Support Vector Machine.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:7a7490558bf2a4b4ae626a6b6fba29ad9dc0109a03822e610b7e4a330dbf09db"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "heading",
     "level": 2,
     "metadata": {},
     "source": [
      "Detect fake profiles in online social networks using Support Vector Machine"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import sys\n",
      "import csv\n",
      "import datetime\n",
      "import numpy as np\n",
      "import pandas as pd\n",
      "import matplotlib.pyplot as plt\n",
      "from datetime import datetime\n",
      "import sexmachine.detector as gender\n",
      "from sklearn.preprocessing import Imputer\n",
      "from sklearn import cross_validation\n",
      "from sklearn import metrics\n",
      "from sklearn import preprocessing\n",
      "from sklearn.linear_model import LinearRegression\n",
      "from sklearn.svm import SVC\n",
      "from sklearn.metrics import roc_curve, auc\n",
      "from sklearn.cross_validation import StratifiedKFold, train_test_split\n",
      "from sklearn.grid_search import GridSearchCV\n",
      "from sklearn.metrics import accuracy_score\n",
      "from sklearn.learning_curve import learning_curve\n",
      "from sklearn.metrics import classification_report\n",
      "from sklearn.metrics import confusion_matrix\n",
      "%matplotlib inline "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 57
    },
    {
     "cell_type": "heading",
     "level": 6,
     "metadata": {},
     "source": [
      "function for reading dataset from csv files"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def read_datasets():\n",
      "    \"\"\" Reads users profile from csv files \"\"\"\n",
      "    genuine_users = pd.read_csv(\"data/users.csv\")\n",
      "    fake_users = pd.read_csv(\"data/fusers.csv\")\n",
      "    # print genuine_users.columns\n",
      "    # print genuine_users.describe()\n",
      "    #print fake_users.describe()\n",
      "    x=pd.concat([genuine_users,fake_users])   \n",
      "    y=len(fake_users)*[0] + len(genuine_users)*[1]\n",
      "    return x,y\n",
      "    "
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 58
    },
    {
     "cell_type": "heading",
     "level": 6,
     "metadata": {},
     "source": [
      "function for predicting sex using name of person"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def predict_sex(name):\n",
      "    sex_predictor = gender.Detector(unknown_value=u\"unknown\",case_sensitive=False)\n",
      "    first_name= name.str.split(' ').str.get(0)\n",
      "    sex= first_name.apply(sex_predictor.get_gender)\n",
      "    sex_dict={'female': -2, 'mostly_female': -1,'unknown':0,'mostly_male':1, 'male': 2}\n",
      "    sex_code = sex.map(sex_dict).astype(int)\n",
      "    return sex_code"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 59
    },
    {
     "cell_type": "heading",
     "level": 6,
     "metadata": {},
     "source": [
      "function for feature engineering"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def extract_features(x):\n",
      "    lang_list = list(enumerate(np.unique(x['lang'])))   \n",
      "    lang_dict = { name : i for i, name in lang_list }             \n",
      "    x.loc[:,'lang_code'] = x['lang'].map( lambda x: lang_dict[x]).astype(int)    \n",
      "    x.loc[:,'sex_code']=predict_sex(x['name'])\n",
      "    feature_columns_to_use = ['statuses_count','followers_count','friends_count','favourites_count','listed_count','sex_code','lang_code']\n",
      "    x=x.loc[:,feature_columns_to_use]\n",
      "    return x"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 62
    },
    {
     "cell_type": "heading",
     "level": 6,
     "metadata": {},
     "source": [
      "function for ploting learning curve"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def plot_learning_curve(estimator, title, X, y, ylim=None, cv=None,\n",
      "                        n_jobs=1, train_sizes=np.linspace(.1, 1.0, 5)):\n",
      "    plt.figure()\n",
      "    plt.title(title)\n",
      "    if ylim is not None:\n",
      "        plt.ylim(*ylim)\n",
      "    plt.xlabel(\"Training examples\")\n",
      "    plt.ylabel(\"Score\")\n",
      "    train_sizes, train_scores, test_scores = learning_curve(\n",
      "        estimator, X, y, cv=cv, n_jobs=n_jobs, train_sizes=train_sizes)\n",
      "    train_scores_mean = np.mean(train_scores, axis=1)\n",
      "    train_scores_std = np.std(train_scores, axis=1)\n",
      "    test_scores_mean = np.mean(test_scores, axis=1)\n",
      "    test_scores_std = np.std(test_scores, axis=1)\n",
      "    plt.grid()\n",
      "\n",
      "    plt.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
      "                     train_scores_mean + train_scores_std, alpha=0.1,\n",
      "                     color=\"r\")\n",
      "    plt.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
      "                     test_scores_mean + test_scores_std, alpha=0.1, color=\"g\")\n",
      "    plt.plot(train_sizes, train_scores_mean, 'o-', color=\"r\",\n",
      "             label=\"Training score\")\n",
      "    plt.plot(train_sizes, test_scores_mean, 'o-', color=\"g\",\n",
      "             label=\"Cross-validation score\")\n",
      "\n",
      "    plt.legend(loc=\"best\")\n",
      "    return plt"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 63
    },
    {
     "cell_type": "heading",
     "level": 6,
     "metadata": {},
     "source": [
      "function for plotting confusion matrix"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def plot_confusion_matrix(cm, title='Confusion matrix', cmap=plt.cm.Blues):\n",
      "    target_names=['Fake','Genuine']\n",
      "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
      "    plt.title(title)\n",
      "    plt.colorbar()\n",
      "    tick_marks = np.arange(len(target_names))\n",
      "    plt.xticks(tick_marks, target_names, rotation=45)\n",
      "    plt.yticks(tick_marks, target_names)\n",
      "    plt.tight_layout()\n",
      "    plt.ylabel('True label')\n",
      "    plt.xlabel('Predicted label')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 65
    },
    {
     "cell_type": "heading",
     "level": 6,
     "metadata": {},
     "source": [
      "function for plotting ROC curve"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def plot_roc_curve(y_test, y_pred):\n",
      "    false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test, y_pred)\n",
      "    print \"False Positive rate: \",false_positive_rate\n",
      "    print \"True Positive rate: \",true_positive_rate\n",
      "    \n",
      "    roc_auc = auc(false_positive_rate, true_positive_rate)\n",
      "\n",
      "    plt.title('Receiver Operating Characteristic')\n",
      "    plt.plot(false_positive_rate, true_positive_rate, 'b',\n",
      "    label='AUC = %0.2f'% roc_auc)\n",
      "    plt.legend(loc='lower right')\n",
      "    plt.plot([0,1],[0,1],'r--')\n",
      "    plt.xlim([-0.1,1.2])\n",
      "    plt.ylim([-0.1,1.2])\n",
      "    plt.ylabel('True Positive Rate')\n",
      "    plt.xlabel('False Positive Rate')\n",
      "    plt.show()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 71
    },
    {
     "cell_type": "heading",
     "level": 6,
     "metadata": {},
     "source": [
      "Function for training data using Support Vector Machine"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def train(X_train,y_train,X_test):\n",
      "    \"\"\" Trains and predicts dataset with a SVM classifier \"\"\"\n",
      "    # Scaling features\n",
      "    X_train=preprocessing.scale(X_train)\n",
      "    X_test=preprocessing.scale(X_test)\n",
      "\n",
      "    Cs = 10.0 ** np.arange(-2,3,.5)\n",
      "    gammas = 10.0 ** np.arange(-2,3,.5)\n",
      "    param = [{'gamma': gammas, 'C': Cs}]\n",
      "    cvk = StratifiedKFold(y_train,n_folds=5)\n",
      "    classifier = SVC()\n",
      "    clf = GridSearchCV(classifier,param_grid=param,cv=cvk)\n",
      "    clf.fit(X_train,y_train)\n",
      "    print(\"The best classifier is: \",clf.best_estimator_)\n",
      "    clf.best_estimator_.fit(X_train,y_train)\n",
      "    # Estimate score\n",
      "    scores = cross_validation.cross_val_score(clf.best_estimator_, X_train,y_train, cv=5)\n",
      "    print scores\n",
      "    print('Estimated score: %0.5f (+/- %0.5f)' % (scores.mean(), scores.std() / 2))\n",
      "    title = 'Learning Curves (SVM, rbf kernel, $\\gamma=%.6f$)' %clf.best_estimator_.gamma\n",
      "    plot_learning_curve(clf.best_estimator_, title, X_train, y_train, cv=5)\n",
      "    plt.show()\n",
      "    # Predict class\n",
      "    y_pred = clf.best_estimator_.predict(X_test)\n",
      "    return y_test,y_pred"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 72
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"reading datasets.....\\n\"\n",
      "x,y=read_datasets()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "reading datasets.....\n",
        "\n"
       ]
      }
     ],
     "prompt_number": 76
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"extracting featues.....\\n\"\n",
      "x=extract_features(x)\n",
      "print x.columns\n",
      "print x.describe()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "extracting featues.....\n",
        "\n",
        "Index([u'statuses_count', u'followers_count', u'friends_count',\n",
        "       u'favourites_count', u'listed_count', u'sex_code', u'lang_code'],\n",
        "      dtype='object')"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "       statuses_count  followers_count  friends_count  favourites_count  \\\n",
        "count     2818.000000      2818.000000    2818.000000       2818.000000   \n",
        "mean      1672.198368       371.105039     395.363023        234.541164   \n",
        "std       4884.669157      8022.631339     465.694322       1445.847248   \n",
        "min          0.000000         0.000000       0.000000          0.000000   \n",
        "25%         35.000000        17.000000     168.000000          0.000000   \n",
        "50%         77.000000        26.000000     306.000000          0.000000   \n",
        "75%       1087.750000       111.000000     519.000000         37.000000   \n",
        "max      79876.000000    408372.000000   12773.000000      44349.000000   \n",
        "\n",
        "       listed_count     sex_code    lang_code  \n",
        "count   2818.000000  2818.000000  2818.000000  \n",
        "mean       2.818666    -0.180270     2.851313  \n",
        "std       23.480430     1.679125     1.992950  \n",
        "min        0.000000    -2.000000     0.000000  \n",
        "25%        0.000000    -2.000000     1.000000  \n",
        "50%        0.000000     0.000000     1.000000  \n",
        "75%        1.000000     2.000000     5.000000  \n",
        "max      744.000000     2.000000     7.000000  \n"
       ]
      }
     ],
     "prompt_number": 77
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"spliting datasets in train and test dataset...\\n\"\n",
      "X_train,X_test,y_train,y_test = train_test_split(x, y, test_size=0.20, random_state=44)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "spliting datasets in train and test dataset...\n",
        "\n"
       ]
      }
     ],
     "prompt_number": 78
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"training datasets.......\\n\"\n",
      "y_test,y_pred = train(X_train,y_train,X_test)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "training datasets.......\n",
        "\n",
        "('The best classifier is: ', SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0,\n",
        "  decision_function_shape=None, degree=3, gamma=31.622776601683793,\n",
        "  kernel='rbf', max_iter=-1, probability=False, random_state=None,\n",
        "  shrinking=True, tol=0.001, verbose=False))"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "[ 0.91796009  0.92904656  0.92461197  0.9556541   0.93777778]"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "\n",
        "Estimated score: 0.93301 (+/- 0.00651)\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAEbCAYAAAAbCrnpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x7f385ca44a50>"
       ]
      }
     ],
     "prompt_number": 79
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print 'Classification Accuracy on Test dataset: ' ,accuracy_score(y_test, y_pred)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Classification Accuracy on Test dataset:  0.904255319149\n"
       ]
      }
     ],
     "prompt_number": 80
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cm=confusion_matrix(y_test, y_pred)\n",
      "print('Confusion matrix, without normalization')\n",
      "print(cm)\n",
      "plot_confusion_matrix(cm)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " Confusion matrix, without normalization\n",
        "[[262   6]\n",
        " [ 48 248]]\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAVQAAAEpCAYAAADMLtqqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x7f385ce8afd0>"
       ]
      }
     ],
     "prompt_number": 82
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "cm_normalized = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
      "print('Normalized confusion matrix')\n",
      "print(cm_normalized)\n",
      "plot_confusion_matrix(cm_normalized, title='Normalized confusion matrix')"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Normalized confusion matrix\n",
        "[[ 0.97761194  0.02238806]\n",
        " [ 0.16216216  0.83783784]]\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAVEAAAEpCAYAAAAqBxHuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\no04AAAAASUVORK5CYII=\n",
       "text": [
        "<matplotlib.figure.Figure at 0x7f385d948410>"
       ]
      }
     ],
     "prompt_number": 83
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print(classification_report(y_test, y_pred, target_names=['Fake','Genuine']))"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "             precision    recall  f1-score   support\n",
        "\n",
        "       Fake       0.85      0.98      0.91       268\n",
        "    Genuine       0.98      0.84      0.90       296\n",
        "\n",
        "avg / total       0.91      0.90      0.90       564\n",
        "\n"
       ]
      }
     ],
     "prompt_number": 84
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plot_roc_curve(y_test, y_pred)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "False Positive rate:  [ 0.          0.02238806  1.        ]\n",
        "True Positive rate:  [ 0.          0.83783784  1.        ]\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEZCAYAAAB4hzlwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xm8lGX9//HXh01AdjFRQCkl9zVF08qjmaJWauVC7laa\nuX1N+pqmSVlpaZZIEu7LT6HNCs00s06Zmn6NRcwFzExAllgFDnLA8/n9cV0z3Axn5syBM3PP8n4+\nHvM4c99zz31/7jnnXJ+5lvu6zd0REREB6JR2ACIiUjmUFEREJEtJQUREspQUREQkS0lBRESylBRE\nRCRLSUHaZGYvmdnH0o6jUpjZFWZ2e0rHvsfMrk3j2B3NzE41s8c38b36mywRJYUqY2ZvmlmTma0w\ns3lmdreZbVnKY7r7Hu7+11IeI8PMtjCz68zsP/E8Z5rZ6HIcO088DWY2O7nO3a9z9y+V6HhmZheb\n2QwzW2lms83s52a2R+bw8ZEqMxtjZvdvzj7c/QF3P6qIY22UCMv5N1lvlBSqjwOfdPfewD7AvsAV\n6YbUfmbWJc9LvwAOA44GegGnA+ea2c0liMHMzDp6v5vpZuBi4CKgP/BB4DfAMR19IDPr3NH7rIZj\nSxvcXY8qegD/Bg5PLP8AeCSxfBDwDLAUmAYcmnhtAHA3MBdYAvw68don4/ZLgaeBPROvvQkcDmwH\nNAH9E6/tC/wX6ByXzwFejvt/DNg+sW0L8BVgFvCvVs7t48BqYHDO+hHAOuADcbkRuA54DlhOKDT7\nF/kZNALfiefYBOwInB1jfgf4F3Bu3HbLGM97wIr4+rbAGOD+uM2weF5nAP+Jn8WVieP1AO6Nn8fL\nwP8Cs/P8bofH89y/wO//bmAc8EiM5++ZzyW+fjPwVvxcXgA+knhtDPBL4P74+jnAAcCz8bN6G7gF\n6Jp4z+7AE8BiYD7hC8hRwBqgOX4uU+O2fYE7437mANcCneJrZ8XP/CZgUXztLOCp+LoBPwIWxNhe\njMc+Nx5nTTzWbxN/kx+PzzsDVwKvx8/kBWBI2v+r1fpIPQA92vkLC0kh888wJP7z/CguD47/cCPj\n8hFxeau4/DtgYvzn7QJ8NK7fN/4zHhD/Oc+Ix+maOObh8fmTwBcT8dwA3BqfH0co8Hcm1EK/ATyd\n2LYFeBzoB2zRyrldD/w5z3m/CXwpPm+Mhc5uQM9MQVfkZ9AY97VrjLEL4Vv4++PrHwNWAfvG5UPJ\nKcSBa9g4KUwAtgD2At4Fdk6eU/zMB8ff11t5zvHLwL/b+P3fE89n/1gY/j9gYuL1Uwk1jE7AV4F5\nQLf42hhCAfvpuNwd2I+QdDsBOxAS1yXx9d7x/ZcC3Qg1txGJz+C+nNh+DYwnJMKtCUk7k2DPAtYC\nF8RjdWfDpHAUoTDvE5d3BgbF53cD327l/yDzN/m1+LkOj8t7AgPS/l+t1oeaj6qPAb8xs3cI3wgX\nEP5BAU4DHnX3xwDc/Y+Ef7RjzWxbYCTwZXdf7u7r3P2p+L5zgQnu/n8e3Ef4ZnZQK8d/EBgFofkF\nODmug1CoXefur7l7C+Hb/D5mNjTx/uvcfZm7r2ll3wMJ30ZbMy++DqEJ7T53f9ndm4CrgZPMrFOh\nzyDx3nvc/RV3b4mfw6Pu/u+4/V+BPwAfjdu31rzU2rpvufsad38RmA7sHdefCHwvfuZzCd/k8zVZ\nbVXg/DMceMjdX3D394AHCM2IxPgfcPel8dxuIiSqnRPvf8bdJ8dt33X3Ke7+fNz+P8BthEQIofb4\ntrv/yN2b3X2luz+f+Ayy52Fm2xCa/C5199Xu/l/gx8ApiWO/7e4/icd6N+e81hKS0K5m1in+DSU/\ni0LNfF8EvuHus+J5zXD3JQW2lwKUFKqPA8e5ex+ggfCNd+v42g7AiWa2NPMADgEGAUOBJe6+vJV9\n7gBclvO+IYTmolwPAR82s0GEb9Ut7v63xH5uTuxjcVw/OPH+DTptc/yX0DzTmu0I35Bb289bQFdC\n0ij0GbQag5kdbWZ/N7PFcftjCAV0eyQLsCbCt+pM3MnjzSmwj8XkP/+kBYnnqxPHwsxGm9nLZrYs\nnktf1ifTjY5vZh80s0fioIXlwHdZf+5DgTeKiAfC594VmJf43H/K+r9NKPC7d/c/EZrFfgIsMLMJ\nZta7yGMPITT7SQdQUqhi8VvtPcCNcdVbhGaN/olHb3f/AeEfcoCZ9W1lV28B3815Xy93/1krx1xK\n+CZ9MvB5QnNUcj/n5uxnS3f/e3IXBU7pj8CBZjYkudLMDiT84/8psXr7nOdrCUml0GewUQxmtgXw\nK0LfzPvcvT/wKOu/mbYWb3tG/8wjFK4ZQ/NtSGiaG2JmH2rH/rPM7KOEppQT3b1fPJflbPgtOzf2\n8YQmo53cvS+hyS9TLrwFfCDP4VpylmcTapdbJT73vu6+Z4Fjb8Ddb3H3/QnNgh+M59Lm++Kxd2pj\nGymSkkL1+zHwCTPbi9C+/CkzO9LMOptZ9zikcrC7zwN+D9xqZv3MrGtinPftwJfNbEQckLOlmR1r\nZr3yHPNB4Ezgs6xvOoLwzfBKM9sNwMz6mtmJxZ6Iuz9JKBh/ZWa7xXM4iNAxequ7Z74NGnCame1q\nZj2BbwO/cHcv9BkkDpUsJLvFxyKgxcyOBo5MvL4A2MrM+uR5f1t+DlwRP/PBwIXkKeRi88etwEQz\nO9TMusX4TzGzy4s4dm9CR/Wi+N5vAn0KbA+hlrECaDKzXYDzE6/9DtjWzC6JQ4V7m9mI+NoCYFhm\n9Fb8+/oDcFPcrpOZ7WhFXktgZvub2YFm1pVQ03qX0MGfOVa+5ARwB3Ctme0U/373MrMBxRxXNqak\nUOXcfRFwH3C1u88hdPZeCSwkfNO7jPW/59MJ36hfJfyjXRz38Q/gS4Tq+xJCZ/EZ5P+GNpnwzWye\nu89IxPIb4PvApNgUMYPQgZjdpIhT+iyhY/YxQmF1P3CHu1+Us5/7CbWkeYRCPXMu+T6DVr8tu/uK\n+N6fx3MfBfw28fqrhNrQG2a2JPbN5F4rUOi8vk1osvk3odD8BaGzt1XufjHrm1GWEkbUHEf4zDPH\nyj1eZvmx+JhJ6ExfTTj/5Ha57x1NqPG9Q+hPmJTZJn42nwA+RficZxKaLInnAbDYzF6Iz88g/C4y\no89+wfpmu3xxZ9b1icdfEmNfRBjEAGFE026xWeohNnYT4ff3B0LN6HZCR7ZsAgtfrkq0c7O7CB18\nC3OqkZnXTyUM0TNCAXB+7KgTycvM/kxoIror7Vjay8zOB05y98PSjkWkNaWuKdxNGPGSzxvAx9x9\nL8K45dtKHI/Ujkq76KxVZjbIzA6JzSk7E4aJ/jrtuETyyXdVaYdw96fMbFiB159NLD5H6EwUKUbq\nUz0UqRuhr+X9wDJCU9StqUYkUkBJk0I7fYEw6kOkoGpqenH3twgXU4lUhYpICmZ2GOGS+0PSjkVE\npJ6lnhTiUMrbCdMSLM2zTbU0FYiIVBR3b1f/W6pDUs1se8IVsqe5++uFts03T0dHPq655prU5x3R\nudT++dTSudTa+dTSubhv2nfpktYUzGwiYR6VgRbmpL+GcCk87j4B+CZh8q7x8RqYte4+Is/uRESk\nxEo9+mhUG69/kTCZlYiIVABd0ZzQ0NCQdggdppbOBWrrfGrpXKC2zqeWzmVTlfSK5o5iZl4NcYqI\nVBIzw6upo1lERCqLkoKIiGQpKYiISJaSgoiIZCkpiIhIlpKCiIhkKSmIiEiWkoKIiGQpKYiISJaS\ngoiIZCkpiIhIlpKCiIhkKSmIiEiWkoKIiGQpKYiISJaSgoiIZCkpiIhIlpKCiIhkKSmIiEiWkoKI\niGQpKYiISJaSgoiIZCkpiIhIlpKCiIhklTQpmNldZrbAzGYU2Gasmc0ys+lmtm8p4xERkcJKXVO4\nGxiZ70UzOwbYyd2HA+cC40scj4iIFFDSpODuTwFLC2zyaeDeuO1zQD8z26aUMYmISH5dUj7+YGB2\nYnkOMARYkE44IlIvVq2COXNg9uzwc84c6N8fLrgg7cjSlXZSALCcZW9tozFjxmSfNzQ00NDQULqI\nRKSqrVy5cYGf+3zNGhgyZP1j6FDYZZe0I988jY2NNDY2btY+zL3VMrjDmNkw4GF337OV134KNLr7\npLj8KnCouy/I2c5LHaeIVIcVKwoX9nPmQHNzKOQzhX2y4M88HzAAbPo0GDgwrKhBZoa7537xLijt\nmsJk4EJgkpkdBCzLTQgiUj+WL2+9sE8ur1sXCvdkAX/AAXDCCeuX+/cHK1QUNjfDmO/C+PHw4IM1\nmxQ2RUlrCmY2ETgUGEjoJ7gG6Arg7hPiNuMII5RWAWe7+5RW9qOagkgVc89f4Ceft7RsXODnPu/b\nt40Cvy3TpsFZZ4Ud3nYbbLddR51mxdmUmkLJm486gpKCSOVyh2XLChf2c+aEgrxQYT9kCPTps5kF\nflu+8x0YOxZuvBFOP73EB0tfNTYfiUgFc4elS9su8Dt33riQ/8hHNi7wUzd8eKgp1HDtYHOppiBS\np9xhyZK2O2232KLwt/vBgyukwJeNqPlIRIBQ4C9a1HpHbfJ5jx4bF/a5y716pX02sqmUFETqgDv8\n97+FC/u5c6Fnz8JDMocMgS23TPtsSqC5Gb77XRgxAo49Nu1oUqWkIFLlWlrWF/j5mnXmzg3f3gt1\n2g4eHJJC3amjkUXFUFIQqWAtLbBwYf7x95kCv0+ftgv8Hj3SPpsKk6kdjB9fNyOLiqHRRyIpee+9\njQv83Ofz5kG/fhsX9nvuuWGnbffuaZ9NFTrppHBVm0YWbTbVFETa8N57MH9+4U7b+fPDVbSFOm23\n204FfsnMnw/bbKPaQQ41H4m003vvhW/whTpt58+HrbYq3Gm73XZh6KZIJVFSEElYt259gZ+vWWfB\ngjAfWqEROtttB926pX02AoS+g5YWVbmKpKQgdWPt2rYL/IULYeutC3fabrstdO2a9tlIUTIjiy66\nCL7whbSjqQpKClIT1q6Ft98u3Gm7aFFoQi50pe2gQSrwa4JGFm0yjT6SitfcHIZdFpoeefHiUKAn\nC/lhw8JcOskCv4v+emtf8roDjSwqC9UUpMOsWZO/wM88X7IkNNkU6rQdNChMsCbCl78MBx+s2sEm\nUvORlMy7764v8PM16yxbFgr8Qp2222yjAl+kXJQUZJOsXt12gb98eai5F+q0fd/7oFOntM9GRDKU\nFGQjTU2hwC/UabtyZbiSNt+3+6FDwygeFfhSMlOnhgmdhg9PO5KaoqRQZ1atyn+VbWZ51ar8UyJn\nng8cqAJfUpIcWXTvvXD00WlHVFOUFGrIypVtz4W/enXhwj5T4Kt/TirS1KlhZNH228OECRpZVAJK\nClVixYq2b2C+Zk3bc+FvtZUKfKlS110HP/oR/PCHcNpp+kMuESWFCvDOO/mbcjLP164tXOAPHRom\nV9P/idSshx+GD31ItYMSU1IoIffWC/zc5y0thb/dDx0apk9WgS8ipaak0MHeew8+9zl47bVQ4EPh\nb/dDhkDfvirwRaQyKCl0sGXLQkH/3HPrC3wRKVJmZNEuu8CoUWlHU5c2JSloIGIBTU3QuzfsvrsS\ngki7TJ0KBxwAU6bAoYemHY20Q0mTgpmNNLNXzWyWmV3eyut9zexhM5tmZi+Z2VmljKe9Vq2CLbdM\nOwqRKtLcDNdcA0cdBaNHw+TJ6kyuMiWbZ9LMOgPjgCOAucD/mdlkd38lsdkFwEvu/ikzGwi8Zmb/\nz93XlSqu9mhqgp49045CpIqceWa4yEYzmlatUk4+PAJ43d3fBDCzScBxQDIptAB94vM+wOJKSQig\npCDSbuPGwYABGm1RxUrZfDQYmJ1YnhPXJY0DdjOzt4HpwCUljKfdVq1SUhBpF11RWfWKrimYWU93\nb2rHvosZLjQSmOLuh5nZjsATZra3u6/I3XDMmDHZ5w0NDTQ0NLQjlE3T1KQ+BZFWNTeHqzD1D1JR\nGhsbaWxs3Kx9tDkk1cwOBu4Aerv7UDPbBzjX3b/SxvsOAsa4+8i4fAXQ4u7fT2zzCHCduz8dl58E\nLnf3F3L2lcqQ1EmT4Ne/hp/9rOyHFqlcmTmLvvQluPDCtKORAko1JPXHhG/0iwDcfRpQzBizF4Dh\nZjbMzLoBJwOTc7Z5i9ARjZltA+wMvFFc6KWnPgWRhNyRRRdckHZEUgJFNR+5+1u2YTthm53B7r7O\nzC4EHgc6A3e6+ytmdl58fQJwLXCPmb0IGPC/7r6knedQMupTEImSM5pqZFFNKyYpvGVmhwDEb/wX\ns+EIorzc/ffA73PWTUg8nwccVXS0ZaY+BZFo4sRQO9CMpjWvmKRwPnAzYeTQXOAPhOsLap6aj0Si\nH/wg7QikTIpJCh90988nV8Saw9OlCalyNDWF21CKiNSLYjqaxxW5ruZomgupO1OnwvTpaUchKcpb\nUzCzDwMHA1ub2VcJHcEAvamTifTUfCR1I3mv5DvvhL33TjsiSUmh5qNuhATQOf7MeAf4XCmDqhRK\nClIXNLJIEvImBXf/C/AXM7snM39RvVFSkJp3442hE1n3SpaomI7mJjO7EdgN6BHXubsfXrqwKoP6\nFKTm7befageygWL6Bh4AXgU+AIwB3iRcrVzzVFOQmnf44UoIsoFiksJW7n4H0Ozuf3H3s4GaryWA\nkoKI1J9ikkJz/DnfzD5pZvsB/UsYU8VQUpCakJmzaMKEtreVuldMUviumfUDLgNGE2ZMvbSkUVUI\n9SlI1UveK/lTn0o7GqkCbU6d3eqbzA7JTHddDmlNnd2nD8yeDX37lv3QIpsned2BRhbVrU2ZOrvQ\nxWtdgJOA7YDH3P0lM/sUcAXQE9hnc4KtdO5qPpIqdv75sHChRhZJu+WtKZjZvcAQ4HngQGAe8CHg\n6+7+m7JFSDo1hebm0HS0dm1ZDyvSMZYvD1Vd1Q7qWofWFID9gT3dvcXMugPzgR3dffHmBFkt1J8g\nVU1tnrKJCnU0N7t7C4C7vwv8u14SAqjpSKpEczMsW5Z2FFJDCiWFXcxsRuYB7JxYfrFcAaZFSUEq\nXmZk0W23pR2J1JBCzUe7li2KCqSkIBWrtZFFIh2k0IR4b5YxjoqjPgWpSJrRVEqsmAnx6pJqClKR\nfvc73StZSkpJIQ8lBalIV12VdgRS44q6g5qZ9TSznUsdTCVR85GI1KM2k4KZfRqYCjwel/c1s8ml\nDixtqilIqqZOhWeeSTsKqUPF1BTGEK5oXgrg7lMJ91aoaUoKkorMjKZHHQULFqQdjdShYvoU1rr7\nMtuwU6ulRPFUDCUFKTuNLJIKUExN4Z9mdirQxcyGm9ktQFH1WjMbaWavmtksM7s8zzYNZjbVzF4y\ns8biQy8t9SlIWf34x6F2MHo0TJ6shCCpKaamcBHwDWANMJHQt3BtW28ys87AOOAIYC7wf2Y22d1f\nSWzTD/gJcJS7zzGzge0/hdJoaoJttkk7CqkbBx6o2oFUhGKSws7ufiVwZTv3PQJ4PXMRnJlNAo4D\nXkls83ngV+4+B8DdF7XzGCWj5iMpqw9/OO0IRIDimo9uik1A15rZHu3Y92BgdmJ5TlyXNBwYYGZ/\nNrMXzOz0duy/pJQURKQetZkU3L0BOAxYBEyIE+JdXcS+i7kBQldgP+AY4CjgajMbXsT7Sk59CtLh\nMiOLfvCDtCMRyauoK5rdfR5ws5n9Cbgc+CZt9yvMBYYmlocSagtJs4FF7r4aWG1mfwX2Bmbl7mzM\nmDHZ5w0NDTQ0NBQT+iZTTUE6VHJk0YQJaUcjNaqxsZHGxsbN2keb92g2s90It+X8HLAY+BnwS3df\n2Mb7ugCvAR8H3ibcwW1UTkfzLoTO6KOALYDngJPd/eWcfZX9zmuHHQZXXw2HH17Ww0qt0b2SJUUd\nfee1jLuASYQRQnOL3bG7rzOzCwmjlToDd7r7K2Z2Xnx9gru/amaPAS8Srn24PTchpEU1BekQo0fD\nv/+tkUVSNdqsKVSCNGoKe+wBEyfCnnuW9bBSa1atCt8uVDuQFHRoTcHMfuHuJ8a7ruVyd9+r3RFW\nEdUUpENotIJUmULNR5fEn58EcjNN5VcvNpOSgrRLczMsXw5bb512JCKbJe+QVHd/Oz79iru/mXwA\nXylLdClSUpCiZe6V/JOfpB2JyGYr5uK1I1tZd0xHB1JJ3Nc3BYvklZzRdPTo8FykyhXqUzifUCPY\nMadfoTfwdKkDS1NzM3TuDF27ph2JVCzNaCo1Ku/oIzPrC/QHridcsJbpV1jh7ovLE142lrKOPlq6\nFN7/fli2rGyHlGpzyy3Qr5+uO5CKtimjjwolhT7u/o6ZbUUrHcvuvmTTwmy/cieFuXNhxIjwU0Sk\nWnX0xWsTgWOBf9D6aKP3t+dA1UT9CSJSr/ImBXc/Nv4cVrZoKoRGHknW1KmweDEccUTakYiURZuj\nj8zsEDPrFZ+fbmY3mdkOpQ8tPUoKssHIoqVL045GpGyKGZL6U6DJzPYGvgq8AdxX0qhSpmmz61zm\nuoMpU8LIohNPTDsikbIpJimsc/cW4HjgJ+4+jjAstWapplDHbr1V90qWulbMLKkrzOxK4DTgo/He\nyzU9gl9JoY4dfLCuO5C6VkxN4WRgDXCOu88n3FLzhpJGlTIlhTq2zz5KCFLXirkd5zzgAaCfmX0S\neNfd1acgIlKDihl9dBLhjmgnEu7A9ryZ1XTPm2oKNS4zsuiqq9KORKTiFNOncBVwQOb2m2a2NfAk\n8ItSBpYmJYUapnslixRUTJ+CAf9NLC9m4/sr1BQlhRqUvO7gsss0skgkj2JqCo8Bj5vZg4RkcDLw\n+5JGlTL1KdSga66BGTM0skikDW0mBXf/mpl9BvhIXDXB3X9d2rDSpZpCDbrmGthiC81oKtKGQvdT\n+CBh6OlOwIvA19x9TrkCS5OSQg3q3j3tCESqQqE+hbuAR4DPAlOAsWWJqAIoKVSx5mZ4++22txOR\nVhVKCr3c/XZ3f9Xdb6CGp8rOpT6FKjVtWrgRxs03px2JSNUq1KfQ3cz2i88N6BGXDXB3n1Ly6FKi\nmkKVaW6G734Xxo+HG26AM85IOyKRqlUoKcwHflhg+bCSRFQBlBSqyLRp4bqDIUPCNQiDB6cdkUhV\nK3STnYYyxlFRmprUfFQ1pk2DSy8NtQONLBLZbMVcvLbJzGykmb1qZrPM7PIC2x1gZuvi0NfU6Xac\nVeSss+DMM5UQRDpIyZJCnGJ7HDAS2A0YZWa75tnu+4SL5CriP1vNRyJSr0pZUxgBvO7ub7r7WmAS\ncFwr210E/JINp9JIjbuSQkWaNg1++9u0oxCpecXMktop3pv5m3F5ezMbUcS+BwOzE8tz4rrkvgcT\nEsX4uMqLirqEmpuhS5fwkAqQmbPoyCNDthaRkiqm6LsVaAEOB74NrIzr9m/jfcUU8D8Gvu7ubmZG\ngeajMWPGZJ83NDTQ0NBQxO7bT/0JFUQji0TapbGxkcbGxs3ah7kXLrvNbKq775v5GddNd/e923jf\nQcAYdx8Zl68AWtz9+4lt3mB9IhgINAFfcvfJOfvytuLsKHPmwIEHwty5ZTmc5HPHHXDlleuvO1BH\nski7mRnu3q5/nmJqCs2xMzhzkK0JNYe2vAAMN7NhwNuE2VVHJTdw9w8k9ns38HBuQig39SdUiI9+\nVLUDkRQUkxRuAX4NvM/Mvgd8jnDjnYLcfZ2ZXQg8DnQG7nT3V8zsvPh6Rd7hRNcoVIidd047ApG6\n1GbzEUAcSvrxuPiku79S0qg2Pn7Zmo+efhq+9jV45pmyHE4gDPlS85BIhytJ85GZbQ+sAh6Oq9zM\ntnf3tzYhxoqn5qMyysxZtHQpjK2bSXhFKloxzUePsn4kUXfCbKmvAbuXKqg0KSmUSXJkke6VLFIx\nirnz2h7J5ThT6gUliyhlmja7xFqb0VRNRyIVo92XaLn7FDM7sBTBVALVFErshhvgH//QyCKRClVM\nn8JlicVOwH5AzY7iV1Iosa99Dbp2Ve1ApEIVU1PolXi+jnCLzl+VJpz0KSmUWLduaUcgIgUUTArx\norU+7n5Zoe1qifoUOkjmXsnDhqUdiYi0Q94J8cysi7u/BxwS5yWqC6opdIDMvZJvuintSESknQrV\nFJ4n9B9MA35rZr8gzE0E4R7ND5U6uDQoKWwG3StZpOoVSgqZ2kF3YDFhltQkJQVZTzOaitSEQklh\nazP7KjCjXMFUAvUpbKI33tC9kkVqQKGk0BnoXa5AKoVqCpvoMxVxe20R2UyFksJ8d/9W2SKpEEoK\nIlLPSnmP5qqkqbPbMG0aPPBA2lGISIkUSgpHlC2KCqLbceaRvFeyiNSsvM1H7r64nIFUCjUftUIj\ni0TqhpqPcigp5LjvvlA7uPRSePhhJQSRGlfUndfSVs47r/XuHWZn6F13467y+M9/oEsXJQORKrQp\nd15TUkhwD+XfmjXhp4hINduUpKDmo4RMMqjbhFAFXxBEpLSUFBLqtj8hM7Lo7LPTjkREUlav34lb\nVZfXKOheySKSoJpCQl1do5C87kAji0QkUk0hoa6aj8aP172SRWQjqikk1FVSuOAC1Q5EZCMlTwpm\nNtLMXjWzWWZ2eSuvn2pm083sRTN72sz2KnVM+dRVn0KXLpriWkQ2UtKkEO/xPA4YCewGjDKzXXM2\newP4mLvvBVwL3FbKmAqpyT6F5maYOTPtKESkSpS6pjACeN3d33T3tcAk4LjkBu7+rLsvj4vPAUNK\nHFNeNdd8lLlX8g9/mHYkIlIlSp0UBgOzE8tz4rp8vgA8WtKICqiZpJAcWfTVr8JPf5p2RCJSJUo9\n+qjoS2TN7DDgHOCQ1l4fM2ZM9nlDQwMNDQ2bGdrGauJWnNOmwZlnwtCh4fl226UdkYiUSWNjI42N\njZu1j5LOfWRmBwFj3H1kXL4CaHH37+dstxfwEDDS3V9vZT9lmfvouutg+XK4/vqSH6p0nngC5s2D\n009XR7JInduUuY9KXVN4ARhuZsOAt4GTgVHJDcxse0JCOK21hFBONdF89IlPpB2BiFSxkiYFd19n\nZhcCjwOdgTvd/RUzOy++PgH4JtAfGG/hm+1adx9RyrjyaWpSa4uI1LeSX9Hs7r8Hfp+zbkLi+ReB\nL5Y6jmJA6SMoAAAPvUlEQVRU1ZDUqVPhuefgy19OOxIRqSG6ojmhKpqPMiOLjjoKevVKOxoRqTGa\n+yih4pPC1KlhRlONLBKRElFNIaGip7mYNCnUDi67LMxZpIQgIiWgmkJCRfcpNDSodiAiJaekkFDR\nzUeDBqUdgYjUATUfJVRMUmhpSTsCEalTSgoJqfcpZEYWnXRSikGISD1TUkhItU9h6lQ44IBwN7Sx\nY1MKQkTqnZJCQirNR8nrDjSySERSpo7myD2lpHDffaF2oJFFIlIBSjpLakcpxyyp774L/fqFn2XV\n0hJmM9WMpiLSwSpxltSqkVp/Qie14IlI5VCJFJW86ai5GWbMKOEBREQ2n5JCVNKkkBlZpHsli0iF\nU1KISnKNQu7Iorvv7uADiIh0LPUpRB3ep/Dii+GWmJrRVESqiJJC1OHNR01NoXageyWLSBVRUog6\nvPnooIPCQ0SkiigpRBU9bbZIiZlqs1Wvo67lUlKINrn5aOpUePJJGD26w2MSKadquJBVWteRSV2j\nj6J2J4XkyKJttilZXCIi5aSaQtSuPgXdK1lEapRqClHRfQq/+Y1mNBWRmqWaQtTUBP37F7HhoYeq\ndiAiNUtJISq6T6F//yKzh4hI9Slp85GZjTSzV81slpldnmebsfH16Wa2bynjKaTVPoV161KJRUQ2\n1tDQwIABA2hubt5o/Z133rnBusbGRoYOHZpddnfGjh3LnnvuSa9evRg6dCgnnXQSL730UofGuGTJ\nEk444QR69erFsGHDmDhxYt5t16xZw6WXXsrgwYMZMGAAF1xwAesSZc64cePYf//96d69O2effXaH\nxllIyZKCmXUGxgEjgd2AUWa2a842xwA7uftw4FxgfKniacsGfQqZkUXHHptWOCKS8Oabb/LUU0/R\nqVMnJk+evMFrZtbmkMxLLrmEsWPHcsstt7B06VJmzpzJ8ccfz+9+97sOjfOCCy6ge/fuLFy4kAce\neIDzzz+fl19+udVtr7/+eqZMmcI///lPZs6cyZQpU/jOd76TfX3w4MFcffXVnHPOOR0aY1tKWVMY\nAbzu7m+6+1pgEnBczjafBu4FcPfngH5mlsr4zmzzUfJeyZrATqQi3HfffXz4wx/mzDPP5N57723X\ne2fNmsWtt97KpEmTaGhooGvXrvTo0YPPf/7zXH55qw0Ym2TVqlU89NBDXHvttfTs2ZNDDjmE4447\njvvvv7/V7R955BEuuugi+vXrx8CBA7n44ou56667sq+fcMIJHHfccWy11VYdFmMxSpkUBgOzE8tz\n4rq2thlSwpjyal7ZzG4/072SRSrRfffdx2mnncapp57K448/zsKFC4t+75NPPsnQoUPZf//9i37P\nV77yFfr379/qY5999mn1PTNnzqRLly7stNNO2XV77703//znP/MeJ3nBYEtLC3PmzGHFihV5tymH\nUiaFYs8kt96XymWVI97+Df3+Fe+VfMYZmsROJCFzx9jNfWyKv/3tb7z11lucdNJJ7Lfffuy44448\n+OCDRb9/8eLFDBo0qF3HvPXWW1m6dGmrj2nTprX6npUrV9KnT58N1vXu3XujQj5j5MiR3HzzzSxa\ntIj58+czduxYzIympqYNtiv3FCSlHH00FxiaWB5KqAkU2mZIXLeRMWPGZJ83NDTQ0NDQETFmXfiX\nE+kx8ETormQgkivNGTDuvfdejjzySAYMGADAqFGjuPfee/mf//kfALp06cLatWs3eM/atWvp2rUr\nAFtttRXz5s0reZy9evXinXfe2WDd8uXL6d27d6vbf+Mb32DZsmXss88+dO/enS9+8YtMmzaNbXJm\nSGhPTaGxsZHGxsZ2x77RAUvxICScfwHDgG7ANGDXnG2OAR6Nzw8C/p5nXy4ipVOp/2NNTU3ep08f\n79Wrlw8aNMgHDRrk/fv3dzPz6dOnu7v7Oeec45dffvkG75swYYIfdthh7u4+c+ZM79y5s7/wwgtF\nH/e8887zXr16tfrYY489Wn3PypUrvVu3bj5r1qzsutNOO82vuOKKoo45YcIEP/jggzdaf9VVV/lZ\nZ51V8L35fn9xffvK7va+oV07h6OB14DXgSviuvOA8xLbjIuvTwf2y7Ofgh+IiGyeSv0fe/DBB33A\ngAE+e/ZsX7BggS9YsMDnz5/vH/vYx/yyyy5zd/fHH3/c3/e+9/nzzz/vLS0t/tprr/muu+7qEyZM\nyO7noosu8uHDh3tjY6OvWbPGV69e7RMnTvTrr7++Q+M95ZRTfNSoUb5q1Sp/6qmnvG/fvv7yyy+3\nuu3cuXN97ty53tLS4s8++6wPHTrUn3jiiezr69at89WrV/vXv/51P/300/3dd9/1devWtbqvqkkK\nHfWo1D9YkVpRqf9jI0eO9NGjR2+0/uc//7lvu+22/t5777m7+1133eW777679+nTx3faaSf//ve/\n7y0tLRu85+abb/bdd9/de/bs6YMHD/ZTTjklb4G9qZYsWeLHH3+8b7nllr7DDjv4xIkTs6/95z//\n8V69evns2bPd3f2vf/2rDxs2zHv27Om77LKLP/jggxvs65prrnEz2+DxrW99q9XjdmRSsPC+ymZm\nXg1xilQrM0P/Y9Ur3+8vrm9XR6kmxBMRkSwlBRERyVJSEBGRLCUFERHJUlIQEZEsJQUREcnSTXZE\nBCj/HDtSmZQURETXKEiWmo8SNnsiqQpSS+cCtXU+tXQuUFvnU0vnsqmUFBJq6Q+ils4Faut8aulc\noLbOp5bOZVMpKYiISJaSgoiIZFXNhHhpxyAiUo3aOyFeVSQFEREpDzUfiYhIlpKCiIhk1V1SMLOR\nZvaqmc0ys8vzbDM2vj7dzPYtd4zt0db5mNmp8TxeNLOnzWyvNOIsRjG/m7jdAWa2zsw+U8742qvI\nv7UGM5tqZi+ZWWOZQyxaEX9nfc3sYTObFs/lrBTCLIqZ3WVmC8xsRoFtqqkMKHg+7S4D2nurtmp+\nAJ0J94MeBnQFpgG75mxzDPBofH4g8Pe0497M8/kw0Dc+H1mp51PMuSS2+xPwCPDZtOPezN9NP+Cf\nwJC4PDDtuDfjXK4ErsucB7AY6JJ27HnO56PAvsCMPK9XTRlQ5Pm0qwyot5rCCOB1d3/T3dcCk4Dj\ncrb5NHAvgLs/B/Qzs23KG2bR2jwfd3/W3ZfHxeeAIWWOsVjF/G4ALgJ+Cfy3nMFtgmLO5/PAr9x9\nDoC7LypzjMUq5lxagD7xeR9gsbuvK2OMRXP3p4ClBTappjKgzfNpbxlQb0lhMDA7sTwnrmtrm0ot\nSIs5n6QvAI+WNKJN1+a5mNlgQmE0Pq6q5KFzxfxuhgMDzOzPZvaCmZ1etujap5hzGQfsZmZvA9OB\nS8oUWylUUxnQXm2WAfU2IV6xhUjuuN5KLXyKjsvMDgPOAQ4pXTibpZhz+THwdXd3C1N6VvK0nsWc\nT1dgP+DjQE/gWTP7u7vPKmlk7VfMuYwEprj7YWa2I/CEme3t7itKHFupVEsZULRiy4B6SwpzgaGJ\n5aGEbwGFthkS11WiYs6H2LF0OzDS3QtVm9NUzLl8CJgUp3geCBxtZmvdfXJ5QmyXYs5nNrDI3VcD\nq83sr8DeQKUlhWLO5SzgOgB3/5eZ/RvYGXihHAF2sGoqA4rSnjKg3pqPXgCGm9kwM+sGnAzkFiiT\ngTMAzOwgYJm7LyhvmEVr83zMbHvgIeA0d389hRiL1ea5uPsH3P397v5+Qr/C+RWaEKC4v7XfAh8x\ns85m1pPQqflymeMsRjHn8hZwBEBsf98ZeKOsUXacaioD2tTeMqCuagruvs7MLgQeJ4youNPdXzGz\n8+LrE9z9UTM7xsxeB1YBZ6cYckHFnA/wTaA/MD5+w17r7iPSijmfIs+lahT5t/aqmT0GvEjoqL3d\n3SsuKRT5u7kWuMfMXiQ0vfyvuy9JLegCzGwicCgw0MxmA9cQmvKqrgyAts+HdpYBmuZCRESy6q35\nSEREClBSEBGRLCUFERHJUlIQEZEsJQUREclSUhARkSwlBakYZvZenEY689i+wLYrO+B495jZG/FY\n/4gXKrV3H7eb2S7x+ZU5rz29uTHG/WQ+lxlmNtnM+rax/d5mdnRHHFvqj65TkIphZivcvXdHb1tg\nH3cDD7v7Q2b2CeBGd997M/a32TG1tV8zuweY6e7fK7D9WcCH3P2ijo5Fap9qClKxzGxLM/tj/Bb/\nopl9upVttjWzvya+SX8krj/SzJ6J7/25mW2Z7zDx51PATvG9X437mmFmlyRi+Z2Fm8jMMLMT4/pG\nM/uQmV0P9Ihx3B9fWxl/TjKzYxIx32NmnzGzTmZ2g5k9b+EmKOcW8bE8S5yh1MxGxHOcYuHmKR+M\n01B8Gzg5xnJijP0uM3subrvR5yiSlfYNIvTQI/MA1gFT4+NXhCkVesfXBgKzEtuuiD8vA66MzzsB\nveK2fwF6xPWXA1e3cry7iTfqAU4kFLj7Eaad6AFsCbwE7AN8Frgt8d4+8eefgf2SMbUS4/HAPfF5\nN8I8QVsA5wLfiOu3AP4PGNZKnJn9dAZ+DhwZl3sDnePzI4BfxudnAmMT7/8ecGp83g94DeiZ9u9b\nj8p81NXcR1LxVrt79taHZtYVuM7MPkqYG2g7M3ufuy9MvOd54K647W/cfbqZNQC7Ac/EuV66Ac+0\ncjwDbjCzq4CFhLnmPwE85GHmUszsIcKdrR4Dbow1gkfc/W/tOK/HgJvjt/ijgb+4+xozOxLY08w+\nF7frQ6itvJnz/h5mNpVQQ3gF+GNc3w+4z8x2IkztnPl/zp1W/EjgU2Y2Oi5vQZgF9LV2nIPUCSUF\nqWSnEr717+fu71mYjrl7cgN3fyomjU8SJmS7iXAXqifc/fNt7N+B0e7+UGaFmR3BhgWqhcP4LAv3\n6j0W+I6ZPenu1xZzEu7+roX7Lx8FnARMTLx8obs/0cYuVrv7vmbWgzAp3QXALYRJ6J509xPMbAeg\nscA+PuOVd58GqUDqU5BK1gdYGBPCYcAOuRvEEUr/dfc7gDsI96r9O3CIhZu9ZPoDhuc5Ru7NVJ4C\njjezHrEf4njgKTPbFnjX3R8AbozHybXWzPJ90foZ4QYnmVoHhAL+K5n3xD6BnnneT6y9XAxcZmad\nCZ/P2/Hl5Eye7xCaljIej+8jHqeib0Qv6VJSkEqSOxTuAWD/OB3z6YSmk9xtDwOmmdkUwrfwmz3c\n6/gsYKKZTSc0He1czDHdfSpwD6FZ6u+E6aynA3sCz8VmnG8C32llX7cBL2Y6mnP2/QfgY4QaTObe\nxXcQ7p8wxcxmEG4z2lpSye7H3acR+jxOAX5AaF6bQuhvyGz3Z8KtMafGDvFrga6xs/4l4Ft5PgsR\nDUkVEZH1VFMQEZEsJQUREclSUhARkSwlBRERyVJSEBGRLCUFERHJUlIQEZEsJQUREcn6/yq+4x2q\n+H0OAAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x7f385e847f50>"
       ]
      }
     ],
     "prompt_number": 85
    }
   ],
   "metadata": {}
  }
 ]
}