documents/Notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5bAr1Da2Wcg3"
   },
   "source": [
    "# Credit Risk Prediction Model with LightGBM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TNugCJqqWcg5"
   },
   "source": [
    "## 1 Introduction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our goal for this project is not only to find a relatively accurate and robust model for risk prediction, but more importantly, to improve its interpretability and give understandable explanations for sales representatives in a bank/credit card company can use to decide on accepting or rejecting applications. \n",
    "\n",
    "The variable names are somewhat cryptic and the user might want more intuitive information for both of the mechanism and the outcome. For that purpose, we manually classified 23 variables into groups and labeled them using a name that makes sense in the business context. Combined with the most important contributing factors, users can easily get insights about  the predictive relationship. To improve the user experience, the full model and explanations can be conditionally displayed in an interactive interface. We deployed our best models on GitHub and Heroku. Users can access our web app by:\n",
    "\n",
    "- opening  http://credit-risk.herokuapp.com/ via web browser, or\n",
    "- running these command lines on terminal\n",
    "\n",
    "```\n",
    "pip install --upgrade streamlit\n",
    "streamlit run https://raw.githubusercontent.com/chriskhanhtran/credit-risk-prediction/master/app.py\n",
    "```\n",
    "\n",
    "A 5-minute explanation and live demonstration can be found at  https://youtu.be/IEk8rmLnJDk/."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "W8hxyNoxWcg5"
   },
   "source": [
    "**Import libraries**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:10.108433Z",
     "start_time": "2019-12-18T04:16:09.561579Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "NlzycM_IWcg6"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import joblib\n",
    "\n",
    "import warnings\n",
    "warnings.simplefilter(action='ignore', category=FutureWarning)\n",
    "\n",
    "%matplotlib inline\n",
    "sns.set_style('darkgrid')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4cp-rJH-Wcg-"
   },
   "source": [
    "**Load data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:10.155305Z",
     "start_time": "2019-12-18T04:16:10.108433Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "TpVP0QK0Wcg_"
   },
   "outputs": [],
   "source": [
    "data = pd.read_csv('../data/heloc_dataset_v1.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:10.249052Z",
     "start_time": "2019-12-18T04:16:10.155305Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 179
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 465,
     "status": "ok",
     "timestamp": 1576269162561,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "EF0YjmMEWchB",
    "outputId": "c96d4142-4f13-42c8-a068-04210023009b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10459, 24)\n"
     ]
    },
    {
     "data": {
      "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>RiskPerformance</th>\n",
       "      <th>ExternalRiskEstimate</th>\n",
       "      <th>MSinceOldestTradeOpen</th>\n",
       "      <th>MSinceMostRecentTradeOpen</th>\n",
       "      <th>AverageMInFile</th>\n",
       "      <th>NumSatisfactoryTrades</th>\n",
       "      <th>NumTrades60Ever2DerogPubRec</th>\n",
       "      <th>NumTrades90Ever2DerogPubRec</th>\n",
       "      <th>PercentTradesNeverDelq</th>\n",
       "      <th>MSinceMostRecentDelq</th>\n",
       "      <th>...</th>\n",
       "      <th>PercentInstallTrades</th>\n",
       "      <th>MSinceMostRecentInqexcl7days</th>\n",
       "      <th>NumInqLast6M</th>\n",
       "      <th>NumInqLast6Mexcl7days</th>\n",
       "      <th>NetFractionRevolvingBurden</th>\n",
       "      <th>NetFractionInstallBurden</th>\n",
       "      <th>NumRevolvingTradesWBalance</th>\n",
       "      <th>NumInstallTradesWBalance</th>\n",
       "      <th>NumBank2NatlTradesWHighUtilization</th>\n",
       "      <th>PercentTradesWBalance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>Bad</td>\n",
       "      <td>55</td>\n",
       "      <td>144</td>\n",
       "      <td>4</td>\n",
       "      <td>84</td>\n",
       "      <td>20</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>83</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>43</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>33</td>\n",
       "      <td>-8</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>Bad</td>\n",
       "      <td>61</td>\n",
       "      <td>58</td>\n",
       "      <td>15</td>\n",
       "      <td>41</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>100</td>\n",
       "      <td>-7</td>\n",
       "      <td>...</td>\n",
       "      <td>67</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-8</td>\n",
       "      <td>0</td>\n",
       "      <td>-8</td>\n",
       "      <td>-8</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>Bad</td>\n",
       "      <td>67</td>\n",
       "      <td>66</td>\n",
       "      <td>5</td>\n",
       "      <td>24</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>100</td>\n",
       "      <td>-7</td>\n",
       "      <td>...</td>\n",
       "      <td>44</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>53</td>\n",
       "      <td>66</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>86</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 24 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  RiskPerformance  ExternalRiskEstimate  MSinceOldestTradeOpen  \\\n",
       "0             Bad                    55                    144   \n",
       "1             Bad                    61                     58   \n",
       "2             Bad                    67                     66   \n",
       "\n",
       "   MSinceMostRecentTradeOpen  AverageMInFile  NumSatisfactoryTrades  \\\n",
       "0                          4              84                     20   \n",
       "1                         15              41                      2   \n",
       "2                          5              24                      9   \n",
       "\n",
       "   NumTrades60Ever2DerogPubRec  NumTrades90Ever2DerogPubRec  \\\n",
       "0                            3                            0   \n",
       "1                            4                            4   \n",
       "2                            0                            0   \n",
       "\n",
       "   PercentTradesNeverDelq  MSinceMostRecentDelq  ...  PercentInstallTrades  \\\n",
       "0                      83                     2  ...                    43   \n",
       "1                     100                    -7  ...                    67   \n",
       "2                     100                    -7  ...                    44   \n",
       "\n",
       "   MSinceMostRecentInqexcl7days  NumInqLast6M  NumInqLast6Mexcl7days  \\\n",
       "0                             0             0                      0   \n",
       "1                             0             0                      0   \n",
       "2                             0             4                      4   \n",
       "\n",
       "   NetFractionRevolvingBurden  NetFractionInstallBurden  \\\n",
       "0                          33                        -8   \n",
       "1                           0                        -8   \n",
       "2                          53                        66   \n",
       "\n",
       "   NumRevolvingTradesWBalance  NumInstallTradesWBalance  \\\n",
       "0                           8                         1   \n",
       "1                           0                        -8   \n",
       "2                           4                         2   \n",
       "\n",
       "   NumBank2NatlTradesWHighUtilization  PercentTradesWBalance  \n",
       "0                                   1                     69  \n",
       "1                                  -8                      0  \n",
       "2                                   1                     86  \n",
       "\n",
       "[3 rows x 24 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(data.shape)\n",
    "data.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:10.264659Z",
     "start_time": "2019-12-18T04:16:10.249052Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 510
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 874,
     "status": "ok",
     "timestamp": 1576269177792,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "jPtDDWCZWchD",
    "outputId": "7370348d-bdc6-4e90-8b21-f25abf7d7d07"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 10459 entries, 0 to 10458\n",
      "Data columns (total 24 columns):\n",
      "RiskPerformance                       10459 non-null object\n",
      "ExternalRiskEstimate                  10459 non-null int64\n",
      "MSinceOldestTradeOpen                 10459 non-null int64\n",
      "MSinceMostRecentTradeOpen             10459 non-null int64\n",
      "AverageMInFile                        10459 non-null int64\n",
      "NumSatisfactoryTrades                 10459 non-null int64\n",
      "NumTrades60Ever2DerogPubRec           10459 non-null int64\n",
      "NumTrades90Ever2DerogPubRec           10459 non-null int64\n",
      "PercentTradesNeverDelq                10459 non-null int64\n",
      "MSinceMostRecentDelq                  10459 non-null int64\n",
      "MaxDelq2PublicRecLast12M              10459 non-null int64\n",
      "MaxDelqEver                           10459 non-null int64\n",
      "NumTotalTrades                        10459 non-null int64\n",
      "NumTradesOpeninLast12M                10459 non-null int64\n",
      "PercentInstallTrades                  10459 non-null int64\n",
      "MSinceMostRecentInqexcl7days          10459 non-null int64\n",
      "NumInqLast6M                          10459 non-null int64\n",
      "NumInqLast6Mexcl7days                 10459 non-null int64\n",
      "NetFractionRevolvingBurden            10459 non-null int64\n",
      "NetFractionInstallBurden              10459 non-null int64\n",
      "NumRevolvingTradesWBalance            10459 non-null int64\n",
      "NumInstallTradesWBalance              10459 non-null int64\n",
      "NumBank2NatlTradesWHighUtilization    10459 non-null int64\n",
      "PercentTradesWBalance                 10459 non-null int64\n",
      "dtypes: int64(23), object(1)\n",
      "memory usage: 1.9+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "MlV_KxnwWchG"
   },
   "source": [
    "## 2 Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:10.998988Z",
     "start_time": "2019-12-18T04:16:10.264659Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "EJQjtKNS-mJI"
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.impute import SimpleImputer\n",
    "\n",
    "# Specify X and y\n",
    "X = data.drop('RiskPerformance', axis=1)\n",
    "y = data.RiskPerformance\n",
    "\n",
    "# Factorizing `y`\n",
    "y = y.map({'Bad': 1, 'Good': 0})\n",
    "\n",
    "# One-hot encode for 'MaxDelq2PublicRecLast12M' and 'MaxDelqEver'\n",
    "dummies = pd.get_dummies(X[['MaxDelq2PublicRecLast12M', 'MaxDelqEver']].astype(str), drop_first=True)\n",
    "X.drop(['MaxDelq2PublicRecLast12M', 'MaxDelqEver'], axis=1, inplace=True)\n",
    "\n",
    "# One-hot encode for -7, -8, -9 values\n",
    "def get_special_dummies(X, col):\n",
    "  \"\"\"\n",
    "  One-hot encode for -7, -8, -9 values in each column\n",
    "  \"\"\"\n",
    "  X[col + '_-7'] = X[col].apply(lambda row:int(row==-7))\n",
    "  X[col + '_-8'] = X[col].apply(lambda row:int(row==-8))\n",
    "  X[col + '_-9'] = X[col].apply(lambda row:int(row==-9))\n",
    "\n",
    "for col in X.columns:\n",
    "    get_special_dummies(X, col)\n",
    "\n",
    "# Concatenate X and dummies\n",
    "X = pd.concat([X, dummies], axis=1)\n",
    "\n",
    "# Replace negative values with NaN\n",
    "X[X < 0] = np.nan\n",
    "\n",
    "# Train test split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X,\n",
    "                                                    y,\n",
    "                                                    test_size=0.1,\n",
    "                                                    random_state=1)\n",
    "# Imputing Missing Values and Feature Scaling\n",
    "preprocessor = Pipeline(steps=[\n",
    "    ('imputer', SimpleImputer(strategy='mean')),\n",
    "    ('scaler', StandardScaler())\n",
    "])\n",
    "\n",
    "# Feature scaling\n",
    "X_train = preprocessor.fit_transform(X_train)\n",
    "X_test = preprocessor.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qmLfiBP-WchK"
   },
   "source": [
    "## 3 Modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:11.123979Z",
     "start_time": "2019-12-18T04:16:10.998988Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "SX8KXml3WchL"
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import KFold, cross_val_score, GridSearchCV\n",
    "from sklearn.metrics import accuracy_score, roc_curve, auc, roc_auc_score\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from lightgbm import LGBMClassifier\n",
    "from sklearn.svm import SVC\n",
    "\n",
    "def gridSearchCV(model, params):\n",
    "    \"\"\"\n",
    "    @param    model: sklearn estimator\n",
    "    @param    params (dict): Dictionary of possible parameters\n",
    "    \n",
    "    @return   cv_results (DataFrame)\n",
    "    \"\"\"\n",
    "    model_cv = GridSearchCV(model, param_grid=params, scoring='roc_auc', cv=5)\n",
    "    model_cv.fit(X_train, y_train)\n",
    "    cv_results = pd.DataFrame(model_cv.cv_results_)[['params', 'mean_test_score']]\n",
    "    \n",
    "    return cv_results\n",
    "\n",
    "def evaluate(model, plotROC=False):\n",
    "    \"\"\"\n",
    "    1. Print AUC and accuracy on the test set\n",
    "    2. Plot ROC\n",
    "    \"\"\"\n",
    "    model.fit(X_train, y_train)\n",
    "    probs = model.predict_proba(X_test)\n",
    "    preds = probs[:,1]\n",
    "    fpr, tpr, threshold = roc_curve(y_test, preds)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    print(f'AUC: {roc_auc:.4f}')\n",
    "    \n",
    "    # Find optimal threshold\n",
    "    rocDf = pd.DataFrame({'fpr': fpr, 'tpr':tpr, 'threshold':threshold})\n",
    "    rocDf['tpr - fpr'] = rocDf.tpr - rocDf.fpr\n",
    "    optimalThreshold = rocDf.threshold[rocDf['tpr - fpr'].idxmax()]\n",
    "    \n",
    "    # Get accuracy over the test set\n",
    "    y_pred = np.where(preds >= optimalThreshold, 1, 0)\n",
    "    accuracy = accuracy_score(y_test, y_pred)\n",
    "    print(f'Accuracy: {accuracy*100:.2f}%')\n",
    "    \n",
    "    # Plot ROC AUC\n",
    "    if plotROC:\n",
    "        plt.title('Receiver Operating Characteristic')\n",
    "        plt.plot(fpr, tpr, 'b', 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])\n",
    "        plt.ylim([0, 1])\n",
    "        plt.ylabel('True Positive Rate')\n",
    "        plt.xlabel('False Positive Rate')\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "But5zX2IWchO"
   },
   "source": [
    "### 3.1 Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:29.230674Z",
     "start_time": "2019-12-18T04:16:11.123979Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 68
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 20511,
     "status": "ok",
     "timestamp": 1576273623755,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "giryrdeXWchP",
    "outputId": "5cbeb28c-aad7-4e9d-f1ee-a02179c97f05"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                  params  mean_test_score\n",
      "0   {'n_estimators': 50}         0.783394\n",
      "1  {'n_estimators': 200}         0.789508\n"
     ]
    }
   ],
   "source": [
    "params = {'n_estimators': [50, 200]}\n",
    "rfc = RandomForestClassifier(random_state=1)\n",
    "print(gridSearchCV(rfc, params))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:38.377185Z",
     "start_time": "2019-12-18T04:16:29.232673Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 329
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 9603,
     "status": "ok",
     "timestamp": 1576273737669,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "7i3rKX2gWchQ",
    "outputId": "e8aec356-c5ca-4503-c4d8-63fe1975864c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.7981\n",
      "Accuracy: 74.00%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rfc = RandomForestClassifier(n_estimators=500, random_state=1)\n",
    "evaluate(rfc, plotROC=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mnrgA3hdWchS"
   },
   "source": [
    "### 3.2 Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:43.965460Z",
     "start_time": "2019-12-18T04:16:38.380186Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 4406,
     "status": "ok",
     "timestamp": 1576273871959,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "MNVgrnAfWchT",
    "outputId": "dd34f715-3a49-42b0-8e4d-3d9ef0c1055e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        params  mean_test_score\n",
      "0  {'C': 0.01}         0.796635\n",
      "1   {'C': 0.1}         0.796564\n",
      "2     {'C': 1}         0.796482\n",
      "3    {'C': 10}         0.796407\n"
     ]
    }
   ],
   "source": [
    "params = {'C': [0.01, 0.1, 1, 10]}\n",
    "logit = LogisticRegression()\n",
    "print(gridSearchCV(logit, params))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:44.402916Z",
     "start_time": "2019-12-18T04:16:43.967457Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 329
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1215,
     "status": "ok",
     "timestamp": 1576272515968,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "DCXmmEDWWchU",
    "outputId": "df0b80ec-4c59-47df-809f-c9329b359084"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.7992\n",
      "Accuracy: 72.94%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "logit = LogisticRegression(C=0.01)\n",
    "evaluate(logit, plotROC=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "FLOiA-3VWchW"
   },
   "source": [
    "### 3.3 LightGBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:48.041285Z",
     "start_time": "2019-12-18T04:16:44.405914Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 4839,
     "status": "ok",
     "timestamp": 1576272501355,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "Ds5jvRqWWchX",
    "outputId": "505d47c3-62d2-4734-db86-d9903fa55500"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                     params  mean_test_score\n",
      "0  {'learning_rate': 0.051}         0.799131\n",
      "1  {'learning_rate': 0.052}         0.798935\n",
      "2  {'learning_rate': 0.055}         0.798296\n"
     ]
    }
   ],
   "source": [
    "params = {'learning_rate': [0.051, 0.052, 0.055]}\n",
    "lightGBM = LGBMClassifier(learning_rate=0.05,\n",
    "                          n_estimators=200,\n",
    "                          num_leaves=5,\n",
    "                          min_data_in_leaf=4,\n",
    "                          max_depth=5,\n",
    "                          max_bin=55,\n",
    "                          bagging_fraction=0.5,\n",
    "                          bagging_freq=5,\n",
    "                          feature_fraction=0.24,\n",
    "                          feature_fraction_seed=9,\n",
    "                          bagging_seed=9,\n",
    "                          min_sum_hessian_in_leaf=11)\n",
    "print(gridSearchCV(lightGBM, params))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:16:48.493642Z",
     "start_time": "2019-12-18T04:16:48.043360Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 329
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1190,
     "status": "ok",
     "timestamp": 1576274017068,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "6yCCIOdLWchZ",
    "outputId": "29224cc6-9bf8-4cb0-f909-e5604041e71c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.8012\n",
      "Accuracy: 73.80%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lightGBM = LGBMClassifier(learning_rate=0.05,\n",
    "                          n_estimators=200,\n",
    "                          num_leaves=5,\n",
    "                          min_data_in_leaf=4,\n",
    "                          max_depth=5,\n",
    "                          max_bin=55,\n",
    "                          bagging_fraction=0.5,\n",
    "                          bagging_freq=5,\n",
    "                          feature_fraction=0.24,\n",
    "                          feature_fraction_seed=9,\n",
    "                          bagging_seed=9,\n",
    "                          min_sum_hessian_in_leaf=11)\n",
    "evaluate(lightGBM, plotROC=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-02T23:38:42.497186Z",
     "start_time": "2019-12-02T23:38:42.493186Z"
    },
    "colab_type": "text",
    "id": "-X0LZQQXWcha"
   },
   "source": [
    "### 3.4 SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:20:57.523573Z",
     "start_time": "2019-12-18T04:20:00.650192Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 211386,
     "status": "ok",
     "timestamp": 1576272737949,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "IC8v_jDVWchb",
    "outputId": "0f30d8bd-0ff0-4bb1-d487-ce95ca5013e5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       params  mean_test_score\n",
      "0  {'C': 0.1}         0.787765\n"
     ]
    }
   ],
   "source": [
    "params = {'C': [0.1]}\n",
    "svc = SVC(gamma='auto', random_state=1)\n",
    "print(gridSearchCV(svc, params))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:22:17.986674Z",
     "start_time": "2019-12-18T04:20:57.527570Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 329
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 58103,
     "status": "ok",
     "timestamp": 1576273991582,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "umzIRgUzWchd",
    "outputId": "ea0744cb-21a8-422b-c995-b30e0d4c502f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.7918\n",
      "Accuracy: 72.56%\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAESCAYAAAD5d3KwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzt3XmcTfX/wPHXXWZfDNGCSvXVJ0UKZSckJSRRtohUEop2aZO0+qbNV4t+WpUWUURJC6EshSwfqZSlzTLMdu/c5fz+OHfMHc1yZ8zd38/Hw8Pce889533PnTnvcz7L+1gMw0AIIYQoYg13AEIIISKLJAYhhBAlSGIQQghRgiQGIYQQJUhiEEIIUYIkBiGEECXYwx2ACB6llAH8CHgAA0gFDgE3aq3XBGF7PwAXaK2zq3vdvvWPBG4EEjA/zzrgHq3178HYXinbHwEkaq2n+2LJ0lo/Wk3rtgE3AwMx/y4TgY+A+7TWTqXULOBHrfWT1bG9SsR1KdBSa31fJd83CdiutX6tnGXuA9ZrrecFsrwIHUkMsa+T1npv0QOl1G3As0Dr6t6Q1vqc6l5nEaXUk0BToIfWeqdSygoMBlYqpVpqrXcFa9t+2mEmWrTWM6p53f8DagJdtNYHlVJpwJvAy8DV1bytyjgPqFXZNwWYSDoDmyuxvAgRSQxxRCllB04C9vs9dw9wBWaz4g5glNZ6j1LqeGAGcAbgBWZorZ9RStUAngaaYJ65fw7crrV2+65Q6gDzgala6/d923gMQGt9p1LqWmCUb3v7gNFa662+M+JawGnAx1rrO/1irA+MBE7UWh/wrcsLvKaUag7cDdyklNoBzAa6Alm+GP7nW0dPYCLmmXg+cJvWeqVS6gHMJFkXWA/cCrwAHAccD/wGXAm0BXoBXZVSBb7PWVtrPdq33VlAF9/+fU1rfa9vu3cB1wI5wNdAb611gyO+lwbAIOAErfUh3+fL812VtPVbtI1SaoUvth+Bgb7lhgM3+D5bLeBRrfX/lFLX+LadBhwEemAmoIbAMb6YBmqtdWnfN/Ctb7/blFIHtdb3BPr9FcWotX5SKfUgcDlQ6HvPNUAfoAXwhFLKA1zmt3xL4Blf3IW+72opImSkjyH2faGU2qCU2gNs8z03DEApNQTzAH++72x/IeYZKsB0YJvW+gzMA+f1Sqn/AE8Ba7XWzYFzgdrA+CO2+ZLfNmyYZ/YvK6U6AkOB9lrrc4HHgbl+70vVWp/lnxR8WgJbipLCEZZgnskXqYV5lnsBMEkp1UQp1RCYAnT3bfd64APfWTnAycC5WuvBQH9gpda6NXAqZhK5Wms9FzPhPaW1fr6UONK11u2BNsBtSqlTlFLdMA+C5wHNgYxS3ofvtU1FSaGI1vrPouTqUw+4EDgdqA/0UUqlA9f5fbarMPdrkbMwm/c6AZcA2Vrr1lrr04HVwGjfcv/6vjEP4jOAd3xJodLfn1LqROAW4DytdQvgU8ymqeeBNZgnFXP9lk8APgQmaa0b+z7b074rRBEicsUQ+zpprfcqpZphHvi/0Fr/7XutB3A+sEYpBWDD7IcA8wB0B4DW+iDQGEAp1QM433fmCJBSyjbfAZ70nYU2wzzg/KSUug74D7DCtz2AmkqpoqaK5eV8joQynk/C7G8o8rzW2gB2KaUWARcBBcAJwOd+2/X6YgFYpbV2+z7r00qp9kqp8Zhn1o0xz5wrMs/3/t1Kqb8xE1R34N2iPhel1POYVxVH8hLYSdqHWut837p+BI7VWuf6vpNLfQnwHCDd7z0b/K5C3lNK/aKUGuP77BcAK33LlfV9+2//Uir//e3GvBJbp5T6BPhEa/15OZ+xCeDRWi/wxbLW95wIIcnCcUJrvQ4YB8zyNV2AmQge01qf47tiaEFx04UbvwOuUupUpVSm7z39/N7TkuKzzqJt5QPvYnakDqP4KsQGvO733ma+bRZdCeSWEf4qoKEv0RypE7DC77Hb72crZse7Dfi8aLu+bbfC11/gv11fs9ck4B/gRcwzXEsZcfkr8PvZ8L3HfcR7PWW891ugkVKqxBWFUqqeUmqBUqoo+bqO3Iavme0HzKue5ZjNZf78P9uNwEzMq6C3MJvdiuIr6/v2V+nvz9fk1xHzymkf8JRS6vEjl/NTIg5fLI19zaAiRCQxxBGt9WzgO8zmIIDFwAi/A8Ak4HXfz0sobg6qgdmX0ND3nnFKKYtSKgmzeaVEYvB5CbPZoS1Q1ByyGBiglDrB93ikb70Vxb0bs815tlKqXtHzSqlhmP0jj/ktPsT32kmYVwuf+LZxkVLqDN9r3YENlH610w2YprV+Hfgbs7/C5nvNTdlXLqVZAFzh239gtvf/q2ql1noPZkfzK0Xfhe//6cA+rXXBke/x0wIziU3GTGI9fO+3lbJsN2CW1nomoIGefp+trO/b/zNX+vtTSjXFTMBbtNaPYP7uned7ubT9qQFDKdXV9/5mwFLkWBVSsrPjz2igu6/9+2XMjsJVSqlNwNmYZ3ZFyzVSSm0AvgEe8V3Wj8XsFNyIeXDdSMk2beBwE4AHeE9r7fA99ynmQfwz33oHAn18TT/l0lrfDbwBzFNK/aiU+gmz+aO11vo3v0VPUUqtBRYBY7VpM2ab+dtKqfXAQ0AvrXVpVyiTMJvBNmAmveUUNzl9AoxUSt1dUby+mJdiJsiVSqk1QA3Ms/XSjMIcobNCmcN+v/U9HlHBZj4FdmEeULdgdn7/4xezvyeBG3yfbRnmcN+i5cr6vpcC3ZRSz1bl+9NarwfmYDZXrgGGU9wnNR94RCk11G95J2bH9P2+/TDDt43CCvaDqEYWKbstYoVvdFBfHYQ5GlWhlGoBtNFaP+N7PB6z4/Wq8EYmRPmk3U6I4NkG3KmUuh6zCel3zCsXISKaXDEIIYQoIWh9DEqplkqpL0t5vqdSarVSaqVv+KIQQogIEpTEoJS6A7NjM/mI5xMwRyVchDmE7foyhiAKIYQIk2D1MfyMObLg9SOeb4RZKOsAgFJqOdAec8x7mQzDkBYvH4sFZF+YZF8Uk31RLFr3hdcLBw9C/hHj1vbtA7fb/FwVMQxzoZP5jRpkk9ii6V7M8i2VEpTEoLV+328Slb9MzJotRXIwh/CVyzBg376y5j7Fl6ysVLKzyxrxGF9kXxSTfVEsmPuioADWrrXhKWuqYhkWLbKza5cVi6XsjLVoUfGUjpSU4uUKfQN1b7yxghG7vmyYmGThlqS5eLL/4fgWTX8r/02lC/WopEOUrBeTAQSlRLMQQgTqr78sLFxox+stf7kZMxL57beqt8A3blx2RmnUyEPDhl7GjSvkrLMqCOQI1j/2kH7HOJyX9cHZ9yrMuZRVF+rEsAWztEEtzOnzHTAn3QghRLXZvx+ys4vbXr791sbGjbbDzTFbt1pZtsw8/FksxuEmmEDNn1/5K5KGDb0cc0w1t3EZBslvvEraAxOxuF0UXtitWlYbksSglBqIWX3yRd8kn8WYHd+v+ModCCFEuf76y8KGDRWfrVssMHBg6YVsa9QwD8wuX9WpIUMKqV3bfK5uXYPu3d2lvs9ferpBcnKFiwWd9ddfyLh1LInLv6awXQdypj6D95RTq2XdQUsMWusdmIXK0Fq/5ff8R5h3phJCiHIZBmzfbmXJEhv331+5o3GnTm769SuuO9i0qZeGDSvXRBPJ7Fs2Y1//AzlTn8ExeGhgvdOBrrva1iSEEJX0zz8W9u41D2h5efD664kkJBQ3t7z2WmKJ5Rs18jBtmqPcdWZkJFNQ4OCss7xYY6wanG3LZuwbfsB51UAKu/dgf6v1GLWOqfbtSGIQQlS7tWut/PFHyaOy1laWLLGTnGwe+D0eWLWq9ENQnTrmmX3NmgbHH+/l9tsL6djRTVoaFR7ss7IgOzt2rgwAKCwkddqTpD49FW+dY3Fe1geSk4OSFEASgxCiFA6HOXa+Ips22Vi5srhTd9q0ROx2OHiw7GaNs8/2kJZmJofmzT20auWhWTNztE5amsEFF3hi7kz/aNjXriZj3GjsW7fg6HsVuQ89SrA7OSQxCCHweMyz/JkzE9m0ycq2baXdziEwGRkGV13lont3NyedVPLM/ZhjDI4/Pgpnn4WJ9Y89ZPW6GG+dYzn45hwKu14cku1KYhAizv38s4XWrdNLPJeUZDBmTOHhM/vynHuul+bNzTN+iwUSEyt4g6iQ7eef8JzWEO8JdTn04ixcHTpiZBx5Q73gkcQgRBzKzoatW2306pV6+Dmr1eC99wpo00aacsLFcjCbtEn3kfzGqxz8cCGu1m0pvLRnyOOQxCBEjCgshA0brGXO3t2928q8eXa++MJOQUFxH8CJJ5qzbQcPdpX+RhESiYsWkn7HOKx//0XBTTfjOqdZ2GKRxCBElMnNhaFDU8jOthzu9LXZrHz/femTuo5Uu7aXtDSD8eMLadLES8uWlSz8I6pd+rjRpLz5Gu5GZ5H92mzcYUwKIIlBiKiwbp2VPn1ScbuhsNDMBgkJBp06mQd1ux3q1HFjtRoMG1b2mX/t2gZNmsTYUM5oVVQC1mLB3fRc8uqfSP6YcRHRSSOJQYgIt2pVcV9Aly5uGjb0kppqMGpUIZm+/kizomhBGKMUlWHdvYv022/B2fsKnFcOwHHN0RW9q26SGISIYAcPcjgpDBxYyFNPOauz8oEINa+X5FdfIe2h+7F4PRR2D33HciAkMQgRYdavN5uNHA5wucws0KGDm2nTnGGOTBwN2y/bSR83hsSV31DYoRM5U5/Ge3KDcIdVKkkMQkSY8eOTycmxMHhwIcccY5CSAtdfX8FNWkTEs2mNffMmDj09HWf/QdVa9K66SWIQIsT8bzu5dauVkSOTyc+3HJ478Ouv5g8PPeQkLS0MAYpqY/txI/YfN+DsP4jCSy5l/+r1GFk1wx1WhSQxCBFCd9+dxMyZ/x510qyZhwYNzNFCzZt7GDmyUJJCNHM6SX3qcVKfeQrvccfj7H2FWfQuCpICSGIQImR27LAwc2Yi9ep5GTiweEhprVoGw4e7IrllQVSCffW3ZtG7bRrHlQPInTQl6EXvqpskBiGCaM8eC3Pn2jEMmDTJPDj06uXm9tulzyAWWf/YQ1bv7niPPY6Ds9+jsMtF4Q6pSiQxCBEkOTlwzjkli9P95z8e7rtPRhfFGts2jed0ZRa9e+lVs+hdemAz0SORJAYhKmHNGit79lhZudLG+vU27Payq486nWbbUMeObmbNKsBqhZSUUEUqQsGSfYC0++8hZfYbZM9fhKtVGwq79wh3WEdNEoMQ5Vi61MYPP9h45plEbDbIySnZEdCuXdl3s7HbDTp3djNlikM6kmNQ4oKPSL9zPNZ9e8m/+dawFr2rbpIYhCiFYUDPnil8913xn0iNGgaDBhVy8cVuTj7Z4NhjvdSqFcYgRdik3zyKlNlv4Gp8Nofeehf32eeEO6RqJYlBiCN8/LGdiROT2LPHnE/w2Wd5nHWWF7v8tcQ3/6J3zc8j99TTKBg1FhISwhtXEMjtOITw+eorGx06pDJ8eAp79lhp0sTDl1/m0bSpJIV4Z935OzX69yFpzmwAHEOGUXDzrTGZFECuGITg0CG4+eZkFiwo/iN///182reX+xTEPa+X5P97mbTJD2AxDJy9Lg93RCEhiUHEDYfDbCZyOi28+moChYVmuZpNm4pvfP/oow6GD5c7mQmwbf+JjHGjSfh2JYUXdCbnyafxnnRyuMMKCUkMIi4YBvTvn8KKFSV/5S++2MWJJ3o54QSDyZOdsdoyIKrAtv0nbHoLh575H86rBkZ00bvqJolBxDzDgKuvLk4KX3yRR1aWwQknGHLTe1GCfeN6bD9uxDlgMIUXd2f/6g0YNbLCHVbISWIQMe2bb2wMGpRCfr55trd5cy61a5c9KU3EKYeDtKmPkfLcNLwn1MV5eV+z6F0cJgWQUUkihi1bZuPhh5PIz7dw4YVuvvoqT5KC+Bf7t6uo2bktqU9PxXHlAA4sXR51Re+qm1wxiJi0ZImNgQPNW2Kec46HN94okGYj8S/WP/aQ1edSvCfUJfudubg6dQl3SBFBEoOIGR98YGfy5CQsFti508wCt9ziZMIEqWQqSrLprXjUGWbRu1feoLBte0hPr/iNcUISg4gJf/1lYeTI4gp1ffq46NXLTffuZdcyEvHHcmA/6fdNIPmdt8ie9wmu1m0p7HZJuMOKOJIYRFTbt8/CW28l8P775q/yI4845KY3olSJH80j465bsRzYT96423Cd2zzcIUUsSQwiqj30UCJvvWXeKrNpUw8dOngkKYh/yRgzkuR33sJ19jnkvP0BniZnhzukiCaJQUQsw4CtW624jpiIvGSJnR07rPz+u5UVK2zUrGnwxRd51K0rI46EH7+id67zWuJuqCgYNQYpfFWxoOwhpZQVmA40BZzACK31dr/XbwMGAF5gitZ6bjDiENHniy9sPPlkEoYBa9bYyl22Xj2D1FSDyZMdkhRECdbfdpBx6804+l6Js/8gHEOGhTukqBKs1NkbSNZat1ZKtQKmApcBKKWygLHAf4A04AdAEkMc+/ZbG199ZeP3363MmVNck6JDBzd5eRZGjy7EZit54G/a1EujRilkZ+eHOlwRyTwerM89S62J92BYrND3ynBHFJWClRjaAYsAtNarlFIt/F7LA37DTAppmFcNIg45nfDmmwncdVfJyUTTpxdw+eVubOVfMAhRgm2bJuOWm7Ct+Q5nl67kPjENb/0Twx1WVApWYsgEDvo99iil7FrrorGDO4HNgA14pKKVWSyQlZVa/VFGIZvNGhP7wjCgXTsrq1ebPcUvv+zl6qvNqwKLJRFIrHAdsbIvqoPsC7D8vQvbL9vxvvoa1v4DyJRRCFUWrMRwCMjwe2z1SwqXACcAp/geL1ZKfaO1/q6slRkG0mTgk5WVGpX7Yvt2C3PmJBzuD9y+3crq1eYlwcKFeTRv7uXgwXJWUIpo3RfBEK/7wr7+e+ybfsQx8Gpo1wXL6g3UOPH4uNwXpalTJ6PihUoRrMTwDdATmOPrY9jo99oBoABwaq0NpVQ2EJ+VquLIHXcks3y5nYQEMzO4XBasVoN58wpo0UJaE0UlFRSQ9uSjpEx/Bm+9+jj69DOL3mVkhjuymBCsxDAX6KqUWgFYgGFKqfHAdq31fKXUhcAqpZQXWA58FqQ4RJg9+mgi06cn4nCYiWD37txwhySiXMLKb0gfNxr7Lz9TMGgIeQ9Mjvuid9UtKIlBa+0FRh7x9Fa/1+8H7g/GtkVkcLlg+PAUFi82f8Xat3czcaIzzFGJaGf9Yw81ruiJt259st+bj6vDBeEOKSbJTA9R7bxeGDs2mcWL7fTp4+LOO52ccorMMxBVZ9u8Cc+ZZ5lF72a9SWHbDpCWFu6wYpYUIhbVyjBg4sQk3n8/gXvucTJjhkOSgqgyy759ZIy6jloXtCZh5TcAFF50iSSFIJMrBlGtnn8+gZdfTmTkyELGjpVy16KKDIOk+XNJv/s2LNnZ5N12F65mLSp+n6gWkhhElRmG2bm8Zo3tcOG6r7+207KlmwcecEoxO1FlGaNvIPndt3Gdcy45732E58yzwh1SXJHEICrNMGD06GTefbe4fMV553kO/z9smEvuliYqz7/oXZt2uM9sTMENo6ToXRjIHheVduONyXzwgZkUevd2MWGCkwYNpB9BVJ11x69k3DoWR9+rcA4YjGPQkHCHFNckMYiAbNtmZetWK//7XyI//WReDmzZkssxx0hCEEfB4yHl5RmkPfIQhtUG/fqHOyKBJAYRgD/+sNCuXclRII8/7pCkII6KTW8l45ZRJKxdg7NrN7PoXd164Q5LIIlBVGDBAjvDhpn3Ur78che33VbIaad5pQ9BHDXb7zuw7fiVQzNm4ry8LzJaIXJIYhClMgy4+eZk3n7b7Evo1MnNjBkO+dsVR8X+/VrsP27EcfU1FHa9mP2rN2CkV63QmwgeSQziMMOAmTMT2LvXPPq//XYCDRp4GT/eSf/+7greLUQ58vNJe+xhUl54Hm/9k3D0628WvZOkEJEkMYjDdu60MGGCWYzMYjGw2w0mTHDSu7ckBVF1Cd8sI2PcaGw7fqVgyHDy7ntQit5FOEkM4jCPORWB554r4MorJRmIo2fds5sa/S7DW/9Esj/4GFe7DuEOSQRAEoPA6zVHHo0aZXYyS8eyOFq2HzfiadwEb916HHptNoVt2kNqfN9hLprIISCOrVtn5Zprkjn++AzOPTedtWvNO6p16SJXC6JqLHv3kjFyOLU6tyVhxXIACi/sJkkhysgVQxxbuNDOwoUJnHmmB7sdhg1z0aqVm5o1wx2ZiDqGQdLc90i/5w4shw6Rd8cEXC3OD3dUoookMcQphwMOHLCQmGjw5Zdyf1xxdDJGXUfy+3NwNW9BzlPP4zmjUbhDEkdBEkMcmj/fzogRZn9CRobMXhZV5PWak9IsFlztOuBueg4F190INlu4IxNHSfoY4ojTCT17phxOCq1bu3nnHblaEJVn/eVnalzRk+TZbwDgGDSEgpGjJSnECLliiGGFhTBpUhJO362Wly2z88sv5rnA22/n07mzJ4zRiajkdpPywnTSHpuMkZiEY+DV4Y5IBEHAiUEpVVNrfSCYwYjq43ZD69Zp7NxpJoLatb14PBbq1fOyaFE+xx0nTUiicmxbNpNx840k/PA9zosvJffx/+I9/oRwhyWCoMLEoJTqCDwP2JRS7wK/aa1nBj0ycVSWL7cdTgq//pojt8gVR822eye2XTs59OL/4bysjxS9i2GB9DE8BHQA/gSmAKOCGpGoFg6H+f/cufmSFESV2deuJvm1/wPM+Qj7vtuAs/cVkhRiXCCJwau13g8YWmsHkBPkmMRRMAxYutTGkCHmhCIZdSSqJC+PtHvvJqv7haQ+N43DHVXp6eGNS4REIIlhu1LqEeAYpdRdwG9Bjkkchb59U+jfv3iW6WmnecMYjYhGCcu+otYFrUl94XkcQ4dz4PNlkJQU7rBECAXS+TwSGAEsB/J8P4sI8uefFt54I4GZMxPYt8/M9Z98kse558oNdUTlWPfspsZVl+M56WSy532Cq3XbcIckwiCQxDBNaz266IFS6jVA7tQdZjk5sHixnc8+szN3bsLh5zMzDRYsyEcpuVIQgbNvXI+7SVO8detx8I13cLVuBykp4Q5LhEmZiUEpdRMwEaillOrje9oCbA5FYKJ8vXqlsmlT8WSi8eOdDBniom5d6VMQgbP8/Tfp99xB8rwPyP5wIa427XB17hrusESYlZkYtNbPA88rpSZoraeEMCZRgV9/tRxOCqtW5XLccYaMPBKVYxgkvfcO6RPvxJKXR97d9+I6r2W4oxIRIpCmpBlKqQFAAuYVQ12t9SPBDUuUZds2aNnSHBny7LMFnHqqXCGIyssYOZzkue/janE+OdOex3O6CndIIoIEkhjeA7YBZwMFgBTXCZPNm6088YTZm3zBBW6uuELumyAqwa/oXeEFXXC3OJ+C4ddLfSPxLwGNWdFajwS2Al0BqdYfJq++msCCBRbq1/fy4osF2KXSlQiQ7eefqHH5pSS/9ToAzgGDpRKqKFNAiUEplQykAQYgM1zCwOOBefPs1KplsG5dHllZ4Y5IRAW3m5Rnp1GzU1vsmzdhJCeHOyIRBQI553weuAX4FNiJOZ9BhNimTVb277eSni59CiIwtk0/knHLTSSs/x5n957kPjYV73HHhzssEQUqTAxa6/eLfvYV0ZPxLyG2bp2VlSvNS/5XX5X5CSIwtj92Y9u9i4MzX6Owx2VS30gErLx5DM2AB4H9wG1a63+AwcA9QL3QhCeys+GSS1IxDPOPuk6dMAckIpr9u2+xb/4RxzXXmkXvVm9AxjKLyirviuEl4G7gZOBhpVQaZkLoUNFKlVJWYDrQFHACI7TW2/1evwS43/dwHXCT1lraSErhdFowDAs33+zkyivdnHdeMtnZ4Y5KRJzcXNLuuYuUl1/A0+AUHAMGm/WNJCmIKiiv8zlPa/2p1vol4BLgV6CT1vrnANbbG0jWWrcG7gKmFr2glMoAngB6aK1bATuA2lWMP27Ur2/QsKE0I4l/S/jic+znNiXl5RdwDL+ObCl6J45SeVcM/oPk92itJ1Zive2ARQBa61VKqRZ+r7UBNgJTlVKnAi/7mqnKZLFAVlZqeYvEpDfftLBpk/lzamoiWVkJ2GzWuNwXpZF9AezciX1QPzjtNDxLv8Deth01wh1TmMnvxdErLzFYlVIJmFcVBb6fLQBa68IK1psJHPR77FFK2bXWbsyrg07AOUAusEwptVJrva2slRkGZGfHz7y6t96yM25c8uF+BZvNoHZtB9nZHrKyUuNqX5QnnveFff33uJueCxnHkPDWe6RdfCHZDi/E6f7wF8+/F0eqUyejSu8rrynpZEADW/x+1pgT3SpyCPCPyOpLCgD7gNVa6z+11rnA15hJQvhMm5ZEUhKccIKXdety2bkzl06dPOEOS0QAy19/kXntEGp27UjCCnPkuOuCziDzE0Q1Kq+I3ilHsd5vgJ7AHKVUK8ymoyJrgcZKqdpANtAKs6NbAB9+aGfHDit9+7qYPt0R7nBEpDAMkt55i/T77sZSUEDuPfdL0TsRNMEqqjAX6KqUWoHZ/DRMKTUe2K61nq+UuhtY7Ft2jtb6xyDFEVVycuD6680a+MOGVdRaJ+JJxvXDSJ73Aa7zW5Hz1HN4Gp4e7pBEDAtKYtBaezHv/OZvq9/rbwNvB2Pb0ezQIbNPoWdPF+edJyOQ4p5/0bsuXXG1ao1j2HXIbflEsAWUGJRSmZj9DL9orfOCG1L8WrDA/Do6d5b+hHhn+2kbGeNG4+g/CMfgoTj7Dwp3SCKOVHjqoZTqC3wFvAWMV0pVZtiqCNDSpTYmTjQ7EE87Ta4W4pbLReq0J6nZqQ22bVsxZIKaCINArknHYXYQ7wUmA5cHNaI4dcMNZt/Cm2/m06qVXDHEI9vGDWR160TalEkUduvO/mWrcV7eN9xhiTgUSGLwaq2dgOErWyFNSdXM44GDB83+BWlGil/Wf/7C+vdfHHzlDQ7NfA3juOPCHZKIU4H0MSxTSs0G6iulZgCrgxxTXMnOhtNPN6d8TJjglPumxBn7qpVm0bubBqwrAAAgAElEQVTh1+Hq3JX9362HVJm1K8KrwisGrfUE4FXMuQYLtNa3Bj2qOLJ5s5kJ7HZDhqjGEUtuDul33UrNXt1IeXE6OJ3mC5IURAQIpPN5DXAq8ILW+qPghxRfcnLM/+fMKaBGvBe5iRMJS5dQs0Mrkv/vZfKvv5EDS6TonYgsgfQxXAqkAp8rpWYppdoGOaa4csst5kik5GSpOh4PrLt3UWPwlRgpKWR/9Cl5kx+DdLlbrogsgTQl/aW1fhK4AkgG5KqhGlksZk2kZs1kiGrMMgzs69YA4K1Xn4Oz3+fA58txny8lLURkCqQpaYhS6nPgNWAhcve2auH1Qvv2qezda6VzZ7dMZo1R1r/+JHPYYGpe3Lm46F3HTlL0TkS0QEYlNQVGaa11sIOJJw8/nIjWZsfzmDHS6RxzDIOkt98k/b4JWJwOcu+dhOv8VuGOSoiAlHfP5x5a64+BbUBHpVTHote01i+GIrhY9uWX5q5ftSqXU0+V/oVYkzliKEkffUhhqzbkPvUsntMahjskIQJW3hXDMb7/jz/ieTmKVZFhwNatVhYssLNxo43Wrd2SFGKJx2N2GlmtOC+6mMJ2HXAMHS5F70TUKe9+DK/6fvRorScXPa+UeiToUcWoH36w0q1bce2b226TJqRYYdumybjlJhwDBuO4+hqcVw0Md0hCVFl5TUnXAiOARkqp7r6nrUAicHcIYos5RWW177/fQfv2Hs4+W0YiRT2Xi9RnnyL1v49jpKVhZGaGOyIhjlp5TUlvAJ8DE4CHfc95gb+DHVSseuWVBACaN/dKUogB9o3ryRhzo1nSoncfch9+AqNOnXCHJcRRK6/xs4nWegfwPqB8/xoBHct5jyjDF1/Y+OQTMzFIWe3YYPnnbyz793Hw1dnkvDhLkoKIGeVdMXQB1gD9j3jeAD4NWkQxauVKc2jq++/nU6eOdDhHq4SV32DbvAnHtdebRe++/QFSUsIdlhDVymIYFR+klFI2zHs3twa+1VqHtNfU6zWMfftyQ7nJauVyQb16ZgXVv//OOap1ZWWlkp2dXx1hRb1Q7gtLziHSHrqflFkzcZ/2Hw58uTKi6hvJ70Ux2RfF6tTJWAu0qOz7KpzgppR6DPgF89aezYA/gWsqu6F4tnOn2el83HHShBSNEpcsJv22W7D++Qf5I0eTd+c9EZUUhKhugQywbqe1fgForbW+GDgxyDHFFJcLPvjA7Ft46CFnmKMRlWXdvYvMIQMwMjPJXvAZeZOmgNxuU8S4QEpi2JRS5wM7lFKJgPSwBei776zcfnsyW7bYuOgiN127usMdkgiEYWBfuxp3i/PNondzPjTLWSQmhjsyIUIikCuG14BngSeBx4GngxpRDNi/H8aPT6JHjzQOHbIwa1YBr79eICeaUcD65x9kDh1Aze4XFhe9a9dBkoKIK4GU3Z4OdMe8J8NkrfXMoEcVpQwD3n7bTtu2acyencCoUYUsW5ZH9+5uLJZwRyfKZRgkv/EqNdudT+KXS8l94GEpeifiViBlt68EVgD3AKuUUoODHlUU0tpK794pjB2bwqmnGixZks8DDzjlHixRInP41WSMH4O7cRP2f7mSglFjwB5IS6sQsSeQ3/xxQHOtda5SKgNYijkrWgD5+fDUU4k8/3wi6enw3/86GDjQJXXTooF/0btLLqXwgs44rr5Git6JuBfIX4BXa50LoLXOARzBDSl6fPaZjQ4d0nj66ST69nWzYkUegwdLUogGti2byerRleQ3XwPAeeUAqYQqhE8gVww/K6WmAl8DHYCfgxtS5Nuzx8I99ySxYEECSnmYNy+f1q094Q5LBKKwkNSnp5I67UmMzEy8WVnhjkiIiBNIYhgO3AB0BbYAdwU1ogjmdsNLLyXw+ONJeL0wcaKTkSMLZcBKlLCv/56MsTdi37IZR59+5E5+DKN27XCHJUTEKa/sdhowDMgF/qe1jutpu6tXW7njjmQ2bTLnJEyZ4uCkk6TmUTSx7N+P5eBBDr7xDoUXXRLucISIWOVdMbwKbAeygNMxy2/HnQMHYPLkJF5/PZG6db3MmlXAJZfI8NNokbD8a+xbNlFw3Y24OnVh/6rvITk53GEJEdHK62mrrbW+CxgFnB+ieCKGYcA775hzEt56K4Ebbyxk+XKZkxAtLIcOkn7rzWT16UHyrJng9JUjkaQgRIXKSwxeAF8TUlwN1fj5Zwt9+qQwZkwKDRqYcxIefFDmJESLxMWfULPd+SS/+Sr5o8Zy4LOvpeidEJVQXlOSVSmVgJkUin62AIS67HYoZWdD69ZmBpg61cGgQTL8NJpYd+8ic/hgPP85nexX38J9bvNwhyRE1CkvMZwMaN/PFt/PFswb9Zwa5LjCwjCgWzezoNGQIYVcfbUrzBGJgBgG9tXf4T6/ZXHRu/NaSn0jIaqozMSgtT6lqitVSlmB6UBTwAmM0FpvL2WZBcA8rfWMqm6rurjdMHp0Mr/+al4ePPqolMiOBtY9u0m/YxxJny4i+8OFuNq0w9W2fbjDEiKqBauRpDeQrLVujTnvYWopy0wGagVp+5U2cWLS4fsmLF2aJ2VyIp3Xi/WlF82id8u+InfSFFwtW4c7KiFiQrASQztgEYDWehVH3FpOKdUXs3P7kyBtv1LWr7fyyitms8OqVbk0bhzXUzaiQuawwdhuGoX73Gbs/2oVBSNHg80W7rCEiAkBnRcrpTIx+xx+0VrnBfCWTOCg32OPUsqutXYrpRoDA4G+wH2BbN9iMe/jGgyGAWPGmPnxlVe8NGsW2Td2t9msQdsXEc/tNmsZWa1YruyHt3cvLNcMI1PGD8f378URZF8cvUDu+dwXs+S2HZijlDK01pMreNshIMPvsVVrXXT7siFAPcwqrQ2AQqXUDq31orJWZhgE7ebeP/5oZevWNFJTDXr0yCM7OyibqTbxeqNz26YfyRh3E45BQ81id5deHrf7ojSyL4rJvihWp05GxQuVItCy260wm4YmA2t8/5fnG6AnZiJpBWwsekFrfUfRz0qpB4A/y0sKwVZQYP4/daoUjY1ITiep054k9empGFlZeI+R2kZCBFugZbedgKG1NoBAmpLmAg6l1ArgKWCcUmq8UqrXUcQaVDVrSt2jSGP/fi01L2xP2tTHcF7el/3LV1PYI2J/hYSIGYFcMSxTSs0G6iulZgCrK3qDb7b0yCOe3lrKcg8EEmQwffKJDD+KVJbsbCx5eRyc/R6FXS4KdzhCxI0Kj4pa6wlKqYuBdcAWrfXHwQ8rNN59185zz5mlEurVkyuGSJCw7Cuz6N31o8yidyvXSTkLIUIskHs+DwGOBf4Cavkex4Tdu82P/9VXeSglQ1TDyXIwm/TxY8i6oifJr/1fcdE7SQpChFwg7SiNfP9bgHOA/cBrQYsoDE47TZJCOCV+soD0O8Zh/edv8kffQt7td0tCECKMAmlKurvoZ6WUBYiZpiQRftZdO8kcMQRPQ0X262/jPqdZuEMSIu4FMo/BvxLZCUCVaygJAYBhkPDtSlyt2uCtfyIH35uPq/l5UvROiAgRyHBVjTmiSGOWsHgiqBGFyP33J/HYY+aBSCbOho51104yB/Ylq9fFJKxYDoCrdVtJCkJEkED6GO7VWr8R9EhCyOOBN99MoE4dgzvucJKQEO6I4oDXS/KsmaQ9dD8WwyBnyuNS9E6ICBVIYrgOiKnEsHixnUOHLDRs6GXwYLnnQihkXjOIpEULKOzYiZypz+A96eRwhySEKEMgiSFJKfU9ZlNS0e0+BwY1qiDL883dfvhhKYMRVH5F75y9++C85FKc/QdJ250QES6QxHBn0KMIsSefNIdCZmXJpLZgsf24kYxbbsIxeCiOa67F2adfuEMSQgSozMSglHpHa32V1vqrUAYUCkV3aatbVxJDtXM4SH3qcVKfnYaRVRPvsceFOyIhRCWVd8VQJ2RRhJjNZjB2bCHJyeGOJLbY160hY8xI7D9tw3HVQHInTcGoGTE36RNCBKi8xHCaUmpKaS9orScEKR4RxSw5OVgcDrLf/gBX5wvDHY4QoorKSwz5mB3OMcfjkc7P6pLwxefY9RYKRo7G1bET+1eslXIWQkS58hLDn1rrV0MWSQjk5UHTpumA3B74aFmyD5B+3wSS334T9xmNKBh2nZkQJCkIEfXKm/m8NmRRhMiIESkcOmReLQwdKvMXqirx4/nUbHc+Se++Tf7Nt3Lg068kIQgRQ8q8YtBa3xbKQIJt924Ln39uftytW3OoJX2iVWLdtZPMG4bhPuNMDs1+D3eTpuEOSQhRzeLm9mVz5ph1LyZMcEpSqCzDIGHlN7jatMNb/0Sy3/8Yd/MWSC0RIWJTIEX0YoLHY/4/dmxheAOJMtadv1Ojfx+yenc/XPTO3aq1JAUhYlhcJIa8PHj8cWkDrxSvl+SZL1CrfUsSvl1FziNP4GrVJtxRCSFCIOabkjwe+PJL82O2auXGGhep8OhlDh1A0uJPKOzUhZwnn8Z74knhDkkIESIxnxgWLrRz7bUpAFx3nYxEKpfLZY7jtVpxXt4XZ4/LcF45QIreCRFnYvr82ePhcFJ4/fV8und3hzmiyGXf8ANZ3TqRPGsmAM4+/XBeNVCSghBxKKYTw0MPmf0KJ5/s5aKLPDKprTQFBaRNfoCsbp2w/v0X3nr1wx2RECLMYropae9e82x3yZI8OfEthX3Nd2bRu5+3UzDwavIemIyRVTPcYQkhwiymE8OePRbq1/dSo0a4I4lMlvx8LC432e/Ow9WxU7jDEUJEiJhNDPPn21m+3E7t2t5whxJREpZ+hn3rVgpGjcHV4QL2r1gDiYnhDksIEUFito/h9tvNmy1MnuwMcySRwbJ/HxmjbyCr/xUkv/MWFPom+klSEEIcISYTw9KlNg4csNC6tZs+feJ8JJJhkPjRh9Rqdz5JH7xL3vjbOfDpl5IQhBBlirmmpH/+sdC/fyoA/frFeVLAV/Ru5LW4z2xMzpwP8TRuEu6QhBARLuYSQ26u+f/IkYUMHhynE9oMg4TlX+Nq3xHviSeRPXch7mbNwR5zX7cQIghirinp44/N4m6NG3vCHEl4WH/bQY1+vcm6omdx0bvzW0pSEEIELKYSw5Yt1sOT2s46K85GI3k8pLw4nVodW2Fft4acx5+SondCiCqJqdPIu+82k8LEic64SwyZQ/qT9NlinBdeRO4T02QGsxCiymImMRw8CCtWmB/n2mvj5J4L/kXv+vU3C99dcaXUNxJCHJWYaUqaNMm8WnjoIQdpaWEOJgTsP6yjZteOJP/fywA4e1+Bs+9VkhSEEEctKFcMSikrMB1oCjiBEVrr7X6vjwP6+x4u1Fo/eLTbzMszD4jXXBPjI5EKCkibdB8p05/BW+dYvCeeGO6IhBAxJlhXDL2BZK11a+AuYGrRC0qpU4FBQBugNXCRUurs6tjoKad4SYrhG7XZV3+LvXkzUp+bhmPg1RxY/h2FF10S7rCEEDEmWH0M7YBFAFrrVUqpFn6v7QQu1lp7AJRSCYCjvJVZLJCVlVruBhMTLdhslgqXi2YWuwGGF/eixdg7dyHeawPabNaY/r4rQ/ZFMdkXRy9YiSETOOj32KOUsmut3VprF7BXKWUBngC+11pvK29lhgHZ2fnlbnDnzhRcLmuFy0WbxCWLsW3dSsHom+HcVmRt+JHsPBfE2Oesiqys1Jj7vqtK9kUx2RfF6tTJqNL7gtWUdAjwj8iqtT5cn0IplQy86Vtm1NFubMMGK998Yz9cFy4WWPbtI+PGEdQY2I/k9+cUF71LSAhvYEKImBesxPAN0B1AKdUK2Fj0gu9KYR6wXmt9Q1GT0tEouiHPuHExkBkMg6S571GrXQuS5s8l77a7OLD4Cyl6J4QImWA1Jc0FuiqlVgAWYJhSajywHbABHYEkpVRRz+ndWuuVld1IXh48+mgSL7xgHjRjoQyGdddOMsaMxH1WY3Keeh7PmWeFOyQhRJwJSmLQWnuBkUc8vdXv5+Tq2M5116WwZIn5EaZMcdC0aZTOdjYMEr7+ElfHTmbRuw8X4j63OXKTaiFEOET1BLcDByyceaaHJUvyGDHCFZXN79Zff6HGFT3J6ndZcdG7FudLUhBChE1UJwaAY481OPvsKLxS8HhI+d9z1LqgNfb1P5Dz5NNS9E4IERFiplZStMkcfCVJn3+G86KLyX38Kbx164U7JCGEAKL4imHVKhtr19rwRtPFQmEhRQE7+w/i0IyZHHr9HUkKQoiIErWJYetWM/QBA6KjNpJ93Rpqdu1A8v+9BIDzsj44+/STondCiIgTtYmhSLt2ET5ENT+ftPsmkNX9QizZ2XgbnBLuiIQQolzSxxBE9lUryRxzA7bfdlAwZDh59z2IkRnvFY6EEJFOEkMQWdwuDJuN7LkLcLVtH+5whBAiIFGXGAoL4dprU1i3LjJbwRIXf4Jtm6ZgzC242nXgwPLVYI+63SyEiGOReXQtx6pVNhYvtvPPP1ZGjCikTh0j3CEBYNm7l4yRw6lx9VUkzX2vuOidJAUhRJSJuqOWyzcIaf78fFq1ioCOZ8Mg6YN3Sb/nDiw5OeTdeQ/5Y8ZJ0TshRNSKusRQJCEhMq4UrLt2knHzKNxNzjaL3p3RKNwhCSHEUYnaxBBWXi8JXy7F1flCs+jd/EW4m54r9Y2EEDEh6voYws32y3Zq9OlBVv8+JKz8BgB3sxaSFIQQMUMSQ6DcblKee5qaF7TB/uNGcqY9L0XvhBAxSZqSAlRjUD8Sv/gc58WXkvv4f/Eef0K4QxJCiKCQxFAep9O8x7LVSsHgoTgGXo2z1+VS30iIKvB43Bw48A9ud3BvwfvXXxYMIzIGp4SK3Z5IzZp1sNmq55AedYnhzjvNm78F+9hsX/MdGeNGUzB0OI4RIyns2Tu4GxQixh048A/JyamkpR2PJYh/wDabFY8nmsouHx3DMMjLO8SBA/9Qu3b1tGREVR+DywW//26GfNZZQfri8/JIu/cusi7tiiU3F8+ppwVnO0LEGbe7kLS0zKAmhXhksVhIS8us1iuxqLpieO45c9LYqFGFJCVV//oTVq0gY/RIbL/voGDYCPImPoCRkVn9GxIiTklSCI7q3q9Rkxi8Xpg+3UwMt97qDM5G3G6MBDvZ8z7B1bptcLYhhBARLmoSw/btVg4etGCzGWRkVN96Exd+jO0nTcHNt5pF75Z9J/WNhIhxb7wxi3ffnc2cOfNJSkri4YcfoEuXi2jlNwS9V69uzJ+/GICvv/6Sd9+djWEYOJ1OBg68mk6dLqz0dufPn8u8eR9gs9kYOvRa2h5RdXn06OsP//z7779xySU9uPbaG5gy5UH27NlNWloa48ffyYknnlTFTx6YqDkCFt3C84UXHNWyPsvff5M+4XaS58/FdfY5FNw4xqxvJElBiKB75x07s2cnVOs6BwxwcdVV7oCW/eyzRXTpchGff/4p3bv3LHfZjRvXM2fOWzz++DRSU1M5eDCbG24YRoMGp3LKKacGHN++fXt57723efnl1yksLGTUqGs577yWJPrVVXvuuRcB2L17F/fddzdDh17LRx/NJSUllRdfnMXvv+/gqace57//fS7g7VZF1BwFXdV1B0/DIOndt0m/9y4seXnkTbiP/JtuNoelCiFi3rp1a6hbtz69e1/BpEn3VZgYPvroQ/r1G0BqaioANWpk8eKLr5JxRNPFo48+xK5dOw8/zsyswZQpTxx+vGXLJpo0aUpiYiKJiYnUq3ciP//8E40anfWvbT7zzFRuvHEMqamp/Prrr4evZE46qQE7dvxa5c8eqKhJDLfeag5TTUw8uvHJ1l07yRg/BnfTc8mZ9jyehqdXR3hCiEq46ip3wGf31e3jj+fRs2dvTjqpAQkJCWza9GOpyxX15+7d+w9169Yr8Vpm5r8Hpdx1173lbjcvL4+0tPTDj1NTU8nNzf3Xctu3/0ReXh4tWpwPQMOGp7NixTI6dLiATZt+ZO/ef/B4PNiCWIYnahJD0XyVTp2qUGrb6yXxiyUUdrnILHr38ae4mzSV+kZCxJlDhw6xcuU3HDiwn/fee4e8vFw++OAdUlJScblKDvf0eMxjzXHHncDff/9FQ7+TyA0bfqBWrWOoX//Ew89VdMWQlpZGfn7+4cf5+fn/uuoA+PTThfTqdfnhx5de2ovffvuVMWNuoEmTpih1RlCTAkRRYrBY4MIL3ZUepmr7+SfSx40hcdUKsj9ciKtNO9znNAtOkEKIiPbppwvp0eMybrrpZgAcDgf9+vViwIDBfPXVF7RvfwEA69d/T4MGZv/BpZf2ZMaM52jWrAUpKSkcOLCfKVMmMXnyYyXWXdEVQ6NGZ/Hii9NxOp24XC5+++1XTjnl3/Ok1qxZzaBBQw8/3rp1M2effQ5jx97K1q2b2bNn19HsgoBETWKoNLeblOnPkvbEFIzkFA498z8ZgipEnPvoo3nce++kw4+Tk5Pp2LEzDoeDlJRUrrlmIKmpqSQkJHDHHRMAaNz4bHr1upxx427CbrfjdDoYOfIm/vOfhpXa9jHH1KZv3/7cdNN1eL1err9+FElJSaxdu5oNG35g2LDrANi/fx81amQdfl/9+ifx0kszmD37DdLTM7j77vITUHWwRENNEa/XMJo391K7tsFbbxUE9J4aV/Ym8culOC/tRc6jUzGOOy64QYZIVlYq2dn5FS8YB2RfFIuGffHnn79x/PEnB3078VYSo0hp+7dOnYy1QIvKriu2rhgcDnN0kc1GwdXDKLh6GIU9Lwt3VEIIEVWiolZSQQH88EP5nS32b1dRs3NbUl4xxwEX9rxMkoIQQlRBVCSGnBzz//btSxnelptL2oTbyerVDYvTibuhCm1wQoiARUPTdTSq7v0aFU1Ju3aZA4oHDiw5yy1hxXIyxozEumsnBddeT96E+yE9vbRVCCHCzG5PJC/vkFRYrWZFZbft9sSKFw5QVCQGgHbt3NSo8e/njZQUsucvxt2yVeiDEkIErGbNOhw48A+5udlB3Y7FEr836qm29VXbmoKsaDRS4sfzsW/fRv4tt+Fq044DX62SiWpCRAGbzV5tN5IpTzSM0Ip0QUkMSikrMB1oCjiBEVrr7X6vXwfcALiByVrrj8tbX0qKQcrBv8gYdRtJH8/Ddc655I8aaxa9k6QghBDVKlidz72BZK11a+AuYGrRC0qp44GxQFugG/CIUqrc+cxZ7n3UanceiZ8tInfiA2QvWGImBSGEENUuWImhHbAIQGu9ipITLM4HvtFaO7XWB4HtwNnlrewE1294zmjEgS9WUDB2vFRCFUKIIApWH0MmcNDvsUcpZddau0t5LQcopVu5mLVF873WVSt+q1X9cUalOnWq8U5FUU72RTHZF8VkXxxWpanmwUoMhwD/b8bqSwqlvZYBVDRMofq624UQQpQrWE1J3wDdAZRSrYCNfq99B7RXSiUrpWoAjYDSC6ILIYQIuaAU0fMblXQ2YAGGYSaK7Vrr+b5RSddjJqYpWuv3qz0IIYQQVRIV1VWFEEKETlTUShJCCBE6khiEEEKUIIlBCCFECRFVK6m6S2lEqwD2wzigv+/hQq31g6GPMjQq2hd+yywA5mmtZ4Q+ytAI4PfiEuB+38N1wE1a65jsRAxgX9wGDAC8mANc5oYl0BBSSrUEHtNaX3DE8z2B+zCPm69orV+qaF2RdsVQraU0olh5++FUYBDQBmgNXKSUKnfmeJQrc1/4mQzEw/zH8n4vMoAngB5a61bADqB2OIIMkfL2RRbmsaI1cBEwLSwRhpBS6g7gZSD5iOcTgKcw90NH4HrfsbRckZYYqrWURhQrbz/sBC7WWnu01l4gAXCEPsSQKW9foJTqi3lW+EnoQwu58vZFG8z5QlOVUsuAv7TW/4Q+xJApb1/kAb8Bab5/8XAD6J+BPqU83whzmsABrXUhsBxoX9HKIi0xlFpKo4zXKiylEcXK3A9aa5fWeq9SyqKUehL4Xmu9LSxRhkaZ+0Ip1RgYiHmZHA/K+/uoDXQC7gQuAW5RSp0e4vhCqbx9AeYJ1GbMJrVnQhlYOPjmgrlKealKx81ISwzVXUojWpW3H1BKJQNv+pYZFeLYQq28fTEEqAcsBa4BxiulLg5teCFV3r7YB6zWWv+ptc4FvgbOCXWAIVTevrgEOAE4BTgJ6K2UOj/E8UWKKh03Iy0xSCkNU5n7QSllAeYB67XWN2itPeEJMWTK3Bda6zu01i19nW2zgP9qrReFI8gQKe/vYy3QWClV23fm3ArzjDlWlbcvDgAFgFNr7cA8EGaFPMLIsAVoqJSqpZRKBDoAKyt6U0SNSgLmAl2VUivwldJQSo2nuJTGM8AyzIR2j+9Lj0Vl7gfAhtmJlOQbhQJwt9a6wi87SpX7OxHe0EKuor+Pu4HFvmXnaK1j9cQJKt4XFwKrlFJezHb1z8IYa8gppQYC6VrrF337ZTHmcfMVrfXuit4vJTGEEEKUEGlNSUIIIcJMEoMQQogSJDEIIYQoQRKDEEKIEiQxCCGEKCHShqsKgVKqAbABc9ZqkaVa60llLD8LeLuqcxiUUjuA3wEP5snSPmCo1jqnEuu4C3Oi3QZgsNb6ZaXUNcD+qg6r9YvLizlMOR24Tmu9ppz3jNZaP1eV7QlRRBKDiFSbj6wSGWQXFc2LUUo9hnk72oBLKWitH/W9twEwAnhZaz2rmuPqBjwA9Chn+YmAJAZxVCQxiKihlLIBLwAnAscAn2it7/V7/XTMGdAuzBLDQ7TWu5VSj2DO+LRizo5+t5xtWDFnyWpfZcpXgNMwz9j/q7V+Ryk1ChiKeSa/XGt9e9FVC3AFcKZS6j7f9v4ETsecqf6qr7LlAq1188rE5XMy5qzeouKBN2FO7gLoi1mSvpZSajpwMzADaOhb/0St9ZcVrF8IQPoYROQ6Uyn1pd+/epgJYZXWuhtmdc0bj3hPV8zSEBcCDwM1fbPDT9Fat8UsMnePryzzkT5VSvQq0FIAAAITSURBVH0BLME8+L6GeaDdq7Vu41vnZKVUbcyriZt9JZ9/OaJ428OYVzv+zV4vYSYSgKuB/6tkXN8ppXZhVhi+zff86cClvqsqDXTTWj+M2XQ1CvOqZa/WugNwGfB8KesWolRyxSAi1b+akpRSmcB5SqlOmMXBjrwfx0zM6qKLMCtKTgCaAM2VUl/6lknAPPM+spDY4SYbv+01wkwUaK1zlFKbMa8ehgG3+ZqcVlJ81l4qrfUWpZRdKXUycBVmkrm+MnEppaZgFoX72/f838CrSqlc4Az+Xf+mCWZtsZa+x3al1DFa633lxSoEyBWDiC7XANla60GYN2ZJ9RUVLHIZsExr3QV4FzNJbAW+8CWZzsAc4JcAt7cFX+16341wmgC/AtcBI7XWHYFzMe+FUMRL6X9XM4HHMRNedhXimgjUBUb5ikg+iHkXvxGYBeOK9kPR/1uB2b71X4K5Pw4E+LlFnJPEIKLJ50B3X+G0/wE/YR4si6wBHvbdqGYk8CzwEZDre24tYFRitNGLwDFKqeXAl8CDWuu/MSt5rlZKLcU8c//W7z1/A4m+qwl/72LeefBl3+NKxeW7KdO1mAkiHbO66DrMopIFfvths1LqDcy+mDOUUl8BK4DffOsQokJSRE8IIUQJcsUghBCiBEkMQgghSpDEIIQQogRJDEIIIUqQxCCEEKIESQxCCCFKkMQghBCihP8H3SoF095oKo0AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "svc = SVC(C=0.1, kernel='linear', degree=2,\n",
    "          gamma=0.1, random_state=1, probability=True)\n",
    "evaluate(svc, plotROC=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "lPca6wDkWchg"
   },
   "source": [
    "### 3.5 Averaging Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:22:18.005664Z",
     "start_time": "2019-12-18T04:22:17.993671Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "2-Yf-_fNWchg"
   },
   "outputs": [],
   "source": [
    "from sklearn.base import BaseEstimator, RegressorMixin, TransformerMixin, clone\n",
    "\n",
    "class AveragingModel(BaseEstimator, RegressorMixin, TransformerMixin):\n",
    "    def __init__(self, models):\n",
    "        self.models = models\n",
    "\n",
    "    def fit(self, X, y):\n",
    "        # Create clone models\n",
    "        self.models_ = [clone(x) for x in self.models]\n",
    "\n",
    "        # Train cloned models\n",
    "        for model in self.models_:\n",
    "            model.fit(X, y)\n",
    "\n",
    "        return self\n",
    "\n",
    "    def predict(self, X):\n",
    "        # Get predictions from trained clone models\n",
    "        predictions = np.column_stack(\n",
    "            [model.predict(X) for model in self.models_])\n",
    "        # Average predictions\n",
    "        predictions = np.mean(predictions, axis=1)\n",
    "        predictions = np.where(predictions >= 0.5, 1, 0)\n",
    "        return predictions\n",
    "    \n",
    "    def predict_proba(self, X):\n",
    "        # Get predictions from trained clone models\n",
    "        probs_1 = np.column_stack(\n",
    "            [model.predict_proba(X)[:, 1] for model in self.models_])\n",
    "        # Average predictions\n",
    "        probs_1 = np.mean(probs_1, axis=1)\n",
    "        probs_0 = 1 - probs_1\n",
    "        probs = np.column_stack([probs_0, probs_1])\n",
    "        return probs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:22:28.510532Z",
     "start_time": "2019-12-18T04:22:18.008662Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "8rGt-offWchk",
    "outputId": "5730cd52-bd19-4505-cbe6-40f3af1c91a5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC: 0.8035\n",
      "Accuracy: 73.52%\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "avg_model = AveragingModel([lightGBM, rfc, logit])\n",
    "evaluate(avg_model, plotROC=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "JjYSyU-PWchn"
   },
   "source": [
    "## 4 Save models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:22:28.524524Z",
     "start_time": "2019-12-18T04:22:28.516539Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "bI-wAYC0Wchn"
   },
   "outputs": [],
   "source": [
    "from sklearn.pipeline import Pipeline\n",
    "import joblib\n",
    "\n",
    "def save_model(model, model_name):\n",
    "    print('Saving ' + model_name + ' ...')\n",
    "    pipeline = Pipeline(steps=[('imputer', SimpleImputer(strategy='mean')),\n",
    "                               ('scaler', StandardScaler()),\n",
    "                               ('model', model)])\n",
    "    pipeline.fit(X, y)\n",
    "    filename = 'models/' + model_name + '.joblib.pkl'\n",
    "    joblib.dump(pipeline, filename, compress=9)\n",
    "    print('Complete!\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:22:28.961474Z",
     "start_time": "2019-12-18T04:22:28.527521Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 221
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 148485,
     "status": "ok",
     "timestamp": 1576274890807,
     "user": {
      "displayName": "Khanh Duc Tran",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mBivPPq6ugLCrYw8TwzyYPUVX3s3MzsM0frR-_MxA=s64",
      "userId": "01880894527573408613"
     },
     "user_tz": 300
    },
    "id": "aodTm9PjWcho",
    "outputId": "9d265b04-92c5-48ae-8c0b-ecc90e7d577f"
   },
   "outputs": [],
   "source": [
    "save_model(logit, 'logit')\n",
    "save_model(rfc, 'rfc')\n",
    "save_model(svc, 'svc')\n",
    "save_model(lightGBM, 'lgbm')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-12-18T04:22:28.964473Z",
     "start_time": "2019-12-18T04:20:00.729Z"
    },
    "colab": {},
    "colab_type": "code",
    "id": "09FTycfoWchq"
   },
   "outputs": [],
   "source": [
    "### Load models\n",
    "#import joblib\n",
    "#model_name = 'lgbm'\n",
    "#filename = 'models/' + model_name + '.joblib.pkl'\n",
    "#model = joblib.load(filename)"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Notebook.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": false,
   "sideBar": false,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": false,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}