SVM for Fraud Detection

svm_fraud_detection.ipynb

{
  "cells": [
    {
      "metadata": {
        "_uuid": "d6d416591c00598f04bfef280548e47217fe9456"
      },
      "cell_type": "markdown",
      "source": "**Fraud Detection with SVM**\n\n"
    },
    {
      "metadata": {
        "_uuid": "287b79323023b6a2759534741357a7bb3ad6eff7"
      },
      "cell_type": "markdown",
      "source": "# Fraud detection principal components using SVM and undersampling for correcting unbalance\n\nThis Kernel is based on work by [Davide Vegliante](https://www.kaggle.com/davidevegliante/nn-for-fraud-detection#).\nInstead of a Neural Network, we will train an SVM, and later use a technique described by [Aneesha Bakharia](https://medium.com/@aneesha/visualising-top-features-in-linear-svm-with-scikit-learn-and-matplotlib-3454ab18a14d) to inspect which of the unnamed features V1..V28 contributed most to the decision boundary / margin of the trained SVM. To increase our confidence that our results are sound, we use [Louis Headley's work](https://www.kaggle.com/louish10/anomaly-detection-for-fraud-detection/notebook) to visualize the probability distributions of various features.\n\ntodo: compare to [Variance Threshold](https://scikit-learn.org/stable/modules/feature_selection.html)\n\n## Dataset\n\nThe datasets contains transactions made by credit cards in September 2013 by european cardholders. This dataset presents transactions that occurred in two days, where we have 492 frauds out of 284,807 transactions. The dataset is highly unbalanced, the positive class (frauds) account for 0.172% of all transactions.\n"
    },
    {
      "metadata": {
        "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
        "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
        "trusted": true,
        "scrolled": true
      },
      "cell_type": "code",
      "source": "import numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\nimport seaborn as sns\nimport matplotlib.pyplot as plt\nimport matplotlib.style\n%matplotlib inline\nimport os\n\nimport warnings  \nwarnings.filterwarnings('ignore')\n\n\nmatplotlib.style.use('ggplot')\nfrom cycler import cycler\ncolor_palette = sns.color_palette()\ncolor_palette[0], color_palette[1] = color_palette[1], color_palette[0]\nmatplotlib.rcParams['axes.prop_cycle'] = cycler(color=color_palette)\n\n# read the dataset and print five rows\noriginal_dataset = pd.read_csv('../input/creditcard.csv')\n\ndataset = original_dataset.copy()\nprint(dataset.head(5))\n",
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": "   Time        V1        V2        V3  ...         V27       V28  Amount  Class\n0   0.0 -1.359807 -0.072781  2.536347  ...    0.133558 -0.021053  149.62      0\n1   0.0  1.191857  0.266151  0.166480  ...   -0.008983  0.014724    2.69      0\n2   1.0 -1.358354 -1.340163  1.773209  ...   -0.055353 -0.059752  378.66      0\n3   1.0 -0.966272 -0.185226  1.792993  ...    0.062723  0.061458  123.50      0\n4   2.0 -1.158233  0.877737  1.548718  ...    0.219422  0.215153   69.99      0\n\n[5 rows x 31 columns]\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "52378368d3e73e4934641a70614520403b21a4ca"
      },
      "cell_type": "markdown",
      "source": "Let's see how many examples and features our dataset contains. "
    },
    {
      "metadata": {
        "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
        "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a",
        "trusted": true
      },
      "cell_type": "code",
      "source": "# count how many entry there are for every class\nclasses_count = pd.value_counts(dataset['Class'])\n\nprint(\"{} Bonafide examples\\n{} Fraud examples\".format(classes_count[0], classes_count[1]))\n\n# classes_count is a Series. \nclasses_count.plot(kind = 'bar')\nplt.xlabel('Classes')\nplt.ylabel('Frequencies')\nplt.title('Fraud Class Hist')",
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": "284315 Bonafide examples\n492 Fraud examples\n",
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "execution_count": 2,
          "data": {
            "text/plain": "Text(0.5,1,'Fraud Class Hist')"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "cc01e42af1b5496883736f07fa289d203ea9e974"
      },
      "cell_type": "markdown",
      "source": "The Features `V1`..`V28` seem to be normalized. The creators of the dataset could not disclose what they represent, but did note that this features are already selected from a larger set using [PCA](https://en.wikipedia.org/wiki/Principal_component_analysis). The (transaction) `Amount` and `Time`  are not normalized, however. SVM algorithms are not scale invariant, so it is highly recommended to scale these 2 remaing features.\n\nThe `Time` feature represents the number of seconds since data recording started at the moment the transaction was performed. The total dataset covers a timespan of two days. It might be interesting to replace this feature by two new features `Day` (having value 0 or 1) and `TimeOfDay` (having values between 0 and 1, where 0 is 00:00  and 1 is 23:59)\n\nTodo: rescale the Time dimension. Removed for now."
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "76f0cbb25c13d9fce563e985a70df6fd632ed224"
      },
      "cell_type": "code",
      "source": "# Since most of our data has already been scaled we should scale the columns that are left to scale (Amount and Time)\nfrom sklearn.preprocessing import RobustScaler\n\n# RobustScaler is less prone to outliers.\nrob_scaler = RobustScaler()\ndataset['Amount'] = rob_scaler.fit_transform(dataset['Amount'].values.reshape(-1,1))\n\n# remove the Time Feature\ndataset.drop(['Time'], axis = 1, inplace = True)\n\ndataset.head(5)",
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 3,
          "data": {
            "text/plain": "         V1        V2        V3  ...         V28    Amount  Class\n0 -1.359807 -0.072781  2.536347  ...   -0.021053  1.783274      0\n1  1.191857  0.266151  0.166480  ...    0.014724 -0.269825      0\n2 -1.358354 -1.340163  1.773209  ...   -0.059752  4.983721      0\n3 -0.966272 -0.185226  1.792993  ...    0.061458  1.418291      0\n4 -1.158233  0.877737  1.548718  ...    0.215153  0.670579      0\n\n[5 rows x 30 columns]",
            "text/html": "<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>V1</th>\n      <th>V2</th>\n      <th>V3</th>\n      <th>V4</th>\n      <th>V5</th>\n      <th>V6</th>\n      <th>V7</th>\n      <th>V8</th>\n      <th>V9</th>\n      <th>V10</th>\n      <th>V11</th>\n      <th>V12</th>\n      <th>V13</th>\n      <th>V14</th>\n      <th>V15</th>\n      <th>V16</th>\n      <th>V17</th>\n      <th>V18</th>\n      <th>V19</th>\n      <th>V20</th>\n      <th>V21</th>\n      <th>V22</th>\n      <th>V23</th>\n      <th>V24</th>\n      <th>V25</th>\n      <th>V26</th>\n      <th>V27</th>\n      <th>V28</th>\n      <th>Amount</th>\n      <th>Class</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>-1.359807</td>\n      <td>-0.072781</td>\n      <td>2.536347</td>\n      <td>1.378155</td>\n      <td>-0.338321</td>\n      <td>0.462388</td>\n      <td>0.239599</td>\n      <td>0.098698</td>\n      <td>0.363787</td>\n      <td>0.090794</td>\n      <td>-0.551600</td>\n      <td>-0.617801</td>\n      <td>-0.991390</td>\n      <td>-0.311169</td>\n      <td>1.468177</td>\n      <td>-0.470401</td>\n      <td>0.207971</td>\n      <td>0.025791</td>\n      <td>0.403993</td>\n      <td>0.251412</td>\n      <td>-0.018307</td>\n      <td>0.277838</td>\n      <td>-0.110474</td>\n      <td>0.066928</td>\n      <td>0.128539</td>\n      <td>-0.189115</td>\n      <td>0.133558</td>\n      <td>-0.021053</td>\n      <td>1.783274</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>1.191857</td>\n      <td>0.266151</td>\n      <td>0.166480</td>\n      <td>0.448154</td>\n      <td>0.060018</td>\n      <td>-0.082361</td>\n      <td>-0.078803</td>\n      <td>0.085102</td>\n      <td>-0.255425</td>\n      <td>-0.166974</td>\n      <td>1.612727</td>\n      <td>1.065235</td>\n      <td>0.489095</td>\n      <td>-0.143772</td>\n      <td>0.635558</td>\n      <td>0.463917</td>\n      <td>-0.114805</td>\n      <td>-0.183361</td>\n      <td>-0.145783</td>\n      <td>-0.069083</td>\n      <td>-0.225775</td>\n      <td>-0.638672</td>\n      <td>0.101288</td>\n      <td>-0.339846</td>\n      <td>0.167170</td>\n      <td>0.125895</td>\n      <td>-0.008983</td>\n      <td>0.014724</td>\n      <td>-0.269825</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>-1.358354</td>\n      <td>-1.340163</td>\n      <td>1.773209</td>\n      <td>0.379780</td>\n      <td>-0.503198</td>\n      <td>1.800499</td>\n      <td>0.791461</td>\n      <td>0.247676</td>\n      <td>-1.514654</td>\n      <td>0.207643</td>\n      <td>0.624501</td>\n      <td>0.066084</td>\n      <td>0.717293</td>\n      <td>-0.165946</td>\n      <td>2.345865</td>\n      <td>-2.890083</td>\n      <td>1.109969</td>\n      <td>-0.121359</td>\n      <td>-2.261857</td>\n      <td>0.524980</td>\n      <td>0.247998</td>\n      <td>0.771679</td>\n      <td>0.909412</td>\n      <td>-0.689281</td>\n      <td>-0.327642</td>\n      <td>-0.139097</td>\n      <td>-0.055353</td>\n      <td>-0.059752</td>\n      <td>4.983721</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>-0.966272</td>\n      <td>-0.185226</td>\n      <td>1.792993</td>\n      <td>-0.863291</td>\n      <td>-0.010309</td>\n      <td>1.247203</td>\n      <td>0.237609</td>\n      <td>0.377436</td>\n      <td>-1.387024</td>\n      <td>-0.054952</td>\n      <td>-0.226487</td>\n      <td>0.178228</td>\n      <td>0.507757</td>\n      <td>-0.287924</td>\n      <td>-0.631418</td>\n      <td>-1.059647</td>\n      <td>-0.684093</td>\n      <td>1.965775</td>\n      <td>-1.232622</td>\n      <td>-0.208038</td>\n      <td>-0.108300</td>\n      <td>0.005274</td>\n      <td>-0.190321</td>\n      <td>-1.175575</td>\n      <td>0.647376</td>\n      <td>-0.221929</td>\n      <td>0.062723</td>\n      <td>0.061458</td>\n      <td>1.418291</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>-1.158233</td>\n      <td>0.877737</td>\n      <td>1.548718</td>\n      <td>0.403034</td>\n      <td>-0.407193</td>\n      <td>0.095921</td>\n      <td>0.592941</td>\n      <td>-0.270533</td>\n      <td>0.817739</td>\n      <td>0.753074</td>\n      <td>-0.822843</td>\n      <td>0.538196</td>\n      <td>1.345852</td>\n      <td>-1.119670</td>\n      <td>0.175121</td>\n      <td>-0.451449</td>\n      <td>-0.237033</td>\n      <td>-0.038195</td>\n      <td>0.803487</td>\n      <td>0.408542</td>\n      <td>-0.009431</td>\n      <td>0.798278</td>\n      <td>-0.137458</td>\n      <td>0.141267</td>\n      <td>-0.206010</td>\n      <td>0.502292</td>\n      <td>0.219422</td>\n      <td>0.215153</td>\n      <td>0.670579</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "4c9da2de181f0f0437433e5c037d1516e7104404"
      },
      "cell_type": "markdown",
      "source": "###  Undersampling with ratio 1"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "a03ba047bd2ce7e44ee5d4847960803393cb5c0b"
      },
      "cell_type": "code",
      "source": "X = dataset.loc[:, dataset.columns != 'Class' ]\ny = dataset.loc[:, dataset.columns == 'Class' ]\n\nfrom imblearn.under_sampling import RandomUnderSampler\nrus = RandomUnderSampler(random_state = 0, sampling_strategy = 1.0)\n\nX_resampled, y_resampled = rus.fit_resample(X, y)\n\nfrom sklearn.model_selection import train_test_split\nX_train, X_test, y_train, y_test = train_test_split(X_resampled, y_resampled, test_size = 0.20, stratify = y_resampled)\n\nassert len(y_train[y_train == 1]) + len(y_test[y_test == 1]) == len(dataset[dataset.Class == 1])\nprint(\"train_set size: {} - Class0: {}, Class1: {}\".format( len(y_train), len(y_train[y_train == 0]), len(y_train[y_train == 1]) ))\nprint(\"test_set size: {} - Class0: {}, Class1: {}\".format( len(y_test), len(y_test[y_test == 0]), len(y_test[y_test == 1]) ))\n",
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": "Using TensorFlow backend.\n",
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": "train_set size: 787 - Class0: 393, Class1: 394\ntest_set size: 197 - Class0: 99, Class1: 98\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "2a12bb26240b3169a8283fe4ae59f3c295e12a55"
      },
      "cell_type": "markdown",
      "source": "### Train SVM Structure\n"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "5442cb2946880c16b0715b66bc17675ff42344df"
      },
      "cell_type": "code",
      "source": "from sklearn.feature_extraction.text import CountVectorizer\nfrom sklearn.svm import LinearSVC\n\nclassifier = LinearSVC(dual=False)\nclassifier.fit(X_train, y_train.ravel())",
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 5,
          "data": {
            "text/plain": "LinearSVC(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n     intercept_scaling=1, loss='squared_hinge', max_iter=1000,\n     multi_class='ovr', penalty='l2', random_state=None, tol=0.0001,\n     verbose=0)"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "e7ab25df2181b069f419ca8856be73754e39d991"
      },
      "cell_type": "markdown",
      "source": "### Inspect the coefficients"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "5ff51c723edea0ad7f6160fc5e79528665d8ab7b",
        "scrolled": true
      },
      "cell_type": "code",
      "source": "def plot_coefficients(classifier, feature_names, top_features=-1):\n if top_features == -1:\n    top_features = len(feature_names)\n    \n coef = classifier.coef_.ravel()\n abs_coef = np.abs(coef)\n top_coefficients = np.argsort(-abs_coef)[-top_features:]\n\n # create plot\n plt.clf()\n plt.figure(figsize=(15, 3))\n colors = [color_palette[c > 0] for c in coef[top_coefficients]]\n plt.bar(np.arange(top_features), coef[top_coefficients], color=colors)\n feature_names = np.array(feature_names)\n plt.xticks(np.arange(0, top_features), feature_names[top_coefficients], rotation=60, ha='right')\n plt.title(\"Feature coefficients\")\n plt.ylabel(\"Coefficient\")\n\nfeature_names = list(X.columns.values)\nplot_coefficients(classifier, feature_names)\nplt.show()",
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 0 Axes>"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 1080x216 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "29bda89500394769ddbe6bffa99da325df8e2320"
      },
      "cell_type": "code",
      "source": "dataset.groupby(\"Class\")['V14', 'V15', 'V19'].describe(percentiles=[])",
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 7,
          "data": {
            "text/plain": "            V14                        ...          V19                    \n          count      mean       std    ...          min       50%       max\nClass                                  ...                                 \n0      284315.0  0.012064  0.897007    ...    -7.213527  0.003117  5.591971\n1         492.0 -6.971723  4.278940    ...    -3.681904  0.646807  5.228342\n\n[2 rows x 18 columns]",
            "text/html": "<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead tr th {\n        text-align: left;\n    }\n\n    .dataframe thead tr:last-of-type th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr>\n      <th></th>\n      <th colspan=\"6\" halign=\"left\">V14</th>\n      <th colspan=\"6\" halign=\"left\">V15</th>\n      <th colspan=\"6\" halign=\"left\">V19</th>\n    </tr>\n    <tr>\n      <th></th>\n      <th>count</th>\n      <th>mean</th>\n      <th>std</th>\n      <th>min</th>\n      <th>50%</th>\n      <th>max</th>\n      <th>count</th>\n      <th>mean</th>\n      <th>std</th>\n      <th>min</th>\n      <th>50%</th>\n      <th>max</th>\n      <th>count</th>\n      <th>mean</th>\n      <th>std</th>\n      <th>min</th>\n      <th>50%</th>\n      <th>max</th>\n    </tr>\n    <tr>\n      <th>Class</th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n      <th></th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>284315.0</td>\n      <td>0.012064</td>\n      <td>0.897007</td>\n      <td>-18.392091</td>\n      <td>0.051947</td>\n      <td>10.526766</td>\n      <td>284315.0</td>\n      <td>0.000161</td>\n      <td>0.915060</td>\n      <td>-4.391307</td>\n      <td>0.048294</td>\n      <td>8.877742</td>\n      <td>284315.0</td>\n      <td>-0.001178</td>\n      <td>0.811733</td>\n      <td>-7.213527</td>\n      <td>0.003117</td>\n      <td>5.591971</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>492.0</td>\n      <td>-6.971723</td>\n      <td>4.278940</td>\n      <td>-19.214325</td>\n      <td>-6.729720</td>\n      <td>3.442422</td>\n      <td>492.0</td>\n      <td>-0.092929</td>\n      <td>1.049915</td>\n      <td>-4.498945</td>\n      <td>-0.057227</td>\n      <td>2.471358</td>\n      <td>492.0</td>\n      <td>0.680659</td>\n      <td>1.539853</td>\n      <td>-3.681904</td>\n      <td>0.646807</td>\n      <td>5.228342</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "aa6c3500c1799457d9b201f7b7903cbca69e3a50"
      },
      "cell_type": "code",
      "source": "def remove_outliers(series):\n    return series[np.abs(series-series.mean()) <= (1*series.std())]\n\nfor feature in ['V14', 'V15', 'V19']:\n    ax = plt.subplot()\n    positive = dataset[feature][dataset.Class == 1]\n    negative = dataset[feature][dataset.Class == 0]\n\n    sns.distplot(positive, bins=50, label='Fraudulent')\n    sns.distplot(negative, bins=50, label='Bonafide')\n    ax.set_xlabel('')\n    ax.set_title('histogram of feature: ' + str(feature))\n    plt.legend(loc='best')\n    plt.show()",
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "b9356fb87f8fa3fd43929c2432499f0319df78ec"
      },
      "cell_type": "markdown",
      "source": "On first inspection, it seems that coefficient strength is indeed a good indicator for the importance of a feature. When the coefficient is positive, the feature is positively correlated with the output class. In other words: Examples with higher `V14` values are more likely to be fraudulent than lower values. Conversely, high values for `V19` are indicative for a bonafide transaction. However, the coefficient for `V19` is much weaker than the coefficient for `V14`, and thus we expect `V19` to have much less of an influence on our prediction. We also expect the probability distributions for Fraudulent and Bonafide transaction to overlap more for `V19` than for `V14`. The coefficient of `V15` is even weaker than that of `V19`. And indeed,, we see that there is larger overlap in the probabilities of fraudulent and bonafide transations for `V15` than for `V19`. All of these predictions were made based on the trained SVM coefficients, and all of them are in line with the probability distributions shown above."
    },
    {
      "metadata": {
        "_uuid": "217fdf976652693f375da858156d2d87c902a4c8"
      },
      "cell_type": "markdown",
      "source": "## Report the classifier performance\n\nNote that **preciosion and recall are unreliable in an unbalanced dataset**, but we used undersampling to account for this."
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "0ca7e49ef2578760b4db7ad3abe3f7a54a3d22db"
      },
      "cell_type": "code",
      "source": "from sklearn.metrics import classification_report\ny_test_pred = classifier.predict(X_test) > 0.5\ntarget_names = [\"Bonafide\", \"Fraudulent\"]\nprint(classification_report(y_test, y_test_pred, target_names=target_names))",
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": "              precision    recall  f1-score   support\n\n    Bonafide       0.88      0.93      0.91        99\n  Fraudulent       0.92      0.88      0.90        98\n\n   micro avg       0.90      0.90      0.90       197\n   macro avg       0.90      0.90      0.90       197\nweighted avg       0.90      0.90      0.90       197\n\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "7f7165f5eabee87292413968be3cac1f746435d8"
      },
      "cell_type": "markdown",
      "source": "### Confusion Matrix"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "ea6a94b895724239172956e405e298224dbd4422"
      },
      "cell_type": "code",
      "source": "\nfrom sklearn.metrics import confusion_matrix\ncm = confusion_matrix(y_test, y_test_pred)\n\nplt.clf()\nplt.grid('off')\nplt.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)\nclassNames = target_names\nplt.title('Fraud or Not Fraud Confusion Matrix - Test Data')\nplt.ylabel('True label')\nplt.xlabel('Predicted label')\ntick_marks = np.arange(len(classNames))\nplt.xticks(tick_marks, classNames, rotation=45)\nplt.yticks(tick_marks, classNames)\ns = [['TN','FP'], ['FN', 'TP']]\nthresh = cm.max() / 2.\nfor i in range(2):\n    for j in range(2):\n        plt.text(j,i, str(s[i][j])+\" = \"+str(cm[i][j]), horizontalalignment=\"center\",\n                 color=\"white\" if cm[i, j] > thresh else \"black\")\nplt.show()\n",
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAE+CAYAAACZVTtkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3Xe8FNX5x/HPoQQREAuKBWtixyhKVMQCltgjGnmsKHaDRjGikag/S+w9ahKN0djLE2tiib1EbIhBMUajRhAVpAgqSGd+f5xzcVlu2XvvXvYy9/t+ve4LdmZ25tnZ3e+ePXN2JmRZhoiI5EerShcgIiLlpWAXEckZBbuISM4o2EVEckbBLiKSMwp2EZGcaVHBHkK4M4Twj0rX0dKEEDYIIWQhhJ6VrqVQCOHKEMKEVNuBZVrnayGE68uxruYuhHBJCOHdStchi2qyYA8h3JreMMV/ZXkDLQlCCC+nx3xs0fQ+aXq3eqzr3BDCRyUs96Ma9nuzfwOGEDqmx/luCOG7EMLkEMKrIYRfhBDal3lbOwC/Ag4DVgEeLtOq9wCGlmldNSr4sMxCCOtVM/+9NG9IPdZZ9drZusS7XADsUOr6Gyp9WFb3mi78W7kM2+me1rVZCcuOLNj2rBDCuBDCkyGEI0II9crV+my3VG3KtaIa/BOwomlTq1swhNA2y7I5TVxP2YUQAtCmltpnAueFEO7OsmzaYixtT+Ctgts17tsQwg+yLJvd9CXVLISwHPH1sjxwNjAcmAb8BBgMfAKU89vWusB3WZaV9RtclmVflXN9JfgUOBo4vWpCCGF7YFXg66bYYAihNZCl1/PieE3vAfwg/b8dMBo4Bni0YJkJi6GOYn8Ezifm6MrAbsC1wEEhhL0q+p7KsqxJ/oBbgWdqmX8n8Y06GBgDzAfaEnfOi8BXxA+BF4CeBfdrA2TAgUXrewH4c8HtFYC/AtOBL4HzqrZZR90bAk8QX7DfAn8D1imYfzQxrHcGRhIDc5ca1vUycDvwGXBewfQ+6TF0K5i2DTHYZgBTUq1dCraZFf2dVcM2f5Tmb13D/A3S/AOAp4Dv0r5pC9wM/C/V8HHV9IL7XgK8W7S+ndP6Vi6Ydkhaz8z0mPZLy/SsrqZ0n5vSPu9WzbxWQOf0/0BsEY8GZgMfAScULT8eOBP4fXoNjQcuBVql+fcW7cuZBdMfLVrX0VXz0+01ia37yWk/fQScXDD/NeD6gtvtgCuBL1K9o4D+BfOXSjUcA9yT9sGnwKl1vE6rnsf/I76+C5+nO9NjHw8MKZh+OPED8xtgIvG1/cOiOgr/3i983oFDgf8Cc4G1C18P6Tl6BhhGbOgAtAZeStNblSlXqurcv4b5R6VaZxJfwxcC7Qrm7wK8QcyFr4ERwLbAstU8/pG11DESuKKa6b2JWVb4mjgaeJOYJxOAh4A107watwtsBPw9PY/TgX8B+5Wynyrdx96buFN/BmwGzAM6ANcDW6f5nwD/SC26+rgV2BTYC9gJWC9tp0YhhKWBp4kvyO2BvsQd/0QIoW3Bom2Bi4gfShsQd3hNviO2QE8NIaxaw3ZXJYbsaGILdR+gB/GDCeAuYjiMJnYbrAJcXdtjKcFlwC3AxsBfiI/5c+BA4ofbEGBQ+rdkIYRewB3pb1PgOuCaOu7TJm331izLPiuen2XZ/CzLqlqfvwLOIn7obJzWfXUI4ZCiu51K/HD5SbrPEOCgNO844NfE52YVYliX6iZiWO9I3E/HAeNqWf4KYABwIrAJ8ABwXwhh26LlziO+BjZNj+mKEELvEur5BzHE9gEIISwP/DzVWewHwDnE19ZuxNfx30IIbbIsmwn0SsvtSdwvhTWuDRxB/NDuTlELOcuy+elxrgucmyafC6wPHJrmN6kQwmBikP+W75+b/sTngBBCR+AR4j77MbAl8cNpNjHk+6RV7Ux8/DvVt4Ysy4YRP9z6F0xuQ3zNbkrct52Ah1OXTW3b7UQM9l3Sfe8D/hpC2KKUQpqyxT6X2AKp+vugqFXxFbB0HetpTWxhHJBu19li5/vWTN+iT/rx1NJiJ74QpgHLF0xbBZgFHFzw6ZsBvUrYBy8DNxBbM28DN6fpfShosQMXE7+1FLa6tkjLbJNunwt8VMI2q1rs3xXt+0OK9s1pJaxrKDCq4HadLXbgfuDZomWGUEuLHVgjzR9UQk0TgfOLpv0ReK/g9njAi5Z5HvhLwe3jgWlFy5TSYv8AOKOW+ha02ImNgjnAkUXLPAE8XvC6zIDLipb5BDinlu1UPY89ia32J9P0wcAbBfthSC3rWCWtY4ui187WRctdQnwvr1LN9OLXw25p2f9L/+5W13Nanz9qaLETc2Iyi+bCz9Jz0Jb44ZQBm9Ww7u61zS9attoWe5p3A/BFLfetqmPjBmz3ReDyupZr6hb768SWeNXfrkXz/51l2XeFE0IIP0yjVz4KIXxD/ETrSP1aVRsRd9SrVROy2CJ5s477bUx8oS7oJ82ybBzwYZpXZX4J61ogi62V04CBIYRNatjuq1lBP32WZSOIX782rmb5UhzGwvv+b0Xz3yi+QwhhUAhheBopMo3YuqvPfoe4718pmvZyHfcJpaw4hLAS0IX49b7Qi8C6Rd+qRhYt8wXQtZTt1OEq4LfpoO7FdbSq1yM2RKqrt/h5bUy9NwN9QwhrEbt0qmutE0LYIoTwSAhhdAjhW+LrGkp7jsem90KtsnjM4vfEbyDXZXUcwwghfBxCmJb+RpRQR03WIR6fuaVgfdOIH9ZtgLWyLPuE2PB4JYTwaAhhSAhhnUZssyaBmD/xRghbp+2NSft9VJpV634PIXQOIVwdQvhPCGFKejy967ofNP3B0xlZltU2kmN6NdMeJ76oBxH7pmcTA7rq4EnVDisOg7YsPnOyeh7ozbLsqRDC08DlxJZOU/usPvs+hDCAGFqnE79KfkPsU/11wWLzaZr9/jnxW8VGZVhXleIDVxl1jwKr8/FlWXZjCOExYsu0L/B0OjB+dGOKpWH1VtX0eYjDeP9M/PZzT/EyIYTOxG7Gp4l97V8S31Nv8/17qzbVvVcXkT5cexG7VX9Uwl124vscaszBxqp9dQSx37zYpwBZlvUPIfQgdm/sBlwUQjg8y7JF9lkjbEzsBiSE0AV4ktilcijxG2cnYsOqrv3+B2Ar4nvyI+K38JtKuF/F+9gXEkLoSmzlXJRl2VNZlr1H/BrVpWqZLMvmEb9yrVpwv/bEr6ZV3iO+QXsVLNOO2L1Rm38D3VM/ZdX9ViH2G5ZjuOBpxBfULtVst1dhizP1o3Uo2O5s4tfNprI98HqWZddmWTYiy7IPiV8ZC00AVi4azrV50TLvEQ8EF6q1rzjLsrnE/sOBoZohoCGEViGEzlmWTQAmpVoL7QD8t74fttWYQMHrKil+fGRZ9lmWZX/OsuwQYgPkyPT6KlZ1oLG6ess9/PRGYkjek1U/+qo7sByxG+nFLMvep+B9lVQFa2NeZxcR9+G2QJ8Qwsm1LZxl2egsyz5Kf582YrsfEwcdrFuwvsK/BR8aWZb9K8uyy7Is2xl4kNjdBmV4/Okb3DZ8f3xsU2AZYtfnP9N+X7HobjVtd3ti9+2DWZa9Q/xwKuXDsnkFO/FN+xVwbAhhvRDCNsQDhzOKlnsGGJS+4mxC7M9f8O0j7bzHgT+GEHYIIWxMPFDYoY7t30EcRXFvCKFHiD+ouZd40PL+xj64LMtGpVpPKZp1HXEUzy0hhI1DCNsBtwEvZFlW1Z30CbBqCGHLEEKXUOZx3cS+481DCHuGOJ55CPHAc6HniOFwduoyOwg4tmiZq4jdAuek57A/cFIJ2z+d+MJ9I4RwVAhhkxDC2iGE/YldOVUf0hcTD0QfEUJYN4RwInEkxEUNedBFngE2CyEckx7fL4B+hQuEEG4IIeyW5ndP8z/OsmxW8cqyLJtK7P+/JISwb9of5xC7JC8uQ72FHicGxi9rmP8JsZF0UghhnRDCT4nfHguNJx6I3TWE0DWEsGx9CkjrHEw8WPoa8UPv0lDG8dk1SY2Dc4DfhBBODSFsmP72DyH8LtW3aQjhtyGEXiGENUIcFrolsTEC8ZvjHGC3Eh//0iGElUMIq4UQeoYQziQemH2G2M8O8QNnHjA47ffdWfS5r2m7HwD9UxZtQhxht0ypO6RJ/ihxuGM103ck9kHNBP4D7EsM1rMKllkVeIw4fOhTYri8wMLDHbsQRyB8R2yJXVDTNou2XzXccTqxe6Da4Y4l7oOXgRuKpq2a1r3g4Gmavk1afpHhjml+O+KHzFeUZ7hjz6Lp7YgfflOIxzVuJ44mmVm03PHp+fiO779eLjh4mpY5jBgks4j97T+vbpvV1NaJOC7432k/fEXshjsOWCotE4DfUPdwxyFF0xZ67qnm4Gmafh6xK/DbtA9OYeGDpzcR+6ZnEL85/h3YoGB+TcMdx1H7cMfig4GLvHZKeR5r2w/EUUEfE99bbxJbhAsNRCC+vscQv2ksNNyxmvUXDndcKW3vgqJlbie+jzs0Jk/q2l9Fj/HN9PxUDWcckuatQ3w/f5Fem2OJjaqlC+4/iJgpc6l7uGOW/manx/4ksSuoVdGyhxPfDzOJXTB90/361bZdYk/BC8T32mfExs/9wMN17aeQViAiIjnR3LpiRESkkRTsIiI5o2AXEckZBbuISM4o2EVEcqapf3nakmm4kUjTK+l0FC2Ngr0Jte9xYqVLkHqYMrxFXPgoN5ZSetVIXTEiIjmjYBcRyRkFu4hIzijYRURyRsEuIpIzCnYRkZxRsIuI5IyCXUQkZxTsIiI5o2AXEckZBbuISM4o2EVEckbBLiKSMwp2EZGcUbCLiOSMgl1EJGcU7CIiOaNgFxHJGQW7iEjOKNhFRHJGwS4ikjMKdhGRnFGwi4jkjIJdRCRnFOwiIjmjYBcRyRkFu4hIzijYRURyRsEuIpIzCnYRkZxRsIuI5IyCXUQkZxTsIiI5o2AXEckZBbuISM4o2EVEckbBLiKSMwp2EZGcUbCLiOSMgl1EJGcU7CIiOaNgFxHJGQW7iEjOKNhFRHJGwS4ikjNtKl2ANJ3lO3fg8Rt/CUDXFZZh/vz5TJwyDYBN1+/G7+54ljOuegiAwQN2osPS7bjwxscbtc39f7o5px+1K61bt+KJl97lrGsfAeCkQ3dk4L69mDt3PpOmTOP48+7k03FTGrWtlqJDu9Z0777Jgtv+wMOMGTOa/vvtw1prrc2s2bPobwdy5tnnNGo7hx58AB9+8AEAU7+eyrKdl+X1ESMbtU6pDAV7jn319XS2PvASAM48bg+mfzeLa+54FoApr13NPjtuyuW3PMXkqdPLsr3lO3fgosH92OaQy5g0ZRo3nT+APluuxwtv/JeR74+l9yH/ZMbMORzTf1suPLkfA874S1m2m3ft27dfJGDHjBlN722348FHHmX69Ols1XMz9thzb3psvnmDt3Pn3fct+P+vTzuVzp07N3hdUlnqimmh5s6bz80PvsIvD9mxbOtce7UV+OjTiUxK3wqee/19+u20GQAvvfkhM2bOAeCNd0azWtdly7bdlq5Dhw702HwLPv74o7KsL8syHrjfsQMOKsv6ZPFTi70Fu/G+lxjuQ7nqtqdrXGb7nuty2ZCfLzJ9xszZ9B141ULTPh47kfXWWok1VlmezydM5Wd9N6Vtm9aL3Hdgv148Oey9xj+AFmLGjBlstUX8gFxz7bXx+x9aaP7kyZN54/XXGPqbsxea/u2337Jzn+2qXeetd9zNhhttVO28YS//k64rdeVH665bhuqlEppVsJvZPGAUEIB5wInu/kqZt7EBcC+QAfsDd7j7NtUsdyvwqLvfX87tNyffTp/JXY++waCD+jAztaaLvfTmhwu6c+oy9dsZnHTRfdx56ZHMzzJee/t/rNOty0LLHLjHT9h8ozXY5ejfNbr+lqK6rhiIAbx1zx60atWKIaedwUYbb7zQ/E6dOjWoj9zvvYf+B6q1viRrVsEOzHD3zQDMbFfgYmCHMm+jH3C/u1+Qbi8S6i3J9Xc9z6v3/JrbH3mt2vn1abEDPP7Suzz+0rsAHLlfb+bNm79gXt+t1ufXR+3KT4++htlz5pbpEbRcVX3sNWlIi33u3Lk88vCDDHt9RNnqlMWvuQV7oWWAKQBmFoDLgN2JLe0L3P0+M+sDnAtMAroDI4BD3T0zs/8D9gbaA68Ax6X7DwbmmdlO7t7XzKa5e8e0jeuAXYCxwOyqQsxsC+AqoGPa1kB3H9fEj3+xmPLNdzzw1FsM7NeL26oJ9/q02AFWXK4jE6dMY9lO7TnWtuPQ028B4iic6888kJ+d+IcFI3OkaTWkxf7cs8+w3vob0K1btyaqShaH5hbs7c1sJLAUsApQdWRvP2AzYFOgCzDczF5K83oAGwNfAMOA3sDLwPXufj6Amd0B7OXufzezG4Bp7n5F0bb3BdYHNgK6Au8Bt5hZW2Lg7+PuE83sAOBC4Mji4s3sWOBYAHdv7L5YbH53x3Mcf0B5vhhdcfr+bLLeagBc/Kd/8NGnEwC46JR+dFi6HXdddhQAY8dPof/gG8uyTSmfv953rw6a5kBzC/bCrphewO1m1h3YFrjH3ecBX5rZi8BPgG+AN9z9s3SfkcBaxGDva2anA0sDywP/Bv5ey7a3L9jGF2b2XJq+PvHbwNNmBtAaqLa17u5/Av6Ubmb1f/hNp3h8+oq9T13w/wlffcsK2/yqLNs5fOit1U7f8/jry7L+lmjS1EW/4Wy/Qx+236FP2bd10y23ln2dsvg1t2BfwN1fNbMuwIp1LDqr4P/zgDZmthTwB6Cnu481s3OJ3wIaIgD/dvdeDby/iMhi1WzHsafRK62BycA/gQPMrLWZrUhsXb9Ry92rQnySmXUkjn6py0sF21gF6JumfwCsmL5BYGZtzWzjmlYiIlJpzS3Y25vZyNSlch9weOoaeQh4B3gbeA443d3H17QSd58K3AS8CzwJDC9h2w8BHxL71m8HXk3rmk38YLjUzN4GRtLCR9KISPMWsqxZdQXnSda+x4mVrkHqYcpwHQdYkiwVO5JDhctolppbi11ERBpJwS4ikjMKdhGRnFGwi4jkjIJdRCRnFOwiIjmjYBcRyRkFu4hIzijYRURyRsEuIpIzCnYRkZxRsIuI5IyCXUQkZxTsIiI5o2AXEckZBbuISM4o2EVEckbBLiKSMwp2EZGcUbCLiORMm5pmmNlhpazA3W8vXzkiItJYNQY7cEwJ988ABbuISDNSY7C7+3aLsxARESmP2lrsCzGz5YDdgFXc/SozWxlo5e5fNFl1IiJSbyUdPDWz7YD/AkcB56XJGwA3NFFdIiLSQKWOivkdcIi77wzMTdNeA7ZskqpERKTBSg32td39qfT/LP07G2hb/pJERKQxSg32981s56JpOwLvlrkeERFppFIPng4BHjGzR4D2ZvZ7YN/0JyIizUhJLXZ3Hwb0AD4mjlsfB/Ry99ebsDYREWmAkGVZ3UsVMLPl3H1KE9WTJ1n7HidWugaphynDr690CVIPS8X+hlDhMpqlkrpizKwzcA1wANDOzGYB9wGnuPvUJqxPRETqqdSDp7cAywJbAculf5dJ00VEpBkp9eDpjsCq7j4j3R6VThL2edOUJSIiDVVqi/0jYI2iad2AD8tbjoiINFapp+19EnjKzG4DxgKrA4cBdzRteSIiUl/1OW3vp0DfgttjgR3KXpGIiDSKTtsrIpIzujSeiEjOlDqOfVXiOPYdgC6F89y9dRPUJSIiDVRqi/2GtOyewDTi6XofAwY1UV0iItJApQZ7b2Cgu78JZO4+AjgCGNxklYmISIOUGuzziOdfB/jazFYEviWOZRcRkWak1GAfDuye/v80cDfwV+CtpihKREQartRgHwAMS/8/GXiF+GvUg5uiKBERabh6n7ZXSqbT9i5hdNreJYtO21uz2k4p8H+lrMDdzy9fOSIi0li1jWNft4T7q7kvItLMqCum6WSTps2tdA1SD6vvc3mlS5B6mPHsUFBXTLV0SgERkZxRsIuI5IyCXUQkZxTsIiI5U+o1TzGzvsCBQFd372dmmwOd3P3FJqtORETqraQWu5kNAm4mXjWp6ipKs4ELm6guERFpoFK7Yk4Fdnb3C4D5adp/gA2bpCoREWmwUoO9EzAm/b9q4Hsbvj/jo4iINBOlBvvLwJCiaScA6l8XEWlmSj14+kvgUTM7BuhkZv8mttb3aLLKRESkQUpqsbv758DmwOHAYcBxQE93H9eEtYmISAOUPNzR3TPiOdmH1bWsiIhUTknBbmafUMOZHN19nbJWJCIijVJqi/3ooturEPvd7ylvOSIi0lglBbu7P1s8zcyeBR4Hril3USIi0nCNOVfMDEDdMCIizUypfezFl8lbGtgTeKrsFYmISKOU2sdefJm86cDvgVvLWo2IiDRancFuZq2BpwF395lNX5KIiDRGnX3s7j4PuE6hLiKyZCj14OljZqbTB4iILAFK7WNvBTxoZi8Tz8m+4MdK7n5kUxQmIiINU2qwfwhc3pSFiIhIedQa7GZ2kLvf4+5nL66CRESkcerqY79xsVQhIiJlU1ewh8VShYiIlE1dfeytzawvtQS8uz9X3pJERKQx6gr2dsDN1BzsGTpfjIhIs1JXsE/X+dZFRJYsjTm7o4iINEM6eCoikjO1Bru7d1pchYiISHmoK0ZEJGcU7CIiOaNgFxHJGQW7iEjOKNhFRHJGwS4ikjMKdhGRnFGwi4jkjIJdRCRnFOwiIjmjYBcRyRkFu4hIzijYRURyRsEuIpIzdV1BSXKga+d2bLhx9wW3b7/nAcZ+OoZ+e+zMnfc9xK577AXAwfvvwwkn/4re2+3QqO3ZvnsyYvjrbLV1b+6+/5EF048/agAj33qLtm3b0mOLnlx57R9p27Zto7aVR8sv057HLz8IgK7Ld2D+/IyJU78DYNMfdeXtj76kTetWfPDpJI6+9FFmzJrb4G0t06Edtwzdm9VXWoY2rVtxjb/OHU+OAmD1lZbhD6fuQbcVO5EB/YY6n375daMfnzQ9BXsLsFT79rzwyoiFpo39dAyrrtaNqy+/eEGwl8uJJ5/KjO++47Zbblpo+s/tYP7459sBOO7IAdx5280ccfTxZd12Hnz1zQy2Pu4WAM48bFumz5jNNX99A4CJj566YN5fhv6MY/buwbX3D2/wto7bZ3PeHzOJ/c+6ny6d2/P2rcdx77P/Zs7c+fz513tx6d2v8NyI0XRYqi3zs6zxD04WC3XFtGAbd/8xnTp35oXnninrerfvsyMdOy16jZZddt2dEAIhBDbfoidffP55Wbfb0gwbNZZ1Vl2+UevIMujYvh0AHdr/gCnfzmTuvPlssOYKtGndiudGjAZg+sw5jfpmIIuXWuwtwMwZM+izzRYArLnm2tx2z/0L5p0yZCiXXHAOfXbcucb7X3/Nldzvdy8yvVfv7bj48mvqXc+cOXPwe+/iwkuvrvd9JWrdKvDTLdfh6eH/W2TeHWftw7qrr7DI9Gvvf4O7n353oWk3PDyC+y/Yn//5L+m09A8Y8NuHyTJYt9vyTJ0+i3vP3Y81V+7M82+N5qw/v8D8+Wq1LwkWS7Cb2TxgVMGkfu4+uszbWAt41N2717HcaKCnu09qwDb6Af919/caVGSFVNcVU2WbbbcD4LVXXq7x/icOPpUTB59atnpOP+VEevXejl69ty3bOluK9j9ow2s3HgnAK6PGcusTby+yzIALHllkWk12+cnavPPRl+x26t2ss+pyPHbZgQwbdTNtWreid/dubH38Xxj75dfceXY/Buy6Cbc98U7ZHos0ncXVYp/h7pvVNNPM2rj7kvA9rx/wKLBEBXtdTjltKFddfjFtWlf/cihni/3yi3/L5EmTuPLaPzao1pZuxuy5C/rYa1KfFvuAXX/Mlfe+CsD/vpjC6PFTWX/1Ffh84re88/EERo+bCsDfhn3IlhutqmBfQlSsK8bMBgL7AR2B1ma2J/AIsBzQFjjL3R8pbomb2RCgo7ufa2ZbAFWv8qeK1t3T3U9Mtx8FrnD3F4pqOBQ4CfgB8DowyN3nmdk04HfAXsAMYB/gh8DPgB3M7Czg5+7+cVl3SoX03WkXLrngHL4cP77a+eVqsd9x6808/8xTPPDoU7RqpcM7TaU+LfaxE76hT4+1GDbqM1ZabmnWW30FPhk3lSnTZtK5Yzu6dG7PpK9n0KfHmrz133FNWLWU0+J6d7U3s5Hp76GC6ZsD+7v7DsBMYF933xzoC1xpZqGO9f4F+KW7b1rfgsxsQ+AAoHf6NjEPOCTN7gC8ltb7EnCMu78C/A04zd03y0uoVzllyFA+/2xsWda110/7cNSAA/nni8/x4/XX4rln4mfuaYNPYOLECey+07b02WYLrrjkgrJsTxrukjuHsfXG3Rh+01E8fvnBnHnT80z+Zgbz52cMvfE5Hr/iYIbfdBQhwC2Pjax0uVKiSnfFPO3uX6X/B+AiM9semA+sBnStaYVmtiywrLu/lCbdAexej5p2ArYAhpsZQHtgQpo3m9jlAjAC2KWUFZrZscCxAO5ej1Ka1pjxUxeZ1nu7HRYar77bnnsz8ds5Zdneo0+9UO308VNnlmX9LcmFty987GPFva4s6/rHTZ7G3r++t9p5z40YzZbH3FzW7cniUelRMdML/n8IsCKwhbvPSQc5lwLmsvA3i6VKWG8p9wnAbe4+tJp5c9y96vD/PErcT+7+J+BP6aaGD4hIRTSnjs7OwIQU6n2BNdP0L4GVzGwFM2tH7PfG3acCU82samjFIQXrGg1sZmatzGx1YMtqtvcssL+ZrQRgZsub2ZrVLFfoW2DRAdoiIs1Icwr2u4CeZjYKOAx4H8Dd5wDnA28AT1dNT44Afm9mI4kt8CrDgE+Io1euBd4q3lgasngW8JSZvZPWvUodNd4LnGZm/zKzH9b7EYqILAYh08+Em0o2adqSMIJTqqy+z+WVLkHqYcazQ2HhBp0kzanFLiIiZaBgFxHJGQW7iEjOKNhFRHJGwS4ikjMKdhGRnFGwi4jkjIJdRCRnFOwiIjmjYBcRyRkFu4hIzijYRURyRsEuIpIzCnYRkZxRsIuI5IyCXUQkZxTsIiI5o2AXEckZBbuISM4o2EUVG1pKAAANlUlEQVREckbBLiKSMwp2EZGcUbCLiOSMgl1EJGcU7CIiOaNgFxHJGQW7iEjOKNhFRHJGwS4ikjMKdhGRnFGwi4jkjIJdRCRnFOwiIjmjYBcRyRkFu4hIzijYRURyRsEuIpIzCnYRkZxRsIuI5IyCXUQkZxTsIiI5o2AXEckZBbuISM4o2EVEckbBLiKSMwp2EZGcUbCLiOSMgl1EJGcU7CIiOROyLKt0DXmlHSvS9EKlC2iO1GJvOiGPf2Y2otI16E/PWcGfVEPBLiKSMwp2EZGcUbBLff2p0gVIvek5a2F08FREJGfUYhcRyRkFu4hIzijYRURyRsEuItUyM+XDEkpPnJSdmemHI0s4M1sWWCv9fyszW7qyFUl9tKl0AZIvZhbcPUv/3wX4Emjl7iMrW5nU04+Bvma2ErA9sHWF65F60HBHaRJm9kvgAOBB4AjgIHd/t7JVSX2Y2V3A3sAv3P2uStcjpVNXjJSFmS1T8P/uwF7AdkAX4GPgPTNrW6HypATVdKH9DrgS2NLMdjOzdmk5PY/NnIJdGs3MNgTONrMt06RpwPvA6cCmxNb6fGAfM+taoTKlFkVdaAPN7HhgaXc/DxgN7A9sYWaDgEE6jtK8KdilHCYAHYA9zKyHu48GNgJOcvc93X2GmQ0ETgDmVa5MqUlBqO8BnAasDww0s5Pd/WpgFHA4cBLwXNXy0jypj10arKrV5u6ZmR0NHAZ8RWypLwWcQwzyUcA+wBHuPqpC5UodzOwYYBfgV+7+WTr4fQDwtrtfl0bGtHf3yRUtVOqkFrs0RqsU6scRW3NXA8sBRxFHXA0EXgfGAoco1JuXarpTZhG7XPqk28OAe4FtzOwUd/9Oob5kUItd6s3M+gJvufvXZtYeuAW4x93/lobHXQJ0As529/crWatUr6hPfTNgrLtPNrN+wF3AHu7+opl1ALYE/uPu4ytYstSDgl3qxcw6A8cADox399lmdg6x6+VKd59kZmsCbxBHVVzt7jMqV7HUxsxOAfYEPgTmuPtJZnYYcBXxoPfTFS1QGkTBLiVLrbmTgN2ANYF/AasTf6E4CHgBeBLYHDgRGOTun1WiVqmbmW0PnOPuO5nZvcTjIYe5+zwzOxI4l3gQdaYOli5ZFOxSkjR2+Y/EA6EjgFeA64j9sVsC2wL7AhsAHYHD3f3fFSlWqmVmvYFd3P3cdHtnoBcwFdgD6Ofus8xsa3d/zcyWcfdvKlexNJQOnkpJ3H0O8A/iD4/uApZ19xOA54lB/6q7Hw8cTeyfVag3P58ThzCenW7/h/h8Hubuu6dQPwH4depb/7ZShUrjqMUuJUtjnK8HngVudvfX0vTfEUfF/FCjJpqfNPqlVepiWQN4FHjA3c8zs8HAusBEYDxwHDBAp39YsqnFLjWqZjjcS8Rxzu8CB6Sgx91PBm4gDnWUZsbdsxTqm7n7p8SDpf3N7DTgZuAeYCVgFeBQhfqSTy12qVbRcLgjgU2AV4GqURK/AJYFhrn7I5WpUkqRzqvelRjgj7j71Wa2OvAEcJe7X5yWCzpImg9qsUu1CkJ9EPHsjC8BvyGOlOgK/B6YDfwk9cdKM1NwoYxW7j6O+Nz1MbNfuvtYYHfgBDMbCt8/57LkU7BLjczsh0B3YgCsCnxHDPOTiF/dLyWOU59esSJlEWa2EYC7zzezPsBvzayDu78AXAH8zMyGpnDfCrivYsVKk1CwywLFferu/jFwFrAhcSjcNsDfiP3shwGzdLC0eUmn1r3CzO5MkwKwPHCKmXV0938CfwJ+Y2YD3f1zd/9fpeqVpqFgF2CRPvV9zexIM1sVmEL8VWnVgdHlgTeB69x9dmWqleqYWSt3nwUcBKxkZpe4+/PAbcTus1PTopOJPyR7pjKVSlPTwVNZSPo5+a+AL4jD3+5x96fN7DFiqC8H7K+RE82XmR0IbEM8o+Zf3X2ImW0NHE/8AVkH4nP4QQXLlCakYJcFzGw3Yqtuz3QOmDOANYjnhfkn8VQBn7v7FxUsU2phZgcAFxF/BbwWcVz6x+5+Upr/U+ADdx9TsSKlyakrpgUr7FNPfbNrEy9avHeafBXwKXFUzPbuPlyh3uy1Bv7g7u8AjwODgR3M7EYAd39KoZ5/CvYWqqhPvTNxSNwfiVfPGWhmP0196FcDbxN/fi7NlJn1NLPVgEnAyWa2hrvPdfcPib8/WM/MVq5slbK4qCumhTOzU4HtiV0uFwP/BX4M/Bz4k7s/VsHypARm1g04g3hQ9BLi+XpOBI4F1iN+AzvK3SdWrEhZrBTsLVjqb70A2Jk4Vr0vMBy4n3jul97EbpgZ+vFK82ZmuwI7ES8kfi3Qn3jmzc7Ame7+duWqk8VNXTEtiJmtYGaF53NZiXgg7Rt3vw94gHi90i7Ec4gcly6HplBvhsxsPzO7AsDdnwSeAlYgnhv/bnc/BNhXod7yKNhbiHTCrieAG83sojT5PSAzs60A0tVyhgFd3X26u0+tTLVSnWpOyvYh8RQB5wG4+zPErrQDgUHpHPpzF2+V0hyoK6YFSMMYzwIuB8YQhzQeSXzTnwe0J/bPjgfOBvqkn5tLM1F0sHtlAHcfb2YbEn9J+pK7n2lm+xL71M9w9wmVq1gqScGec2a2PHGkxM/d/SEz2xJ4BHiYeCGFS4h96zsQf7hylS6S0XyZ2RBiX/oKwE3ufpOZrUc8x/rHxHOr7+3uGsXUginYWwAz25N4kHQg8SRQrxD70B8ARrn7kWm5dukn6dJMFLXUjwMOdvcdzOx24o+QznX3K82sPfGEXh/pOrOiYG8hUnfM48Bv3P2SNK0jsfV+oLtP1Pm4m5eiUF8R+CHxVA/7Ea8xexXxOb3K3c+vWKHS7OjgaQvh7v8AdgWOMLNl0+T+xP71mWkZhXozUhDqxwN3Au8A04nDU89y91eAvwN7pB+ZiQAK9hYljXoZDLxccAGNY91dFy1uRgp/IZrOp34AcIi7fwd8TexLP8DMTgbaAP3d/etK1CrNk7piWiAz2wt4EOihA6XNSzoecg6wF/GiJr8ATib+puCRtMzPgR7ElvvROtOmFFOwt1BmtnRqAUozkY6DnAlcmLrOMLOliVesWof4o6MXCpbvoKtXSXUU7CLNQMGw1P3c/WEz+xHxNwWDgFWIY9PXBR5KXWoiNVKwizQTqRvmt8RhqVcDj7n7VWne2sQrI61IHNk0o1J1SvOng6cizUQ6k+ZvgJHA0+5+lZm1SfM+Ae4CfqtQl7qoxS7SzJjZLsB1wFbu/rWZtXX3OZWuS5YcCnaRZsjMdgeuAXq5+1eVrkeWLAp2kWbKzPYhDn3sCWT6AZmUSsEu0oyZWUd3n1bpOmTJomAXEckZjYoREckZBbuISM4o2EVEckbBLkscM1vLzLKqH++Y2RNmdvhi2O65ZnZnDfP6mFlJF7gws4Fm9nIDa2jwfaXlaFPpAiSfzGw00BWYRzyH+BPAiU0xwsPdd69HTUeniz6L5JZa7NKU9nb3jsDmxLHYZxUvYGbBzPQ6FCkjtdilybn752b2BNAdwMxeAIYBfYihv4mZTSRe6m0PYD7wF+Acd59nZq2BS4knx/oGuLJw/Wl9d7r7n9PtY4BfAd2AscChwCnAGsDfzWwecL67X2ZmW6ftbgSMAU6uOjVuOvHWranG14APSn3MZnYGcAywUqrhTHd/qGCRYGbXAwOAccAJ7v5sum/nmvZFqduXlk0tJWlyZrY6MaT+VTB5AHAs0IkYqLcCc4EfES8i8VPg6LTsMcQLT/Qgtvz3r2Vb/YFzgcOAZYCfAZPdfQDwKelbRAr11YDHiBf6Xh4YAjyQri8KcDcwAuhCPOtiffrxPwa2AzoD5wF3mtkqBfO3Sst0If669MF06l7q2BcidVKLXZrSw2Y2l3g5t8eAiwrm3Vp19SYz60oM/mXTmQunm9nVxOC/ETDgGncfm5a/mNjar87RwGXuPjzd/qiW+g4FHnf3x9Ptp83sTeI1RJ8HfgLs7O6zgJfM7O+lPnB3/2vBzfvMbCiwJfHi4QAT0mPK0vxTgT3N7Clq3xcidVKwS1PqV8uByrEF/18TaAuMM7Oqaa0Kllm1aPkxtWxzdWJLuBRrAv3NbO+CaW2B59M2pxRdoWhMWn+dzOwwYnfQWmlSR2LrvMrnRed+GZO2Wde+EKmTgl0qpTDUxgKzgC7uPreaZcexcKCuUct6xwI/LGGbVcve4e7HFC9oZmsCyxVdfm6NataxiHTfm4CdgFfTcYKRQChYbDUzCwXhvgbwN+reFyJ1UrBLxbn7uNQFcaWZnQ1MA9YGurn7i4ADJ5nZo8Shk2fUsro/A1elsd5vEUN+jruPAb4kXju0yp3AcDPbFXiG2FLeGvjI3cekbpnzzOw3xG6UvYnhW5cOxA+AiQBmdgTpwHGBldJj+gPQD9iQ2C00uY59IVInHTyV5uIw4AfAe8AU4H7itT4htn6fBN4mhvWDNa0k9W1fSDzw+S3wMPHAKMDFwFlmNtXMhqQ++32IVy2aSGwtn8b374uDiQc5vyIe4Ly9lAfi7u8RR+68Svww2YQ4CqjQ68RrmE5K9e7v7pNL2BciddLZHUVEckYtdhGRnFGwi4jkjIJdRCRnFOwiIjmjYBcRyRkFu4hIzijYRURyRsEuIpIzCnYRkZz5f/hr/Y+90YSZAAAAAElFTkSuQmCC\n"
          },
          "metadata": {}
        }
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3.6.6",
      "mimetype": "text/x-python",
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "pygments_lexer": "ipython3",
      "nbconvert_exporter": "python",
      "file_extension": ".py"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 1
}