlfEvF55/DDz/425ynjHNw\n7LG+PKdaJ0REJEfkfFLx/POw3XZw5JEpPKgZ3Hsv7LijqmKKiEjOyOmkYu1aePFF6NLFF71KqSOO\nSPEBRUREoi2n2+TffRcWLPBJhYiIiFRNTicVzz8Pu+8OrVuHHYmIiEjmy9mkYsUK+O9/oWvXStzi\nvKjIz+x48cW0xCYiIpKJcjapeOMNnxtUuOujoABatPBVMZcvT0tsIiIimShnk4rhw+HAA333R1Li\nq2JOnQoXXJDWGEVERDJJTiYVS5bAqFEVaKUYN863Tgwb5utOjBunqaIiIiJxcjKpGDHCTyc966wk\nNr7xRjjmGFXFFBERKUdOXh2HD4ejj/ZFr8q1//5qnRAREUlCzhW/WrgQxo+HwYOT3KFz53SGIyIi\nkjVyrqVizBioVQtOOy3sSERERLJLziUVo0fDCSf4W52LiIhI6kQmqTCzXmY2y8xWmNknZnZgOdsf\naWaFZrbSzL41s27JvM706XDOOTErCgrg7berFHuuys/PDzuErKNzmlo6n6mnc5pa2XY+I5FUmNlZ\nwEPArcD+wBRgjJk1LGX7JsAbwDigJfAI8JSZHVvea225JXTqxIZ1J55+OiXvI9dk2y9DFOicppbO\nZ+rpnKZWtp3PSCQVQB9gkHPuWefcdKAHsBzoXsr2lwIznXPXOudmOOf6A68ExynTCSdAzQ8KNqw7\nMXx4qt6HiIhIzgo9qTCzmkBrfKsDAM45B4wFDillt4OD52ONKWP7v5z8y6ANq2Kq7oSIiEhKROFq\n2hCoASyIW78AaFTKPo1K2X4zM9ukrBdr8v6zqjshIiKSBrlUp6I2wDd33w2HHAKTJ4cdT8ZbunQp\nkyZNCjuMrKJzmlo6n6mnc5pamXI+v/nmm5J/1i5rO/M9DeEJuj+WA6c750bGrH8GaOCcOzXBPu8B\nhc65K2PWnQ/0c85tUcrrdAGeT230IiIiOaWrc67UgYiht1Q459aYWSHQDhgJYGYW/PxoKbt9DHSM\nW9c+WF+aMUBXYDawsgohi4iI5JraQBP8tbRUobdUAJjZmcAz+Fkfn+FncZwB7OmcW2Rm9wCNnXPd\ngu2bAF8CA4Cn8QnIw8Dxzrn4AZwiIiJSDUJvqQBwzr0U1KS4A9gWmAx0cM4tCjZpBOwYs/1sM+sE\n9AN6A3OAC5VQiIiIhCcSLRUiIiKS+aIwpVRERESygJIKERERSYmsSSqq64ZkuaQi59TMTjWzt81s\noZktNbMJZta+OuPNBBX9nMbs18bM1phZ9Ce0V6NK/N7XMrO7zGx28Ls/M5iOLoFKnNOuZjbZzP40\ns3lmNtjMtqyueKPMzNqa2Ugzm2tmxWZ2UhL7ZPS1KSuSiuq8IVmuqOg5BQ4H3sZP9W0FvAu8bmYt\nqyHcjFCJc1qyXwNgKH8vTZ/TKnk+XwaOAi4A9gDygBlpDjVjVOJvaRv8Z/NJoDl+1t5BwBPVEnD0\nbYqfeNATKHcAY1Zcm5xzGb8AnwCPxPxs+Bkh15ay/X3A1Lh1+cCosN9LVJaKntNSjvEVcHPY7yUq\nS2XPafDZvB3/h35S2O8jKkslfu+PA34DNg879qgulTinVwHfxa27DPgp7PcStQUoBk4qZ5uMvzZl\nfEtFdd+QLBdU8pzGH8OA+vg/4jmvsufUzC4AmuKTCglU8nyeCEwErjOzOWY2w8weMLMyyw7nikqe\n04+BHc2sY3CMbYHOwJvpjTZrZfy1KeOTCqr5hmQ5ojLnNN41+Ka/l1IYVyar8Dk1s92Bu/FlcYvT\nG17GqcxndBegLbA3cApwOb65vn+aYsw0FT6nzrkJwDnAi2a2GpgPLMG3VkjFZfy1KRuSComY4D4r\ntwCdnXOLw44nE5nZRvh71dzqnPuhZHWIIWWDjfBN0F2ccxOdc6OBK4FumfIHO2rMrDm+3/82/Fiq\nDviWtUEhhiUhikRFzSpaDKzDV+KMtS3wSyn7/FLK9n8451alNryMVJlzCoCZnY0fpHWGc+7d9ISX\nkYepnQkAAAS2SURBVCp6TusDBwD7mVnJN+mN8D1Lq4H2zrnxaYo1E1TmMzofmOucWxaz7ht8srYD\n8EPCvXJHZc7p9cBHzrm+wc9fmVlP4AMzu8k5F/+tW8qW8demjG+pcM6tAUpuSAZscEOyCaXs9nHs\n9oHybkiWMyp5TjGzPGAwcHbwLVAClTinfwD7APvhR4G3BAYC04N/f5rmkCOtkp/Rj4DGZlY3Zl0z\nfOvFnDSFmjEqeU7rAmvj1hXjZzqoZa3iMv/aFPZI0VQswJn426efB+yJb3r7Fdg6eP4eYGjM9k2A\nIvxI22b46T6rgWPCfi9RWSpxTrsE57AHPrMuWTYL+71EZanoOU2wv2Z/VOF84sf4/Ai8COyFnwY9\nAxgY9nuJylKJc9oNWBX83jcF2uBvCjkh7PcShSX4zLXEfzkoBq4Ift6xlPOZ8dem0ANI4X9eT/xt\nzVfgs7oDYp4bAhTEbX84PitfAXwHnBv2e4jaUpFziq9LsS7B8nTY7yNKS0U/p3H7Kqmo4vnE16YY\nAywLEoz7gU3Cfh9RWipxTnvh7xq9DN/iMxTYLuz3EYUFOCJIJhL+XczGa5NuKCYiIiIpkfFjKkRE\nRCQalFSIiIhISiipEBERkZRQUiEiIiIpoaRCREREUkJJhYiIiKSEkgoRERFJCSUVIiIikhJKKkQk\n7cxsVzMrDu5qiZm1C36uW96+IpI5lFSISFLMbEiQCKwLHkv+vUuSh4gv35v2cr5KXkSqVzbc+lxE\nqs9bwPlseAfKRUnuG8ZdKw3dMVOk2qilQkQqYpVzbpFzbmHM4szseDP70MyWmNliMxtpZk2r+mJm\n1svMfjCzlWY2zczyYp7boEslWLdVsO5QM9sVeDt4qihoVXmiqjGJSOmUVIhIKtQBHgBaAe3wLQMj\nqnJAM+sMPIS/PfQ+wNPAMDNrE7NZoi6UknUz8bfyBn9b7u2AK6sSk4iUTd0fIlIRJ5pZUczPo5xz\nZznnNkggzOyfwDwz28M5920lX+sq4Enn3FPBzw+a2SHA1cBHJS+VYD8DCFpQlgTrFjnnllcyDhFJ\nkloqRKQiCoB9gZbB0hvAzHY3sxfMbKaZ/QF8h28x2Km8A5rZRmZWFCx/mNmjwVN7ARPiNv8oWF8i\n7YM9RSR5aqkQkYr40zk3K8H6N4Fvge7AfKAWMCV4LJNzrtjMWsasWppkLMX4VonY1oqaSe4rImmg\nlgoRqRIz2wbYDfi3c268c24GsBUVmELqnJsZs/warP4GaBO3aRtgWvDvklkn28U8v3/c66wOHmsk\n9WZEpErUUiEiVfUrsAS4xMwW4QdF3ptgu4pO63wAeM7MpgDvAqcCJwGHAzjnlpnZROAGM5sDNALu\niDvGj8HjiWb2NrDCuf9v345NEAiCMIz+VmUpJoLYgyCYmQtWYh0mhlZgZAOGMgZnIGLgcRO+Fy/D\nhB/Lbj1G7gH8yU0FMElVPZMsksyTXDPEwObX0ZFzTxkea27fc9dJllV1/ji2yvDz5JLkkGT3NeOW\nZJ/hF8k9yXHMDsA4syrvnACA6dxUAAAtRAUA0EJUAAAtRAUA0EJUAAAtRAUA0EJUAAAtRAUA0EJU\nAAAtRAUA0EJUAAAtRAUA0OIFZZBFgYgBkzEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x23c48630>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Courbe ROC des données de test X et Y selon le classifieur RF\n",
    "# Au mieux, le classifieur à une performance de 0.78 pour la prédiction du sexe\n",
    "draw_roc_curve(rf, X_test, Y_test, \"Sex prediction by RF\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAF4CAYAAACvlrkrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFLUmQ60AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x332b5ba8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Matrice de confusion des données X_test sur lesquelles on fait la prédiction et des données Y_test.\n",
    "# La prédiction du modèle étant de 0.72, au mieux, commence à être correct.\n",
    "# Un grand nombre de Vrais Positifs et Vrais Négatifs sont présents.\n",
    "draw_confusion_matrix(Y_test, rf.predict(X_test), \"\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2.2) SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y.Sex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------ \n",
      " svm_grid_search\n",
      " best_score_ = 0.49 with =  SVC(C=1.0, cache_size=10, class_weight=None, coef0=0.0,\n",
      "  decision_function_shape=None, degree=2, gamma=0.01, kernel='rbf',\n",
      "  max_iter=-1, probability=True, random_state=None, shrinking=True,\n",
      "  tol=0.001, verbose=False) \n",
      "\n",
      "Duration time : 5.44199991226\n"
     ]
    }
   ],
   "source": [
    "#100 rows : 0.82s / 0.5 pred  / 0.57 acc\n",
    "#200 rows : 2.7s  / 0.5 pred  / 0.57 acc\n",
    "#300 rows : 6.1s  / 0.42 pred / 0.57 acc\n",
    "#400 rows : 11 s  / 0.49 pred / 0.57 acc\n",
    "#500 rows : 21s   / 0.50 pred / 0.57 acc\n",
    "#600 rows : 25s   / 0.52 pred / 0.58 acc\n",
    "nb_ligne = 300\n",
    "start_time = print_time()\n",
    "\n",
    "# Apprentissage du classifieur SVM sur X_train et Y_train (<nb_lignes> données)\n",
    "svm = svm_grid_search(X_train[:nb_ligne], Y_train[:nb_ligne])\n",
    "\n",
    "end_time = print_time()\n",
    "print \"\\nDuration time :\", end_time - start_time\n",
    "\n",
    "# REMARQUES :\n",
    "# L'utilisation du classifieur SVM ne semble toujours pas très approprié pour faire des prédictions sur nos données.\n",
    "# D'autre part, le temps nécessaire pour faire un apprentissage pour entrainer le classifieur svm est toujours trop important\n",
    "# au vue du nombre de données que l'on lui fournit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhUAAAGHCAYAAAAHoqCrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noEKSJLXCoEKSJLXCoEKSJLXif7NzVyn3fqx7AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x18b72550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Courbe ROC des données de test X et Y selon le classifieur SVM \n",
    "# Au mieux, le classifieur à une performance de 0.52 pour la prédiction du sexe\n",
    "# Nous pouvons dire que la classification se fait quasiment de manière aléatoire puisque la courbe se rapproche de la droite diagonale\n",
    "draw_roc_curve(svm, X_test, Y_test, \"Sex prediction by SVM\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAF4CAYAAACvlrkrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nA/EfvxxaOqVGAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a07c2b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Matrice de confusion des données X_test sur lesquelles on fait la prédiction et des données Y_test.\n",
    "# La prédiction du modèle étant de 0.58, au mieux, n'est pas satisfaisant.\n",
    "# Un nombre trop important de Faux Positif est présent.\n",
    "# Le classifieur SVM ne semble toujours pas adapté pour effectuer une prédiction sur nos données.\n",
    "draw_confusion_matrix(Y_test, svm.predict(X_test), \"\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2.3) KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y.Sex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------ \n",
      " knn_grid_search\n",
      "{'n_neighbors': [2, 3, 5, 8, 10], 'weights': ['uniform', 'distance'], 'leaf_size': [1, 3], 'algorithm': ['auto', 'ball_tree', 'kd_tree', 'brute']}\n",
      " best_score_ = 0.76 with =  KNeighborsClassifier(algorithm='auto', leaf_size=1, metric='minkowski',\n",
      "           metric_params=None, n_jobs=1, n_neighbors=5, p=2,\n",
      "           weights='uniform') \n",
      "\n",
      "Duration time : 4.15100002289\n"
     ]
    }
   ],
   "source": [
    "#15 rows  : 2.7s   / 0.5  pred / 0.52 acc\n",
    "#30 rows  : 4.5s   / 0.51 pred / 0.53 acc\n",
    "#50 rows  : 9.8s   / 0.51 pred / 0.53 acc\n",
    "#75 rows  : 16s    / 0.53 pred / 0.55 acc\n",
    "#200 rows : 102s   / 0.63 pred / 0.60 acc\n",
    "#400 rows : 289s   / 0.66 pred / 0.62 acc\n",
    "#1000 rows : 1387s / 0.69 pred / 0.68\n",
    "nb_ligne = 30\n",
    "start_time = print_time()\n",
    "\n",
    "# Apprentissage du classifieur KNN sur X_train et Y_train (<nb_lignes> données)\n",
    "knn = knn_grid_search(X_train[:nb_ligne], Y_train[:nb_ligne])\n",
    "\n",
    "end_time = print_time()\n",
    "print \"\\nDuration time :\", end_time - start_time\n",
    "\n",
    "# REMARQUES :\n",
    "# L'utilisation du classifieur KNN ne semble pas très adapté pour effectuer un apprentissage sur nos données.\n",
    "# Bien que la prédiction du modèle semble à peu près correct, le temps nécessaire pour un apprentissage sur un nombre\n",
    "# aussi réduit de données n'est pas viable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhUAAAGHCAYAAAAHoqCrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAIABJREFUeJzs3Xd8VFX6x/HPI6sgFuywtgV3VyyLBWxYcUERe9eASHEt\ngMpid9XV9aei7io2FLFR1Ci2FRugRFcUsAQQC4oFXBVFsEDoJef3x3NnHcZJMkmm5/t+veaV5M69\n95y5meQ+c8pzLISAiIiISH2tlesKiIiISHFQUCEiIiJpoaBCRERE0kJBhYiIiKSFggoRERFJCwUV\nIiIikhYKKkRERCQtFFSIiIhIWiioEBERkbRQUCEiRcfMDjKzSjM7MG7bMDOblcYyekZlbJuuc9ai\n7EozuyPb5YrUREGFZJ2Z9Yj+KcYeK83sazN7yMy2rOa47mb2HzP7ycwWm9l0M7vKzJpWc8xxZvai\nmc0zs+Vm9o2ZPW5mB6dY16PNrNzMlprZl2Z2jZk1SvHYyioel1SxfyczG29mP5vZQjN718xOSrGs\no8zsNTObG12bz6PX2TmV44tU4hoEAais7UnM7HIzO6aK8xftOgdxgdnxCdvXNrPnzWy1mfWMtsX+\nppeY2W+TnOs1M5uesG12dMztqZYt+e83ua6ANFgBuAqYDTQB9gF6AfuZ2Z9CCCtiO5rZWkApcBLw\nOnA1sAQ4IPr+JDPrGEKYF1+AmT0E9ACmALcA3wG/BY4DXjGz/UIIk6uqoJl1AZ4ByoBzgTbAlcDm\nQL8UX+c4YETCtqlJyuoF3B/tfzmwGmgNbFNTAWZ2EXAz8BpwA35t/gB0Ak4BxqZY12L3F+r2Qepv\nwBPAswnbRwCl8e/VIrRG0GRmvwGeAg4DzgghDEvYvzFwGdC/uvPEbQvAmWY2MITwXQrHSJ5TUCG5\nNCaEMCX6/kEz+wG4BDgaeDJuv0vxgOLmEMJlcdvvN7NR+D/7YcARsSeiG20P4NYQwkUJ5Q40s27A\nqhrq9y9gGtA5hFAZnbcCuNzMbg8hzEzhNc4MITxa3Q5m9jvgLuD2EMIFKZwz/thGeKAzNoTQJcnz\nm9XmfLlmZk1DCEsyce4Qwmo8WEvX+QJQzAEFgP3vGw8ongAOB85KElCA/71UFSQk8yEePF8G/LWq\nsqVwqPtD8skE/B/J72MbzKwJcBHwMf6JcQ0hhBeA4cBhZrZX3DGXAR8BFycrKITwSAjh3aoqYmY7\nAjsCQ2MBReRu/O/mxFRflJk1MbPG1ezSJzrn1dH+66V6bmAzYENgYrInQwjzE+qyjpn9w8w+NbNl\nZvZfM7vJzNaJ2yc2VqBnwrF/i7YfVl2Fombt0WZ2iJlNjbqOPjSz4xL2izWZH2hmd5vZXOCruOe3\nNLMHzey7qK4fRC06ieVtZWb/NrNFUffPrfgnZkvY71djKsz1N+9KW2pm35vZS2bWNnq+EmgKxK5J\npZk9mHCdtk04Z9+orsvMu9vuMrNmCfu8FpW5o5m9GnVZfW1mSd+v1Vzrrmb2cVT3d83sgLjnOkT1\n+1XXTXRcpZntnWI5jYDHgaOAc0IIDybZLeAtZb/B//5SMRtv8TnTzFqkeIzkMQUVkk9aRV9/itu2\nP7Ax8GjCzT3eCPwGcmTcMZtEx9S1CXV3/J9kefzGEMK3wNfR86noCSwGYjfWkiT7dMSDpiPM7Cug\nwsx+MLNrzaymT2vfA0uBo8xs4+p2jM71HHAB3rpzLt69MwB4LLZf9An0eeBWM9sqOrYN8HfgvhDC\nmBrqFIDto3O+iN9gVgJPmFnHJPvfDewA/AO4MSpvC+At4M/AHcD5wKfAA2Z2ftxraoJ3Tx0S7Xcd\n/vu/meRjKhK3PQgMAr7EW8kG4tdzn+j50/DWiNej708D7q3qfGZ2Dd7q9DV+nZ8EzgbG2ppjcQL+\nHn0J7w67AJgB3Gipj4PpENV9JN6VuAnwkpntBBBCeA0P0rolObYb8FkI4a0aygh4kPAYcAzQN4Rw\nfzX7z6L2QcL1wNqkHohIPgsh6KFHVh94t8Rq4GBgU2Ar4ARgLn4D3jJu3/OjfY+u5nwb4QPwnoh+\nPq+mY1Ko44XRObZK8txbwJspnGMCfuM+EjgLeC+q59kJ+/0M/ICPhbgaH/MxMtr3+hTKuSaqawXw\nAj4mY/ck+52G39zbJ2w/Kzp+n7htzYH5wBj8H/4U4Atg/RTqMys63zFx2zYAvgHeTXgfVOJjQSzh\nHPfjN+aNErY/CvwINI5+7h+VdXzcPk2AmdH2A+O2PwR8EffzwVH5t9bweiqAB6t5H28b/bwZsAx4\nMWG/vtF+PeK2vRpt6xq3bW1gDjAqhWtcGR2/W9y2baL30JNx266Ptm0Qt20zPFC6qoYyDorKif0+\nz65m39i1aIt/OFgBDEp4vdOTvE9GR98/gP/tN08o+/jq6qhH/j3UUiG5YsB4YB7+aeoJYBEeCMyJ\n22+D6GtFNeeKPbdhwtfqjqnJutHX5UmeWxb3fJVCCAeEEO4KITwfQhgKtAM+AG5I6A5ZHw+M/h5C\n+EcI4ZkQQnf8ht6/pu6QEMI1QFf8xn8o/mm93HzWyg5xu56IfxqeaWabxh74P3zDb7Kxc87FB6Me\nigdHuwC9QwiLanrdkTkhhP8NbAwhVOCfYHePWiH+9xTe+pHYgnA83qrSKKGu4/Br1TbarwvwbQjh\n6biylgFDU6jjCfiN69oUX1NNOuGBwW0J2+/D34tHJGxfFOLG24QQVgJvA9ulWN7EEMK0uOO/wlug\nOse1cI3Ag6z47rpTgUbAIymWswU+/mh2KjuHEGbhQfFZZtY8xTKuQ60VRUFBheRKwMcSdML/ub/A\nL5+g4sUCgw2oWmLgsTCFY2qyNPqabCxEk7jnUxZCWIU3jW+EBxiJZT2WcEgpHrzU2NUSQng8hHAQ\n3lV0KH7D2B0YHTde4o/AznggF//4BP99bJF4Tvz3shd+438tldcZ+SzJttjA1pYJ22fH/2Bmm+PX\n6KwkdX0woa6/q6KsT1Ko43Z48PNzCvum4nfR1zUG8EbBwhdxz8d8neQcP+G/w1RUdY2b4jOUCCF8\nArzDml0gXYHJIYQvUigj4N1CXwFPmVn7FOtWqyChjoGI5CHN/pBceidEsz/M7FngDeBRM2sdfpkB\nMAP/FL0LMLqK8+wSff0o+vpxdEybao6pybfR19/izfbxfot3gdRFbCDiJnHb5uBTQOcm7Ps9/jpS\nvckQtSSMB8ab2SrgdGBvvLVhLeB9fAxFsrEaX8X/YGabAHvgN5adUq1DHSQGaLEPOw/jg3CTmV7F\n9kJS1UyUdM96GAHcZp4DZl18vEjfFI81/G+hE/Am8IKZHRRCeL+6g0IIs8zsYTxIuCnFsq4HuuOz\nvRKn70qBUEuF5IXggzAvx8dXnBv31Bv4mIOu1Qxa7IHf+J6PO+YnoCSFgY5VmYb/Q90jfqN5Yp+t\nSZJrIkWxmS3xOTVig0G3Sth3K/x1zaNuYrNbYsmIPgc2CSG8GkIoS/L4NOH4u/GumcuBA8wsccpf\ndf6QZFvr6OvsGo6dh7c6NaqinmXhl1ktXxI3WyjODkm2Jfoc2NLMNqphv1QH+34ZfW0dv9HM1sbH\nGXz5qyPq549JtrXGx1DEv2cew7t5SvBWihXAqNoUFEKYDXTGr8VYM0t2zRPFWisuTbGML/BA8mx+\nec9KgVFQIXkjhPAfvE/5r7Em+xDCUjxfxA74dLU1mNkReFAxJoTwdtwxN+Gfrm9OVpaZdTOzPZI9\nF53jI7zF46yEwKQv/g/6qbhzrWtmraM+/9i2X+WHMLMN8Ln481lzVsnjeABzRty+hicD+zFh38Rz\nrmtm+1Tx9OHR11hXwChgazM7M8l5mlhcZlIzOxE4Gbg0hHAzfmO6zsySBQvJbGlxU0jNbEP8U+jU\nEML31R0YBZhPASeY2c5J6hp/bV+Myjoh7vmmwK9eYxJPETeVtxqL8e6YmryCD4Q9P2H7X/BxPs//\n6oj6aW9m/+saM7Nt8BwvY+PHqIQQfsBnmXTHu0HGhBB+rG1hIYQP8HEhGwAvW5LMmQn7xwcJqc4E\nuQ5YB+9yUfKrQpTrkaJ6NLwHv4z6b5vkudjgubPitq2F3xBX4zMFzsP/UQ/HB5C9B2yecB7DE2Kt\nxj+xX4ZP77wMmBxt37uGeh4Rnf+VqLzbo5/vSdgvNlL973HbrsZbM66Njv07Ptp9FXBqkrJejp4b\ngo81GRfV8Ywa6rhpVPbEqIxeePfGf6Ljn0y4Js9H5TyKD8Q8H7gHD3TaRvttgX/SfTnu2E3wZvBU\nZr3MwrutfsADwf54d8VKoFOK74Mt8HEIi/Bpk2fin3hHAfPj9lsXH0ewBJ8Oej4+hmAqNcz+iLYN\nj/Z7IXpf9cengfaN2+d5fJzOADxD6V5x9f/f7I+43/tqfJBtX3ya60pgEt7yEtvvV7MhqqpjFde4\nEn/fz8WTn10SXffFwJ+S7H88v8wYOTHFv9OkMzDw6bvL8MRVm1T3u8RbkVZGz1U5+yPJNYjVVbM/\nCuyR8wro0fAecf+Mk91MDM9HMJNfTzM8Hc8X8FP0z3M6cAWwbjVlHYd/SpuHz+T4JroxHZhiXY/G\nWwqW4M3X18TfHKJ9Dopez1Vx2zpFN5Zvon/AP+Cfqg+qopymwK3R/kvx7pdfBR9JjmsE9MY/dX8R\n1bMCD6QGAL9Jsv9F0bVbggcTb0fXcf1onyeja7x1wrFHRa/zohrqNAsfy9Ipeh1LohvQcam+D6Ln\nN8NvyrOja/gNHmz1TthvazzfRgV+k70luvElCyo+T/J+uyCq31I8lfvzrDlVc3s8CFgUnfPBhPpv\nm3DOPtH5luHjZe4ENkzY51XgvSSv+Vd1rOLarMaD3BK8JWoJHkwdUMX+a0fvwR+BdVJ878fe17+6\nseMZblfiAfp61f0u8cG1qxNfb/R+fTbJ/r/Hu2hWJStbj/x+WPRLFBFJC/Osle+HEI7OdV3ERYm3\n5uA38bNyXR8pXhpTISJS/I7DW34SF7cTSStNKRURKVLm6+Hsio+7mBJCeCPHVZIip5YKEUm3gEbu\n54s+wGB8rEiPHNdFGgCNqRAREZG0UEuFiIiIpEWDGVMRJSbqzC/T00RERCQ1TfB1e8YGT6iWVIMJ\nKvCAItVV+UREROTXuuHJ85JqSEHFbICHH36YHXfcMcdVKQ4DBgxg0KBBua5GUdE1TS9dz/TTNU2v\nQrmeM2bM4LTTToMa1u5pSEHFMoAdd9yRtm3b5rouRaFZs2a6lmmma5peup7pp2uaXgV4PasdPqCB\nmiIiIpIWCipEREQkLRRUiIiISFooqJA6KykpyXUVio6uaXrpeqafrml6Fdv1bDAZNc2sLVBeXl5e\naINiREREcmrKlCm0a9cOoF0IYUpV+6mlQkRERNJCQYWIiIikhYIKERERSQsFFSIiIpIWCipEREQk\nLRRUiIiISFooqBAREZG0UFAhIiIiaaGgQkRERNJCQYWIiIikhYIKERERSQsFFSIiIpIWCipEREQk\nLRRUiIiISFrkRVBhZgeY2Wgz+8bMKs3s6BSO6WBm5Wa2zMxmmlmPbNRVREREksuLoAJYD5gG9AVC\nTTubWUvgeWA8sCtwO3C/mR2SuSqKiIhIdX6T6woAhBDGAGMAzMxSOKQP8EUI4ZLo50/MbH9gAPBy\nZmopIiIi1cmXlora2gd4JWHbWKB9DuoiIiIiFG5Q0QKYm7BtLrChmTXOQX1ERERqbfHiXNcgvfKi\n+0NERKSYLV4MM2bAhx/CBx/Ah9NX8eHEhXy9ZGMqKoymTXNdw/Qo1KDiO6B5wrbmwMIQwvLqDhww\nYADNmjVbY1tJSQklJSXpraGIiDQ4y5bBxx978PC/AOJDmDULQjQNodVvl7LzzxPpumoqO5+9L2b7\n5rbSCUpLSyktLV1j24IFC1I61kKocbJFVplZJXBsCGF0NfvcCHQJIewat+1RYKMQwuFVHNMWKC8v\nL6dt27bprraIiDQgK1bAp5/+EjTEAojPPoPKSt9n663hT3+CnXeOvrZczI4PX8H6D9wOHTrAAw/A\ndtvl9HWkasqUKbRr1w6gXQhhSlX75UVLhZmtB/wBiM382M7MdgV+DCF8ZWYDgS1DCLFcFEOAfmZ2\nE/Ag0BE4EUgaUIiIiNTH1Knw3HO/BA8zZ8KqVf5cixYeOHTp8ksAsdNOsEajeFkZ9OwN8+fDXXdB\nnz6wVqEOa6xaXgQVwB7Aq3iOigDcEm0fDvTGB2ZuE9s5hDDbzI4ABgHnA18DZ4QQEmeEiIiI1MnK\nlfD003DnnfDmm7DxxrDLLt7IcO65HkDsvDNsumkNJ7rtNhgwwA8sKyuY1om6yIugIoTwH6qZiRJC\n6JVk2+tAu0zWS0REGp65c2HoUBgyBObM8Vjgqafg6KPhN3W5a3bpAmuvXbStE/HyIqgQERHJtbff\n9laJUaOgUSPo3t1bJNq0qeeJW7f2RwOgoEJERBqsFSvgiSfgjjs8qGjVCm64AXr39u4OqR0FFSIi\n0uDMmQP33uuPuXOhUyd49lk44ghvpZC6UVAhIiINQggwaZJ3cTz5JDRuDD16eBfHjjvW48RlZT49\n5MIL01bXQlXcI0ZERKTBW7YMhg2DPfaA/faDd9+Ff/0LvvkGBg+uR0BRUeGDLzt2hJdegtWr01nt\ngqSWChERKUpffQX33AP33efpIbp0gRdfhM6d0zAJo6zMB14Ued6J2lJQISIiRSMEmDDBuzieeQbW\nWw969YJ+/eCPf0xDARUVcMklPt+0AeSdqC0FFSIiUvCWLIFHH/VgYvp02GEHn9HRvTtssEGaCpk0\nCUpK1DpRDQUVIiJSsGbPhrvvhvvvh59/hiOP9PESnTqBWY2H187663u+CbVOVElBhYiIFJQQ/L5+\n552+HseGG8IZZ0Dfvhm+17dpA2PHZrCAwqegQkRECsKiRfDwwx5MfPSRL9x1zz3QrZuPnZDcU1Ah\nIiJ57fPPferngw/6OMljj/WfDzooA10cUi8aYSIiInmnstJ7Go480mdtDB8O55wDX3zhi3t16JCB\ngGLRInjhhTSftGFRS4WIiOSNhQs9gLjrLpg5E3bbzQdhlpTAuutmsOCyMh+Y8fPPMGsWbLRRBgsr\nXmqpEBGRnPvkEzjvPNh6axgwwIOJCRNgyhTPMZWxgGLRIh/h2bEjtGwJ5eUKKOpBLRUiIpITlZWe\n4fLOO2HcONh8c+jf37s5ttoqCxWItU7Mm6e8E2mioEJERLLq55/hoYd8sOXnn/uaHCNGwMkn+yJf\nGbdokWfFvOceH5wxfrzyTqSJggoREcmKjz7yVomRI2HFCjjpJJ8iuvfeWZ7FMXOmp99U60TaKagQ\nEZGMWb3aE1Tdeaf3NrRoARdfDGef7d/nRNu28N//etYsSSsFFSIiknY//uizNu6+G778Etq398aB\nE06AddbJde1QQJEhCipERCRtpk/3VomHH/aBmCUlPqujXbtc10yyQUGFiIjUy6pV8O9/ezDx+us+\nc+PKK+HMM2GLLXJUqa+/9vmpklUanSIiInUybx7ccAO0auWDLkOAUaM8d9QVV+QooKio8MGXf/gD\nfPZZDirQsKmlQkREaqW83FslHnvMZ2106+ZdHLvumuOKlZV5pqz58+GWWzRNNAcUVIiISI1WrvQ1\nN+64AyZNgm23hWuv9dxRm26a48ol5p0oK1NAkSMKKkREpErffQdDh8KQIfDtt3DwwfDMM3DUUdCo\nUa5rh7Ji5hkFFSIi8itvveVdHKNGwdprQ/fucO658Kc/5bpmcWbPhkMPhQMOUFbMPKGgQkREAFi+\n3IOIO++Ed97xe/SNN0KvXrDxxrmuXRItW/p0k332UetEnlBQISLSwH3zjXdvDB0K33/vH/6few66\ndMmTLo7q7LtvrmsgcRRUiIg0QCHAxIneKvHUU9CkCfTsCf36wQ475Lp2UqgUVIiINDCPPw433QRT\np8L228Ott0KPHnmaubqyUl0bBUS/KRGRBmLFCp8cceqp0Lw5vPQSzJjhOSbyMqAoK/ORoTNn5rom\nkiIFFSIiDcD338Mhh8ADD8B993lAcdhhedoIEMuK2bGjRz95sQKZpELdHyIiRW7aNDjmGJ/d8eqr\nsN9+ua5RNeKzYirvRMHRb0pEpIiNGuUTJDbf3KeJ5m1AEd860aqVL3far58CigKj35aISBGqrPSV\nQk85BY47DiZMgG22yXWtqlBRAbvsAiNHeuuEElkVLHV/iIgUmYUL4bTT4PnnfZbHxRf7wl95a4MN\nvJKHHaZgosApqBARKSKffebjJ77+2oOKww/PdY1S1LdvrmsgaaDuDxGRIjFuHOy5J6xaBW+/XUAB\nhRQNBRUiIgUuBBg0yNNqt2/vi4G1bp3rWklDpKBCRKSALVvmC35dcAFcdJGv2bHRRrmuVYKKCujf\nH2bNynVNJMM0pkJEpEDNmQPHHw/vvQcPPwzduuW6RknE8k7Mmwd//rNPF5WipZYKEZEC9NZbsMce\nPiBzwoQ8DCji8060bAnvv+8jSKWoKagQESkww4fDgQf6h/533/XgIq+UlUGbNjBihOedKCvTVNEG\nQkGFiEiBWLXKx0707Andu/u9ukWLXNcqTmWlTw2Nb51QVswGJW9+02bWz8xmmdlSM5tsZnvWsH83\nM5tmZovNbI6ZPWBmm2SrviIi2fTTTz5F9I474M47fVGwxo1zXasEa63li3+pdaLByouBmmZ2CnAL\ncBbwNjAAGGtm24cQ5ifZfz9gONAfeB7YCrgXGAqcmK16i4hkw0cf+XCEH3/0XBR//nOua1SN227L\ndQ0kh/KlpWIAcG8IYUQI4WPgHGAJ0LuK/fcBZoUQBocQvgwhTMSDir2yU10Rkex47jnYZx9o0sQX\nBMvrgEIavJwHFWa2NtAOGB/bFkIIwCtA+yoOmwRsY2ZdonM0B04CXshsbUVEsiMEuP56b6Ho1Akm\nTVJvguS/nAcVwGZAI2Buwva5QNIhSFHLxGnA42a2AvgW+Ak4N4P1FBHJisWLfXXRK6+Eq6+GJ5+E\n9dfPda0i48f7PFaRJPIhqKg1M9sJuB24BmgLdAZa4V0gIiIF68svYf/94cUX4amnPKjIi8kTsbwT\nnTrBPffkujaSp/JhoOZ8YDXQPGF7c+C7Ko65DHgzhHBr9PMHZtYXmGBmV4QQEls9/mfAgAE0a9Zs\njW0lJSWUlJTUqfIiIuny+utw4oneKjFpkqd6yAvjx8MZZ8D8+T6zo0+fXNdIMqi0tJTS0tI1ti1Y\nsCClY3MeVIQQVppZOdARGA1gZhb9fEcVhzUFViRsqwQCYNWVN2jQINq2bVuvOouIpNuQIXDeeXDA\nATBqFGy2Wa5rhLdOXHKJV65DB00TbSCSfdCeMmUK7dq1q/HYfGhUA7gVONPMTjezHYAheOAwDMDM\nBprZ8Lj9nwNOMLNzzKxVNMX0duCtEEJVrRsiInlnxQo45xz/8N+nD4wdmycBRWJWzPHjFVBIjXLe\nUgEQQhhlZpsB1+LdHtOAziGEedEuLYBt4vYfbmbrA/2AfwE/47NHLstqxUVE6uH77727Y/JkuP9+\n72HIGw895Fkx1TohtZAXQQVACOFu4O4qnuuVZNtgYHCm6yUikgnTpvl00eXL4bXXYN99c12jBPfe\n68kx8mKUqBQKvVtERLJs1CgPIjbf3BNa5V1AAdC0qQIKqTW9Y0REsqSy0nNPnHIKHHecL1m+zTY1\nHydSKPKm+0NEpJgtXAinnQbPPw833wwXXQRW7Vy1DKuo8L6XvBgVKsVCLRUiIhn26ae+fsfrr8ML\nL8DFF+c4oIjN7Dj//BxWQoqRggoRkQwaNw722gtWr4a33oIuXXJYmVhWzI4doVUruO66HFZGipGC\nChGRDAgBBg3yIKJ9ew8oWrfOYYVirRMjRyrvhGSMggoRkTRbtgx69oQLLvCxE889BxttlKPKJLZO\nTJ8O/fppZodkhAZqioik0Zw5PrNj+nR45BHo2jXHFbrsMm+dGDzYU3cqmJAMUlAhIpImb73lAUWj\nRvDGG5DCUgmZd/XV3lzSqlWuayINgEJWEZE0GD4cDjzQ793vvJMnAQXAFlsooJCsUVAhIlIPq1b5\n2ImePaF7dx8P2aJFrmslkhvq/hARqaMff4RTT/VA4s47ffxjTvJPrF7tfS4iOaagQkSkDj780BcE\n+/lnePllOPjgHFSiogIuucTTdT7ySA4qILImdX+IiNTS6NGeIbNpUx8/kZOAIj7vxL77emIMkRxT\nUCEikqIQ4Prr4dhj4ZBDYOLEHIyBrCrvRE7zfos4dX+IiKRg8WLo1QueeAKuuQauuioHKR/KyqB3\nb5g/37Ni9umjvBOSVxRUiIjU4MsvffzEZ5/B0097LoqsGzoUzj4bOnTw4EIptiUPKagQEanG66/D\nCSfABhvApEk+jCEnjjwS7r7bAwu1Tkie0jtTRKQKQ4b40IVddvEBmTkLKAC23FLdHZL39O4UEUmw\nYoUvk9Gnjz/GjIFNN811rUTyn7o/RETifP89nHgiTJ4M998PZ5yR6xqJFA61VIiIRKZOhT33hJkz\n4bXXshxQlJXBFVdksUCR9FNQISICPP447LcfbL45vPuu55PKivi8E5MmwbJlWSpYJP0UVIhIg1ZZ\n6Q0Ep57qU0UnTICtt85S4fFZMQcPhldegSZNslS4SPopqBCRBmvhQs+OOXAg3HwzPPwwrLtuFgpO\nzIr5/vvQt69mdkjB00BNEWmQPv3UE1rNmQMvvABdumSp4HfegZNOUlZMKUp6J4tIgzNuHOy1l3d9\nvPVWFgMKgC228MQXsTU7FFBIEdG7WUQajBDg1ls9iNh3Xw8oWrfOciV+9ztf5lRptqUIKagQkQZh\n2TLo2RMuvBAuvtjv682a5bpWIsVFYypEpOjNmeMzO6ZPh0cega5dc10jkeKklgoRKWqTJ8Mee3hg\n8cYbWQgoKirgsccyXIhIflJQISJFa/hwOOggn7X5zjvQrl2GC4zlnTjzTPjuuwwXJpJ/FFSISNFZ\ntQoGDPAxFKef7vf6Fi0yWGBi3on33stwgSL5SWMqRKSo/PijZ8csK/M0EH37glkGCywrg969lXdC\nBAUVIlKkHJAfAAAgAElEQVREPvzQE1r9/DO8/DIcfHAGC6uogEsugSFDoEMHDy40TVQaOIXTIlIU\nnn0W9tkHmjb18RMZDSjA10h/6ilvnRg/XgGFCGqpEJECFwJcfz1cdRUcf7wPzlx//SwU/Pvfw5df\nZmmxEJHCoKBCRArW4sXQqxc88QT84x9w5ZVZHs6ggEJkDQoqRKQgzZ7tK4x+9hk8/bQntxKR3NKY\nChEpKCtX+vodbdr40uWTJmUwoPjsswydWKQ4KagQkYIxYYInsLr4YujRA6ZM8eAi7WJ5J1q3hqlT\nM1CASHFSUCEieW/uXA8iDjzwl9kdd90FG22UgcJiWTFHjoQ77oBdd81AISLFSUGFiOSt1ath8GBv\nMHj+ebjvPpg4Edq2zUBhiVkxp0+Hfv2UyEqkFjRQU0Ty0uTJng1z2jT4y19g4EDYdNMMFaasmCJp\nob8aEckr8+f7elzt23t67UmTYOjQDAYU338PRxyh1gmRNFBLhYjkhcpKeOABuOwy/37wYDj7bGjU\nKMMFb7GFRy677KJgQqSe8uYvyMz6mdksM1tqZpPNbM8a9l/HzK43s9lmtszMvjCznlmqroikUXm5\nt0ycdRYcdRR88ol3fWQ8oIjZbTcFFCJpkBd/RWZ2CnALcDWwO/AeMNbMNqvmsCeAg4FewPZACfBJ\nhqsqImn000/e27DnnrB0qU8ZHTbMGw9EpPDkS/fHAODeEMIIADM7BzgC6A3cnLizmR0GHABsF0L4\nOdr83yzVVUTqqbISRozwRT6XLfNkVueeC7/J1H+kVasyeHIRiUm5pcLMpprZlFQetamAma0NtAPG\nx7aFEALwCtC+isOOAt4FLjWzr83sEzP7p5k1qU3ZIpJ906d7volevaBTJ/j4Y/jrXzN4zy8rgx12\ngLffzlABIhJTmz/jf2eoDpsBjYC5CdvnAq2rOGY7vKViGXBsdI57gE2AMzJTTRGpj4UL4eqr4c47\nYfvt/V6f0eXJKyq8KWTIEOjQATarrjdVRNIh5aAihPCPTFakltYCKoGuIYRFAGZ2AfCEmfUNISyv\n6sABAwbQrFmzNbaVlJRQUlKSyfqKNFghQGkpXHihBxY33OAtE+usk8FClXdCpM5KS0spLS1dY9uC\nBQtSOjYfOhnnA6uB5gnbmwPfVXHMt8A3sYAiMgMwYGvg86oKGzRoEG0zko5PRBJ99JGPlXj1VTjh\nBBg0CLbZJoMFJrZOlJXBdttlsECR4pPsg/aUKVNo165djcemHFSY2U9ASGXfEMImqZ43hLDSzMqB\njsDoqCyLfr6jisPeBE40s6YhhCXRttZ468XXqZYtIpmxaBH83//5AMxWrWDMGOjcOcOFrlgBe+wB\n33yj1gmRHKlNS8VfM1YLuBUYFgUXb+OzQZoCwwDMbCCwZQihR7T/o8CVwENmdg2wOT5L5IHquj5E\nJLNCgKeeggEDvOfh6qt9RdHGjbNQ+DrrwN//7gkv1DohkhO1GVMxPFOVCCGMinJSXIt3e0wDOocQ\n5kW7tAC2idt/sZkdAtwJvAP8ADwOXJWpOopI9T791Ls6xo2Do4+G227zVoqs6tYtywWKSLx6j6mI\npnGuMeQqhLCwtucJIdwN3F3Fc72SbJsJZLpBVURqsGSJL/Z1882w5Zbw3HNw5JG5rpWI5EKdggoz\nWw+4CTgZSLbMT7aS64pIDo0eDf37w5w5cOmlcPnlsO66ua6ViORKXUcx3Qz8GegDLAf+gqfYngOc\nnp6qiUi+mjXL1+g45hho3Ro++ACuvTbDAUVFhef0njYtg4WISH3UNag4CugbQngKWAVMCCFcB/wN\nUKemSJFatsyDh512gvfe80GZL70Ef/xjhgsuK4M2bWD4cB+8ISJ5qa5BxSbAF9H3C6OfAd4ADqxv\npUQk/4wZ4/f1//s/7/KYMQOOPx7MMlhoRYVPDe3Y0Ud9Tp8OJ52UwQJFpD7qGlR8AcTGdX+Mj60A\nb8H4OekRIlKQ/vtfT1zVpQtsu63f12+8EdZbL8MFx1onRo70vBPjx2uqqEieq2tQ8RCwa/T9jUA/\nM1sGDAL+mY6KiUhurVjhwcOOO8KkSZ5q+5VX/OeMCgHOO2/N1ol+/ZTISqQA1Gn2RwhhUNz3r5jZ\nDvhKo5+FEKanq3IikhtlZX4f//RTOP98uOYa2HDDLBVuBi1aKCumSAFKy9ofIYQvgS/TcS4RyZ05\nc3zhr8ceg/33h1GjvAci6664IgeFikh91ekjgJndYWbnJtl+rpndVv9qiUg2rVzp63S0bu2tFMOH\nw+uv5yigEJGCVdd2xRPwmR6JJgIn1r06IpJtEyZAu3a+RkePHvDJJ3D66Rme1SEiRamuQcWmQEWS\n7QuBzepeHRHJlrlzPYg48EBo2hTeeceHMWy0URYKLyvzddFFpKjUNaj4DOiSZHsXfslfISJ5aPVq\nDx5at4YXXoD77oOJE6Ft2ywUHp934t57s1CgiGRTXQdq3grcZWabA2XRto7AhWR2iXQRqYfJk6Fv\nX890/Ze/+EJgmyZbvScTysqgd29fEz02s0NEikpdp5Q+aGaNgSv4Zbnx2UCfEMKINNVNRNJk/nxf\n7Ov++71FYtIk2HvvLBW+aBFccgnccw906ODBhZJYiRSlOk8pDSHcA9wTtVYsDSEsSl+1RCQdKis9\nkLj8cv9+8GA4+2xolK11hMvK4IwzYN485Z0QaQDqHFSY2W+ADsDvgUejbVsCCxVgiOReebl3dbz9\nNvTsCTfdBFtskeVKvPACtGypFNsiDUSdggoz+x0wBtgWaAy8jM8GuTT6+Zx0VVBEauenn+DKK723\noU0bnzK6//45qszAgfCb36h1QqSBqOtf+u3Au8DGwNK47c/gAzZFJMsqK2HYMJ/VMXKkJ7MqL89h\nQAGwzjoKKEQakLp2fxwA7BtCWGFrZsiZDWxV30qJSO1Mn+5dHW++CV27wj//CVtumetaiUhDU9eP\nEGsByYZ6bU3ypFgikgELF8KAAT6j48cffVzkI49kMaCoqIBvvslSYSKS7+oaVIxjzXwUwczWB/4B\nvFjvWolItUKARx/1ro777oMbbvDcEwcfnMVKlJX5oI0zzshioSKSz+oaVFwI7GdmHwFN8Nkfs/GW\nikvTUzURSeajjzwhZbduPl5ixgxPA7HOOlmqQHxWzFat4O67s1SwiOS7uia/+trMdgVOBXYB1gce\nAB5JY91EJM6iRXDttTBokN/Lx4yBzp2zXIlkWTE1EFNEIvVJfrUKeDj2c5Rhsx9wCdCi/lUTEfCu\njqee8rET8+fD1Vf7iqKNG2exEhUV3hwyZIiyYopIlWr1EcPMGpvZQDN718wmmtmx0fZewCxgADAo\nA/UUaZC+/BIOOwxOOskHY86Y4TkoshpQgGfOGjnSWyeUyEpEqlDbloprgbPxZFf7AU+Y2UPAPsAF\nwBMhhNXpraJIw/Tcc740+QYb+PdHHpnDylx2mXd7KJgQkWrUNqg4CTg9hDDazP4ETI/OsWsIIaS9\ndiIN0MqVcMUVnmvi6KM9odXGG+e4Uuuv7w8RkWrUNqjYGigHCCF8YGbLgUEKKETS4+uv4ZRTfL2O\nf/0LLrgA1swvJyKSv2obVDQCVsT9vArQ4mEiafDSS9C9OzRtCq+/Du3bZ7kCK1fC2mtnuVARKSa1\nDSoMGBa1UIDnqBhiZovjdwohHJ+Oyok0BKtWwd//7mtvHX44jBgBm26axQosWuQzOz79FMaNU9OI\niNRZbYOK4Qk/P5x0LxFJyZw5UFLia3bceKNPFc1q2oeyMs+IOW+ez/AIQUGFiNRZrYKKEEKvTFVE\npKF5+WXPirn22vDqq3DAAVksPNY6cc89nndC00RFJA2UCk8ky1av9gRWnTvD7rvD1KlZDihia3aM\nGKG8EyKSVnXOqCkitffdd9468dprnnL7b3/LcnfHI4/AaaepdUJEMkJBhUiWvPoqdO3q37/ySpZX\nFI058kgYOtTHUWjNDhFJM/1XEcmwykq47jro1Al23NG7O3ISUAA0awZnnqmAQkQyQi0VIhk0b573\nNrz8Mlx1lU8dbdQo17USEckMBRUiGTJhApx6queUGjsWDjkk1zUSEckstYGKpFllpad8OPhg+MMf\nYNq0LAYU48fDOed4vgkRkSxTUCGSRj/8AEcd5Yt6XnKJ3+O33DILBVdUQJ8+PnDjk0/8ZxGRLFP3\nh0iaTJzoi4EtXQovvghdumSp4PHjfTbH/Pmed6JPHw3EFJGc0H8ekXoKAW65BQ46CLbd1md3ZCWg\niG+daNUKpk+Hfv0UUIhIzqilQqQefvoJevaE0aN93Y7rr8/SQp/vvQfHHOPTS9Q6ISJ5QkGFSB29\n/TacfDIsXOhBxVFHZbHwrbeGvfbyVciUFVNE8oQ+2ojUUghwxx2w//7QvLl3d2Q1oABfG33UKAUU\nIpJX8iaoMLN+ZjbLzJaa2WQz2zPF4/Yzs5VmNiXTdRRZsABOOgn69/fhCxMmwO9+l+taiYjkh7wI\nKszsFOAW4Gpgd+A9YKyZbVbDcc2A4cArGa+kNHhTpkDbtr5ux9NPw6BBsM46ua6ViEj+yIugAhgA\n3BtCGBFC+Bg4B1gC9K7huCHAI8DkDNdPGrAQ4O67oX172HhjDy6OOy7DhVZUwL33KomViBSUnAcV\nZrY20A4YH9sWQgh460P7ao7rBbQC/pHpOkrDtXAhlJR4V8dZZ8Gbb2ZhGMP48dCmDVx4IXzxRYYL\nExFJn5wHFcBmQCNgbsL2uUCLZAeY2R+BG4BuIYTKzFZPGqr33oM99vBEVqNGwZ13QuPGGSwwWd6J\n3/8+gwWKiKRXPgQVtWJma+FdHleHED6Pbc5hlaTIhAD33Qf77APrrQfl5T44M6NirRMjR3reifHj\nNbNDRApOPuSpmA+sBponbG8OfJdk/w2APYDdzGxwtG0twMxsBXBoCOG1qgobMGAAzZo1W2NbSUkJ\nJSUldau9FJVFi7yx4OGH4eyz4bbboEmTDBZYUeGLhAwZAh06QFmZggkRyanS0lJKS0vX2LZgwYKU\njrWQBwPBzGwy8FYIoX/0swH/Be4IIfwzYV8Ddkw4RT/gYOAEYHYIYWmSMtoC5eXl5bRt2zYDr0IK\n3QcfeIvEV1/B0KHQtWsWCp07F/be29NxKiumiOSpKVOm0K5dO4B2IYQqUzjkQ0sFwK3AMDMrB97G\nZ4M0BYYBmNlAYMsQQo9oEOdH8Qeb2ffAshDCjKzWWorGsGHQt68PYXj3XdhhhywV3Lw5zJypuaki\nUhTyIqgIIYyKclJci3d7TAM6hxDmRbu0ALbJVf2keC1Z4jM7hg2D3r19MGbTplmuhAIKESkSeRFU\nAIQQ7gburuK5XjUc+w80tVRqacYM7+744gsPKnr0yHWNREQKmzpwpUF6+GHYc0+orIR33slwQPHB\nBxk8uYhI/lBQIQ3K0qWexKp7dzj+eA8odt45Q4XF8k60aQOvvpqhQkRE8kfedH+IZNrMmb5U+Sef\nwP33+xgKy1SGk7IyL2DePM87cdBBGSpIRCR/qKVCGoTHH4d27byl4q234IwzMhRQxFonOnaEli3h\n/fd9JKimiopIA6D/dFLUli3zqaKnngpHHunTRXfZJUOFlZV5V8eIEd46oURWItLAqPtDitbnn3t3\nx4cfesLKs87KYHfHwoVw4okesSiYEJEGSkGFFKWnn4ZevWDzzWHSJNh99wwXuOGGMHEibL+9ujpE\npMHSfz8pKitWQP/+cMIJcOihvhhYxgOKmB12UEAhIg2aWiqkaMye7d0d06Z5Zsx+/TLY3SEiIr+i\noEKKwujRnsBqo428F2KPPTJQyPLl0LhxBk4sIlIc1FYrBW3lSrjoIjjmGE8FMWVKhgKKsjLv3njp\npQycXESkOCiokIL11VceSNx+O9x6KzzzDGy8cZoLScw70bp1mgsQESke6v6QgvTii55qe/31YcIE\n2GefDBQSy4o5fz4MHgznnKOBmCIi1dB/SCkoq1bB5ZfDEUfAvvvC1KkZCCjiWydatfKsmH37KqAQ\nEamBWiqkYHzzDZSU+EDMm27ysRRpv8+vXg377efrod91lwcXCiZERFKioEIKwrhxcNppsM468Npr\nsP/+GSqoUSO4/npfulRZMUVEakUfwSSvrV4NV10Fhx0Gbdt6d0fGAoqYo45SQCEiUgdqqZC89d13\n0LUr/Oc/cN11cNll6okQEclnCiokL5WVeUBhBuPHQ4cOua6RiIjURJ/7JK+sXg3XXguHHAJ/+pOn\n3E5rQBGb2VFWlsaTiogIKKiQPPL99z524ppr4O9/h7FjoXnzNBZQVgZt2sDIkfDtt2k8sYiIgLo/\nJE+8/jqceqq3VLz8sqeISJuKCrjkEhgyxJs9Xn3V80+IiEhaqaVCcqqyEgYOhIMPhu239+6OtAYU\n8a0Tgwf7AA0FFCIiGaGgQnJm/nw48kj42988S+Yrr8Bvf5vGAi68UFkxRUSySN0fkhNvvundHcuW\nwZgx0LlzBgrZfnut2SEikkUKKiSrQoBbbvGcE+3bQ2kpbL11hgo7++wMnVhERJLRxzfJmh9/hGOO\ngYsv9p6JsrIMBhQiIpJ1aqmQrHjrLTj5ZJ+I8dxzPpZCRESKi1oqJKNCgNtugwMOgC239NkdaQso\nyspg8uQ0nUxEROpLQYVkzM8/wwknwIABcN55vobHttum4cSxrJgdO8J996XhhCIikg7q/pCMePdd\n7+748Ud45hk49tg0nbisDHr39vmod93lwYWIiOQFtVRIWoXgszj32w823dSXKk9LQBHfOtGqFUyf\nDv36aaqoiEge0X9kSZuFC+GUU+Dcc3025xtvpCl5ZXxWzLvu8qyY222XhhOLiEg6qftD0mLaNDjp\nJF8U7Ikn4MQT03jyd97x6KSsTMGEiEgeU0uF1EsIcO+9sM8+sP76UF6e5oAC4KKL1DohIlIAFFRI\nnS1aBKed5lmwe/WCSZPgD3/IQEGNGmnshIhIAVD3h9TJ++97d8c338Cjj0JJSa5rJCIiuaaPf1Jr\nDz0Ee+8N66zjU0frHVBUVMBnn6WlbiIikjsKKiRlixdDz56eJqJbN0+93bp1PU8am9lx2mk+QENE\nRAqWggpJyUcfwV57+cyOESM8keW669bjhIl5Jx59FMzSVl8REck+BRVSo5EjYc89/ft33oHu3et5\nQuWdEBEpSgoqpEpLl8Jf/gKnn+6DMt9+G3baqR4nVFZMEZGiptkfktQnn3gg8emn8OCDPmW03oYO\n/aV1ok8fBRMiIkVGQYX8SmkpnHUWbLWVt060aZOmE59/vi9b2rJlmk4oIiL5RB8V5X+WLfMGhK5d\n4eijffxE2gIKgLXXVkAhIlLE1FIhgKeJOPlkn+Vx771w5pmajCEiIrWTNy0VZtbPzGaZ2VIzm2xm\ne1az73FmNs7MvjezBWY20cwOzWZ9i8mTT0K7dj6OcvJk7/qoc0CxdGla6yYiIoUjL4IKMzsFuAW4\nGtgdeA8Ya2abVXHIgcA4oAvQFngVeM7Mds1CdYvG8uVw3nk+ILNzZ18MbLfd6niy2MyO/faDlSvT\nWk8RESkM+dL9MQC4N4QwAsDMzgGOAHoDNyfuHEIYkLDpCjM7BjgKD0ikBrNmeXfH9Ok+GaNv33q0\nTpSVeZrN+fPhppt8ATAREWlwct5SYWZrA+2A8bFtIYQAvAK0T/EcBmwA/JiJOhabZ5+Ftm3hhx9g\n4kRPFVGngEJ5J0REJE4+/PffDGgEzE3YPhdokeI5LgbWA0alsV5FZ8UKuOACOPZYOPhgmDLFx1LU\nibJiiohIgnzp/qgzM+sKXAUcHUKYX9P+AwYMoFmzZmtsKykpoaTI1+7+73+9u6O8HAYNgv7969Hd\n8eyzHpl06ODBhYIJEZGiUVpaSmlp6RrbFixYkNKxFnK8MmTU/bEEOCGEMDpu+zCgWQjhuGqOPRW4\nHzgxhDCmhnLaAuXl5eW0bds2LXUvFC+84Km2118fRo3yZcvrZflyeOwxXwREXR0iIkVvypQptPOm\n7XYhhClV7ZfzO0IIYSVQDnSMbYvGSHQEJlZ1nJmVAA8Ap9YUUDRUK1fCpZfCkUf6pIypU9MQUAA0\nbgw9eiigEBGRNeRL98etwDAzKwfexmeDNAWGAZjZQGDLEEKP6Oeu0XPnA++YWfPoPEtDCAuzW/X8\n9PXXcOqpnnfin/+ECy9UMisREcmsvAgqQgijopwU1wLNgWlA5xDCvGiXFsA2cYeciQ/uHBw9Yobj\n01AbtDFjvGeicWP4z3+8lUJERCTT8qb9OoRwdwihZQhh3RBC+xDCu3HP9Qoh/Dnu54NDCI2SPBp0\nQLFqFVx5JXTpAnvsAdOm1TGgKCuDU07xE4qIiKQob4IKqZ9vv4VOnWDgQLjhBh+cuVlV+UirEp93\n4vvvIcXRviIiIpAn3R9SP+PH+8qijRrBq6/CgQfW4STxWTHvusuDCw3EFBGRWtBdo4CtXg3XXAOH\nHAK77OLdHbUOKJQVU0RE0kQtFQVq7lzo1s0bGK65Bq64og5LbsyY4QMw1DohIiJpoKCiAP3nPz5d\nNAR45RX4859rPiap3/3O83VfdZWyYoqISL3pY2kBqayE66/3IGKHHby7o84BBUDTpvDQQwooREQk\nLRRUFIh58+Dww71R4YorvIWiRarLrYmIiGSBuj8KwBtveHfH8uWe2OrQQ3NdIxERkV9TS0Ueq6yE\nm2/2xUC32867O2oVUFRUwC23+IlEREQyTEFFnvrhBzj6aF8Q7OKLfZbHVlvV4gRlZdCmDVx9NXzw\nQcbqKSIiEqOgIg9Nngy77w6TJnlmzIED4TepdlQlyzuxyy4Zra+IiAgoqMgrIcCgQXDAAbD11r5U\n+eGH1+IEsdaJkSM978T48ZrZISIiWaOgIk/89BMcfzxccAH07++5KLbdNsWDFy1SVkwREck5zf7I\nA++8AyefDD//DM8+62MpaiUEmDBBWTFFRCSnFFTkUAgeB1x4Iey2my8G1rJlHU60wQbw3nt1yNMt\nIiKSPvpImyMLFnjrxPnnQ9++nouiTgFFjAIKERHJMbVU5MDUqXDSSZ4l88kn4YQTcl0jERGR+lNL\nRRaFAEOGQPv20KwZTJlSi4Di3Xf9BCIiInlKQUWWVFRA164+jvKMM+DNN+H3v0/hwEWLvH9kzz3h\n3//OeD1FRETqSt0fWTB9und3fPstPPYYnHJKigeWlXkEMm+ej+g85piM1lNERKQ+1FKRQSHA/ffD\n3ntDkybeg5FSQBFrnejY0UdvKu+EiIgUAN2lMmTxYujRA848E7p399Tb22+fwoGxrJgjRigrpoiI\nFBR1f2TAhx96d8eXX3rG7NNOS/HA5cs9EvnDHxRMiIhIwVFQkWbDh3vPRatW3t2x4461OLhxY8+M\nue226uoQEZGCoztXmixd6mMqe/b0cRNvv13LgCKmZUsFFCIiUpDUUpEGS5f6eh1vvgkPPeSBhYiI\nSEOjoKKeli3z1UXffBNeegkOOqiGA5YsgXXXBbOs1E9ERCRb1M5eD8uXw4knwmuvwfPPpxBQlJXB\nTjv5zA4REZEio6Cijlas8AXBXnkFRo+GP/+5mp0rKjyVZseOPoLzgAOyVk8REZFsUfdHHaxcCSUl\nMGYMPPssHHJINTuXlUHv3jB/vued6NNHAzFFRKQo6e5WS6tWQbdu8Nxz8NRTcNhhVewYnxWzVStl\nxRQRkaKnlopaWLXKs2M+84wvWX7kkVXsGII3X7z/vlonRESkwVBQkaLVq6FXL3jiCRg1qoa1vczg\nxhthm22UFVNERBoMBRUpqKz0xFalpf44/vgUDqpxKoiIiEhxUVBRg8pKOOssX8Pj4Yd9TQ8RERH5\nNQUV1QjBx1o++KCnligpyXWNRERE8pdGD1YhBDjvPBg61IOKNVYarajwaOPxx3NWPxERkXyjoCKJ\nEGDAABg82IOKNdbyKCuDNm286WLJklxVUUREJO8oqEgQAlx0Edx+OwwZAn/5S/REYlbM6dN9OoiI\niIgAGlOxhhDgssvg1ls9vcTZZ0dPjB/v0z+UFVNERKRKujNGQoCrroKbb4bbbvPklwD87W/QqZOy\nYoqIiNRAd8fItdfC9dfDv/4F/fvHPbH77t46MX68ElmJiIhUQ90fwHXXwTXXeBLMCy9MeFKJKURE\nRFLS4FsqbrzRuz2uuw4uvTTXtRERESlcDTqo+Ne/4PLLvZXiiityXRsREZHCljdBhZn1M7NZZrbU\nzCab2Z417N/BzMrNbJmZzTSzHrUp77bb4OKL4Ypus/n7PuPqV/kGqrS0NNdVKDq6puml65l+uqbp\nVWzXMy+CCjM7BbgFuBrYHXgPGGtmm1Wxf0vgeWA8sCtwO3C/mR2SSnl33eXJrS7dbSz/90gr7KEH\n6/8iGqBi+2PIB7qm6aXrmX66pulVbNczL4IKYABwbwhhRAjhY+AcYAnQu4r9+wBfhBAuCSF8EkIY\nDDwZnadaTz7p6bcv3HAoA2eegN11Fzz6aLpeh4iISIOV86DCzNYG2uGtDgCEEALwCtC+isP2iZ6P\nN7aa/f9n4EDoz238c/dS7H3lnRAREUmXfLibbgY0AuYmbJ8LtKjimBZV7L+hmTWurrCTGz3FoDvX\nxsqUd0JERCSdGlKeiiYAR9+8GVP3bQ/TpuW6PgVvwYIFTJkyJdfVKCq6puml65l+uqbpVSjXc8aM\nGbFvm1S3n3lPQ+5E3R9LgBNCCKPjtg8DmoUQjktyzH+A8hDCBXHbegKDQggbV1FOV+CR9NZeRESk\nQekWQqhyIGLOWypCCCvNrBzoCIwGMDOLfr6jisMmAV0Sth0aba/KWKAbMBtYVo8qi4iINDRNgJb4\nvbRKOW+pADCzk4Fh+KyPt/FZHCcCO4QQ5pnZQGDLEEKPaP+WwPvA3cCDeAByG3B4CCFxAKeIiIhk\nQc5bKgBCCKOinBTXAs2BaUDnEMK8aJcWwDZx+882syOAQcD5wNfAGQooREREcicvWipERESk8OXD\nlPJs+68AAAc0SURBVFIREREpAgoqREREJC2KJqjI9oJkDUFtrqmZHWdm48zsezNbYGYTzezQbNa3\nENT2fRp33H5mttLM8n9CexbV4e9+HTO73sxmR3/7X0TT0SVSh2vazcymmdliM5tjZg+Y2SbZqm8+\nM7MDzGy0mX1jZpVmdnQKxxT0vakogopsL0jWENT2mgIHAuPwqb5tgVeB58xs1yxUtyDU4ZrGjmsG\nDOfXqekbtDpezyeAg4FewPZACfBJhqtaMOrwv3Q//L15H7ATPmtvL2BoViqc/9bDJx70BWocwFgU\n96YQQsE/gMnA7XE/Gz4j5JIq9r8JmJ6wrRR4MdevJV8etb2mVZzjA+DKXL+WfHnU9ZpG781/4P/o\np+T6deTLow5/94cBPwIb5bru+fqowzW9EPg0Ydu5wH9z/Vry7QFUAkfXsE/B35sKvqUi2wuSNQR1\nvKaJ5zBgA/yfeINX12tqZr2AVnhQIZE6Xs+jgHeBS83sazP7xMz+aWbVph1uKOp4TScB25hZl+gc\nzYGTgBcyW9uiVfD3poIPKsjygmQNRF2uaaKL8aa/UWmsVyGr9TU1sz8CN+BpcSszW72CU5f36HbA\nAcDOwLFAf7y5fnCG6lhoan1NQwgTgdOAx81sBfAt8BPeWiG1V/D3pmIIKiTPROusXAWcFEKYn+v6\nFCIzWwtfq+bqEMLnsc05rFIxWAtvgu4aQng3hDAGuADoUSj/sPONme2E9/tfg4+l6oy3rN2bw2pJ\nDuVFRs16mg+sxjNxxmsOfFfFMd9Vsf/CEMLy9FavINXlmgJgZqfig7RODCG8mpnqFaTaXtMNgD2A\n3cws9kl6LbxnaQVwaAjhtQzVtRDU5T36LfBNCGFR3LYZeLC2NfB50qMajrpc08uAN0MIt0Y/f2Bm\nfYEJZnZFCCHxU7dUr+DvTQXfUhFCWAnEFiQD1liQbGIVh02K3z9S04JkDUYdrylmVgI8AJwafQqU\nSB2u6ULgT8Bu+CjwXYEhwMfR929luMp5rY7v0TeBLc2sady21njrxdcZqmrBqOM1bQqsSthWic90\nUMta7RX+vSnXI0XT8QBOxpdPPx3YAW96+wHYPHp+IDA8bv+WQAU+0rY1Pt1nBdAp168lXx51uKZd\no2t4Dh5Zxx4b5vq15Mujttc0yfGa/VGP64mP8fkSeBzYEZ8G/QkwJNevJV8edbimPYDl0d99K2A/\nfFHIibl+LfnwiN5zu+IfDiqBv0Y/b1PF9Sz4e1POK5DGX15ffFnzpXhUt0fccw8BZQn7H4hH5UuB\nT4HuuX4N+faozTXF81KsTvJ4MNevI58etX2fJhyroKKe1xPPTTEWWBQFGDcDjXP9OvLpUYdr2g9f\nNXoR3uIzHPhtrl9HPjyAg/6/vbsHsaMKwzj+fxQFbQNiGkGjgiAGbUQXbNIJESwkCIIfIBaChQZE\nO7UREhEsVbTQwiaNoEIKtTDapAlIxIgriqi4hsWs8Quyr8WchbuXa3In9+xqwv8Hw9w5c+bM3Gb3\n4Z2Ze1qYmPl38WL83+SEYpIkqYsL/pkKSZL0/2CokCRJXRgqJElSF4YKSZLUhaFCkiR1YaiQJEld\nGCokSVIXhgpJktSFoULSlkuyK8l6m9WSJHva9pXnOlbShcNQIWkuSd5sQeBMW298vm7OIaZ/vnfL\nf87X8CJtr4th6nNJ2+cD4CE2z0C5Muex/8WslcEZM6VtY6VC0hh/VdVKVf08sVSSu5N8kmQ1yS9J\n3k1y7aInS/J4kq+T/JnkeJL7J/ZtuqXS2na0tjuT7AIOt11rrary6qLXJOnfGSok9XAFcAC4DdjD\nUBk4tMiASe4DXmKYHvpm4A3grSRLE91m3ULZaFtmmMobhmm5dwJPLnJNks7O2x+SxtibZG1i+/2q\n2ldVmwJEkkeBH5LcWFUnzvNcTwGvVdXrbftgkjuA/cCRjVPNOC4ArYKy2tpWqur387wOSXOyUiFp\njA+BW4DdbXkCIMkNSd5JspzkFPAVQ8XgmnMNmOSSJGttOZXklbbrJuDTqe5HWvuGLX/YU9L8rFRI\nGuN0VX0zo/094ATwCPAjcDlwrK3PqqrWk+yeaPp1zmtZZ6hKTFYrLpvzWElbwEqFpIUkuQq4Hnih\nqj6uqi+BHYx4hbSqlieWk635C2BpqusScLx93njrZOfE/lunzvN3W18615eRtBArFZIWdRJYBR5L\nssLwUOSLM/qNfa3zAPB2kmPAR8C9wD3AXQBV9VuSo8AzSb4Hrgaenxrj27bem+Qw8EdVnR55HZLm\nZKVC0kKq6gywD7gd+JwhDOyf1XXkuIcYHtZ8uo37MPBAVX020e1BhjdPjgIHgWenxvgOeI7hLZKf\ngJfHXIOkcVLlc06SJGlxViokSVIXhgpJktSFoUKSJHVhqJAkSV0YKiRJUheGCkmS1IWhQpIkdWGo\nkCRJXRgqJElSF4YKSZLUhaFCkiR1YaiQJEld/AOhNk3H3Z8rUwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x51f49c18>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Courbe ROC des données de test X et Y selon le classifieur KNN\n",
    "# Au mieux, le classifieur à une performance de 0.69 pour la prédiction du sexe (mais nécessitant un temps assez important)\n",
    "# Plus l'apprentissage se fait sur un nombre de données, plus l'indice de prédiction augmente.\n",
    "# Pour obtenir un tel résultat, le temps nécessaire est tout de même trop important.\n",
    "draw_roc_curve(knn, X_test, Y_test, \"Sex prediction by KNN\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAF4CAYAAACvlrkrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x552e4a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Matrice de confusion des données X_test sur lesquelles on fait la prédiction et des données Y_test.\n",
    "# La prédiction du modèle étant de 0.62, au mieux, est tout juste satisfaisant.\n",
    "# Le classifieur SVM semble pouvoir effectuer une prédiction assez correcte, mais le temps nécessaire est\n",
    "# la principale contrainte de ce classifieur.\n",
    "draw_confusion_matrix(Y_test, knn.predict(X_test), \"\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2.4) CLF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y.Sex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------ \n",
      " clf_grid_search\n",
      "{'n_estimators': 10, 'min_samples_split': 2, 'learning_rate': 0.01, 'max_depth': None, 'loss': 'ls'}\n",
      " best_score_ = 0.77 with =  GradientBoostingClassifier(init=None, learning_rate=0.1, loss='deviance',\n",
      "              max_depth=3, max_features=None, max_leaf_nodes=None,\n",
      "              min_samples_leaf=1, min_samples_split=2,\n",
      "              min_weight_fraction_leaf=0.0, n_estimators=100,\n",
      "              presort='auto', random_state=None, subsample=1.0, verbose=0,\n",
      "              warm_start=False) \n",
      "\n",
      "Duration time : 122.983999968\n"
     ]
    }
   ],
   "source": [
    "#100   rows : 2.3s / 0.71 pred / 0.65 acc\n",
    "#1000  rows : 18s  / 0.82 pred / 0.75 acc\n",
    "#5000  rows : 122s / 0.85 pred / 0.78 acc\n",
    "nb_ligne = 5000\n",
    "start_time = print_time()\n",
    "\n",
    "clf = clf_grid_search(X_train[:nb_ligne], Y_train[:nb_ligne])\n",
    "\n",
    "end_time = print_time()\n",
    "print \"\\nDuration time :\", end_time - start_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhUAAAGHCAYAAAAHoqCrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x52d2da20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Courbe ROC des données de test X et Y selon le classifieur CLF\n",
    "# Au mieux, le classifieur à une performance de 0.85 pour la prédiction du sexe\n",
    "# La classification se fait plutôt correctement au vue de la performance du classifieur\n",
    "draw_roc_curve(clf, X_test, Y_test, \"Sex prediction by RF\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAF4CAYAAACvlrkrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x51aad4e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Matrice de confusion des données X_test sur lesquelles on fait la prédiction et des données Y_test.\n",
    "# Le classifieur CLF est  satisfaisant sur nos données\n",
    "draw_confusion_matrix(Y_test, clf.predict(X_test), \"\")\n",
    "\n",
    "# Le Classifier Gradient Boosting semble être le meilleur modèle de prédiction que nous avons testé jusqu'ici\n",
    "# La prédiction et la précision sont les plus élevées que nous avons enregistré jusqu'ici"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [Root]",
   "language": "python",
   "name": "Python [Root]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}