Report 3 - MODELS.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-02-26T12:24:53.530726Z",
     "start_time": "2020-02-26T12:24:53.523729Z"
    }
   },
   "source": [
    "# 3. Predictive models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section we will develop different kind of models exploiting the train-test approach."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:37.309345Z",
     "start_time": "2020-03-04T12:34:21.300461Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\tensorboard\\compat\\tensorflow_stub\\dtypes.py:541: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n",
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\tensorboard\\compat\\tensorflow_stub\\dtypes.py:542: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n",
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\tensorboard\\compat\\tensorflow_stub\\dtypes.py:543: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n",
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\tensorboard\\compat\\tensorflow_stub\\dtypes.py:544: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n",
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\tensorboard\\compat\\tensorflow_stub\\dtypes.py:545: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n",
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\tensorboard\\compat\\tensorflow_stub\\dtypes.py:550: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd \n",
    "import numpy as np \n",
    "import matplotlib.pyplot as plt \n",
    "import seaborn as sns\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "import category_encoders as ce\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix, precision_score, recall_score, f1_score, roc_curve, roc_auc_score\n",
    "\n",
    "import scikitplot as skplt\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.linear_model import LogisticRegressionCV\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import roc_curve\n",
    "from sklearn.metrics import auc\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn import preprocessing\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from mlxtend.evaluate import lift_score\n",
    "\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import classification_report,accuracy_score, precision_recall_curve\n",
    "from sklearn.model_selection import GridSearchCV, RandomizedSearchCV\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "\n",
    "from tensorflow.python.keras.layers import Dense\n",
    "from tensorflow.python.keras import Sequential\n",
    "from tensorflow.python.keras.layers import Dropout\n",
    "from tensorflow.python.keras.initializers import he_normal, glorot_normal\n",
    "from tensorflow.keras.optimizers import Adam, SGD\n",
    "from tensorflow.keras.regularizers import l1,l2\n",
    "\n",
    "#from keras.utils import np_utils\n",
    "#from keras.models import Sequential\n",
    "#from keras.layers import Dense\n",
    "#from keras.layers import Dropout\n",
    "#from keras import optimizers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:37.395047Z",
     "start_time": "2020-03-04T12:34:37.321606Z"
    }
   },
   "outputs": [],
   "source": [
    "data=pd.read_csv('Dataset_end.csv',sep=',',)\n",
    "data.set_index('ID',inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Reparametrization:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Arrangement of the variables to prepare the dataset for the modelling process:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:37.497057Z",
     "start_time": "2020-03-04T12:34:37.401399Z"
    }
   },
   "outputs": [],
   "source": [
    "cat=['job','marital','month','day_of_week','poutcome','pdays_new','education']\n",
    "\n",
    "for i in cat: \n",
    "    data[i]=data[i].astype('category')\n",
    "    \n",
    "for i in data[cat]:\n",
    "    data[i]=data[i].cat.codes\n",
    "    \n",
    "for i in cat: \n",
    "    data[i]=data[i].astype('category')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:37.544155Z",
     "start_time": "2020-03-04T12:34:37.520516Z"
    }
   },
   "outputs": [],
   "source": [
    "data['y']=pd.factorize(data['y'])[0]\n",
    "data['contact']=pd.factorize(data['contact'])[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction goal"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The goal of our analysis is to built a predictive model, able to classify well all the y's but with a focus on the maximization of the precision of 1 ('yes'). \n",
    "\n",
    "Precision is a valid choice of evaluation metric when we want to assess our prediction performances.\n",
    "This choice was made to avoid future useless expenditure in marketing for people that would not buy the product, above all because the current campaign seems not to be quite effective (see EDA analysis).\n",
    "\n",
    "The metrics that we used to evaluate all the models below are:\n",
    "- **AUC**: it evaluates the overall performance of the model;\n",
    "- **Precision** for the positive class ('yes'): it evaluates the correct prediction of 1 (e.g. given a precision value of 80% means that, when the prediction of a client successfully buy a banking product, it is correct 80% of the time);\n",
    "- **Recall** for the positive class ('yes'): it evaluates how many 1's the classifier identifies among all the 1's present in the test set (e.g. given a value of 80%, it correctly identifies 80% of all *yes* response present);\n",
    "- **F1 score**: it is the harmonic average of precision and recall;\n",
    "- **Accuracy**: it sums up the ability of the model to catch the right category of the response for each observation;\n",
    "- **Lift score**: it compares model predictions to randomly generated predictions;\n",
    "\n",
    "All of these metrics, except the AUC, are dependent on the choice of the threshold. Thus, the definition of this parameter will be made according to the goal define above.\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.1 Parametric approach"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### One hot encoding:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:37.650850Z",
     "start_time": "2020-03-04T12:34:37.552101Z"
    }
   },
   "outputs": [],
   "source": [
    "data_dummy=pd.get_dummies(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Avoid correlations:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the first report (1.2.6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We decided to avoid: emp.var.rate, because it has correlations above 0.7 with the other three socio-economic variables and nr.employed, instead of euribor3m, because it is not strickly related to the banking context."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:37.920559Z",
     "start_time": "2020-03-04T12:34:37.899315Z"
    }
   },
   "outputs": [],
   "source": [
    "data_glm=data_dummy.drop(['emp.var.rate','nr.employed'], axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1.1 Logistic regression"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We split in train and test:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:37.960848Z",
     "start_time": "2020-03-04T12:34:37.932961Z"
    }
   },
   "outputs": [],
   "source": [
    "Y=data_glm['y']\n",
    "X=data_glm.drop('y',axis=1)\n",
    "x_train_glm,x_test_glm,y_train_glm,y_test_glm=train_test_split(X,Y, test_size=0.2, random_state=123)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:38.578763Z",
     "start_time": "2020-03-04T12:34:37.980310Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "glm=LogisticRegression(random_state=10091995)\n",
    "glm.fit(x_train_glm,y_train_glm)\n",
    "proba_train= [y for (x, y) in glm.predict_proba(x_train_glm)]\n",
    "glm_proba= [y for (x, y) in glm.predict_proba(x_test_glm)]\n",
    "lr_auc = roc_auc_score(y_test_glm, glm_proba)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:39.454345Z",
     "start_time": "2020-03-04T12:34:38.592697Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set_style('whitegrid')\n",
    "fpr, tpr, threshold = roc_curve(y_test_glm,glm_proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "fig=plt.subplots(figsize=(10,5))\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": {},
   "source": [
    "The overall quality of our logistic regression is indicated by the AUC score(0.79).\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:39.469348Z",
     "start_time": "2020-03-04T12:34:39.460989Z"
    }
   },
   "outputs": [],
   "source": [
    "glm_pred=glm.predict(x_test_glm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:39.493074Z",
     "start_time": "2020-03-04T12:34:39.480526Z"
    }
   },
   "outputs": [],
   "source": [
    "def report (y_t,y_pred,y_proba,c):\n",
    "    auc=roc_auc_score(y_t,y_proba)\n",
    "    p=precision_score(y_t,y_pred)\n",
    "    r=recall_score(y_t,y_pred)\n",
    "    f1 =(2 * (p* r)) / (p+ r)\n",
    "    a=accuracy_score(y_t,y_pred)\n",
    "    score=np.array([auc,p,r,f1,a]).round(2)\n",
    "    table=pd.DataFrame(score,index=['AUC','Prec','Rec','F1','Acc'],columns=[c])\n",
    "    return(table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:39.543995Z",
     "start_time": "2020-03-04T12:34:39.500040Z"
    }
   },
   "outputs": [
    {
     "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>GLM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUC</th>\n",
       "      <td>0.79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prec</th>\n",
       "      <td>0.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rec</th>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1</th>\n",
       "      <td>0.72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Acc</th>\n",
       "      <td>0.73</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       GLM\n",
       "AUC   0.79\n",
       "Prec  0.78\n",
       "Rec   0.67\n",
       "F1    0.72\n",
       "Acc   0.73"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "GLM=report(y_test_glm,glm_pred,glm_proba,'GLM')\n",
    "GLM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With a threshold of 0.5:\n",
    "\n",
    "- Our model has a precision of 0.78;\n",
    "- Our model has a recall of 0.67, in other words;\n",
    "- Our model has a F1 (the harmonic average of precision an recall) of 0.72;\n",
    "- Our model accuracy is 73%;\n",
    " \n",
    "\n",
    "The choice of 0.5 can be a good compromise, given the fact that the dataset is perfectly balanced among the response. <br>\n",
    "\n",
    "Despite that we can decide not to use this threshold, because our aim is to maximise the precision of 1 ('yes'). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:40.646589Z",
     "start_time": "2020-03-04T12:34:39.555137Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "prec, rec, tre = precision_recall_curve(y_test_glm,glm_proba)\n",
    "F1 =(2 * (prec* rec)) / (prec+ rec)\n",
    "def plot_prec_recall_vs_tresh(precisions, recalls, thresholds, f1,x=0.5):\n",
    "    plt.plot(thresholds, precisions[:-1], 'b--', label='precision')\n",
    "    plt.plot(thresholds, recalls[:-1], 'g--', label = 'recall')\n",
    "    plt.plot(thresholds, f1[:-1], 'r--', label = 'F1')\n",
    "    plt.axvline(x=0.5,c='k')\n",
    "    plt.axvline(x=x,c='c')\n",
    "    plt.xlabel('Threshold')\n",
    "    plt.legend(loc='upper left')\n",
    "    plt.ylim([0,1])\n",
    "    \n",
    "fig=plt.subplots(figsize=(10,5))\n",
    "plot_prec_recall_vs_tresh(prec, rec, tre,F1,0.64)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the plot above it is possible to select a set of thresholds (greater than 0.5), that can maximize the precision of the yes predictions. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:40.663489Z",
     "start_time": "2020-03-04T12:34:40.655250Z"
    }
   },
   "outputs": [],
   "source": [
    "glm_pred2=[]\n",
    "for i in glm_proba:\n",
    "    if i>0.6:\n",
    "        i=1\n",
    "    else:\n",
    "        i=0\n",
    "    glm_pred2.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:40.704288Z",
     "start_time": "2020-03-04T12:34:40.669345Z"
    }
   },
   "outputs": [
    {
     "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>GLM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUC</th>\n",
       "      <td>0.79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prec</th>\n",
       "      <td>0.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rec</th>\n",
       "      <td>0.61</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1</th>\n",
       "      <td>0.71</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Acc</th>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       GLM\n",
       "AUC   0.79\n",
       "Prec  0.83\n",
       "Rec   0.61\n",
       "F1    0.71\n",
       "Acc   0.74"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "GLM=report(y_test_glm,glm_pred2,glm_proba,'GLM')\n",
    "GLM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We use the threshold of 0.6, that ensure an 83% of good classification of 1.\n",
    "\n",
    "This logic that drives our prediction process will be used for all the following non parametric models.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.2  Non-parametric approach"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Normalization:\n",
    "We use the normalize dataset to exploit the non parametric approaches "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:40.779118Z",
     "start_time": "2020-03-04T12:34:40.711345Z"
    }
   },
   "outputs": [],
   "source": [
    "data_norm=data.copy()\n",
    "\n",
    "numi=['age','campaign','cons.conf.idx','cons.price.idx','emp.var.rate','euribor3m','nr.employed','previous']\n",
    "data_norm[numi]=preprocessing.normalize(data[numi])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Target Encoding:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are lots of categorical variables in our dataset. Making One Hot Encoding for non-parametric model, lead to an increase of the number of features affecting the quality of our classifier. Therefore we decided to use the 'Target Encoding' method to avoid this issue."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:40.797084Z",
     "start_time": "2020-03-04T12:34:40.782917Z"
    }
   },
   "outputs": [],
   "source": [
    "encoder=ce.TargetEncoder(smoothing=5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The choice of a smoothing parameter was made because we wanted to be sure to not cause overfitting."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:40.831313Z",
     "start_time": "2020-03-04T12:34:40.802561Z"
    }
   },
   "outputs": [],
   "source": [
    "Y=data_norm['y']\n",
    "X=data_norm.drop('y',axis=1)\n",
    "x_train,x_test,y_train,y_test=train_test_split(X,Y, test_size=0.2, random_state=123)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:41.338176Z",
     "start_time": "2020-03-04T12:34:40.841116Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ANDREA\\Anaconda3\\lib\\site-packages\\pandas\\core\\frame.py:3509: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self[k1] = value[k2]\n"
     ]
    }
   ],
   "source": [
    "c=['job','marital','month','day_of_week','poutcome','pdays_new','education','loan','housing','contact']\n",
    "x_train[c]=encoder.fit_transform(x_train[c],y_train)\n",
    "x_test[c]=encoder.transform(x_test[c])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2.1 Decision tree"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first step we made is the cross validation to set the best combination of parameters:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:47.844895Z",
     "start_time": "2020-03-04T12:34:41.346132Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'min_samples_leaf': 0.001, 'max_depth': 6, 'criterion': 'entropy'}   0.7316258351893096\n"
     ]
    }
   ],
   "source": [
    "dt = DecisionTreeClassifier(random_state=123)\n",
    "params = {'criterion': ['gini', 'entropy'], 'max_depth': np.arange(4, 15, 2), 'min_samples_leaf': [0.1, 0.01, 0.001]}\n",
    "cv_tuning = RandomizedSearchCV(dt, params, random_state=123, scoring='accuracy', return_train_score=True, cv=5, n_iter=20)\n",
    "cv_tuning.fit(x_train, y_train)\n",
    "\n",
    "print(cv_tuning.best_params_,' ',cv_tuning.best_score_)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:48.919130Z",
     "start_time": "2020-03-04T12:34:47.860033Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "dt = DecisionTreeClassifier(random_state=123,min_samples_leaf=0.001,max_depth=6,criterion='entropy')\n",
    "dt.fit(x_train,y_train)\n",
    "dt_pred=dt.predict(x_test)\n",
    "dt_proba = [y for (x, y) in dt.predict_proba(x_test)]\n",
    "\n",
    "\n",
    "fpr, tpr, threshold = roc_curve(y_test,dt_proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "\n",
    "fig=plt.subplots(figsize=(10,5))\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": {},
   "source": [
    "The overall AUC is 0.78"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:48.988519Z",
     "start_time": "2020-03-04T12:34:48.942293Z"
    }
   },
   "outputs": [
    {
     "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>DT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUC</th>\n",
       "      <td>0.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prec</th>\n",
       "      <td>0.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rec</th>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1</th>\n",
       "      <td>0.72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Acc</th>\n",
       "      <td>0.73</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        DT\n",
       "AUC   0.78\n",
       "Prec  0.78\n",
       "Rec   0.67\n",
       "F1    0.72\n",
       "Acc   0.73"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "report(y_test,dt_pred,dt_proba,'DT')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With a threshold of 0.5:\n",
    "\n",
    "- Our model obtains a 78% of precision in the classification of yes;\n",
    "- Our model has a 67 % of recall in the classification of yes;\n",
    "- An F1 score of 0.72;\n",
    "- An overall avvuracy of 73%;\n",
    "\n",
    "We can decide to change the threshold:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:49.872067Z",
     "start_time": "2020-03-04T12:34:49.006386Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "prec, rec, tre = precision_recall_curve(y_test,dt_proba)\n",
    "F1 =(2 * (prec* rec)) / (prec+ rec)\n",
    "fig=plt.subplots(figsize=(10,5))\n",
    "plot_prec_recall_vs_tresh(prec, rec, tre,F1,0.54)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:49.897034Z",
     "start_time": "2020-03-04T12:34:49.875264Z"
    }
   },
   "outputs": [],
   "source": [
    "dt_pred2=[]\n",
    "for i in dt_proba:\n",
    "    if i>0.54:\n",
    "        i=1\n",
    "    else:\n",
    "        i=0\n",
    "    dt_pred2.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:34:49.969059Z",
     "start_time": "2020-03-04T12:34:49.908751Z"
    }
   },
   "outputs": [
    {
     "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>DT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUC</th>\n",
       "      <td>0.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prec</th>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rec</th>\n",
       "      <td>0.61</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1</th>\n",
       "      <td>0.70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Acc</th>\n",
       "      <td>0.73</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        DT\n",
       "AUC   0.78\n",
       "Prec  0.82\n",
       "Rec   0.61\n",
       "F1    0.70\n",
       "Acc   0.73"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "DT=report(y_test,dt_pred2,dt_proba,'DT')\n",
    "DT"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With a threshold of 0.54, we obtain an increse in the precision score (82%) and an overall good performance for the other indeces."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2.3 Random forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first step we made is the cross validation to set the best combination of parameters:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:37:13.696996Z",
     "start_time": "2020-03-04T12:34:49.979789Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.734966592427617   {'n_estimators': 150, 'min_samples_leaf': 0.001, 'max_features': 'sqrt', 'max_depth': 5, 'criterion': 'entropy'}\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier(random_state=33)\n",
    "params = {'n_estimators':[50,100,150,200],'criterion': ['gini', 'entropy'], 'max_depth': np.arange(2, 6), 'min_samples_leaf': np.logspace(-1, -3, 3),'max_features': ['sqrt', 0.2, 0.5]}\n",
    "cv_tuning = RandomizedSearchCV(rf, params, random_state=122223, scoring='accuracy', return_train_score=True, cv=5, n_iter=20)\n",
    "cv_tuning.fit(x_train,y_train)\n",
    "print(cv_tuning.best_score_,' ',cv_tuning.best_params_)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the cv we obtained the best model, that is retrained in order to make the predictions on the test set:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:37:18.349139Z",
     "start_time": "2020-03-04T12:37:13.707729Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rf = RandomForestClassifier(random_state=33, n_estimators=150, min_samples_leaf=0.001,max_features='sqrt',max_depth=5,criterion='entropy')\n",
    "rf.fit(x_train,y_train)\n",
    "rf_pred= rf.predict(x_test)\n",
    "rf_proba = [y for (x, y) in rf.predict_proba(x_test)]\n",
    "\n",
    "fig=plt.subplots(figsize=(10,5))\n",
    "fpr, tpr, threshold = roc_curve(y_test,rf_proba)\n",
    "roc_auc = auc(fpr, tpr)\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": {},
   "source": [
    "The overall AUC of the model is 0.79."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:37:18.415344Z",
     "start_time": "2020-03-04T12:37:18.356645Z"
    }
   },
   "outputs": [
    {
     "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>RF</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUC</th>\n",
       "      <td>0.79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prec</th>\n",
       "      <td>0.79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rec</th>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1</th>\n",
       "      <td>0.73</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Acc</th>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        RF\n",
       "AUC   0.79\n",
       "Prec  0.79\n",
       "Rec   0.67\n",
       "F1    0.73\n",
       "Acc   0.74"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "RF=report(y_test,rf_pred,rf_proba,'RF')\n",
    "RF"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With a threshold of 0.5:\n",
    "\n",
    "- Our model has a precision of 0.79 in the classification of yes;\n",
    "- Our model has a recall of 0.67 in the classification of yes;\n",
    "- Our model has a F1 (the harmonic average of precision an recall) of 0.73 in the classification of yes;\n",
    "- Our model accuracy is 74%;\n",
    "\n",
    "We can decide to change the threshold:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:37:19.517552Z",
     "start_time": "2020-03-04T12:37:18.430072Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig=plt.subplots(figsize=(10,5))\n",
    "prec, rec, tre = precision_recall_curve(y_test,rf_proba)\n",
    "F1 =(2 * (prec* rec)) / (prec+ rec)\n",
    "plot_prec_recall_vs_tresh(prec, rec, tre,F1)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the graph, we decide not to change the threshold because there is a constant behaviour of the precision til the 0.7 and a decrease, in the same interval, of the other two indeces."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2.4 Gradient Boosting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:39:44.232839Z",
     "start_time": "2020-03-04T12:37:19.533160Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'subsample': 0.8, 'n_estimators': 200, 'min_samples_leaf': 0.001, 'max_features': 'sqrt', 'max_depth': 5, 'learning_rate': 0.01}   0.7430400890868597\n"
     ]
    }
   ],
   "source": [
    "gb = GradientBoostingClassifier(random_state=123)\n",
    "\n",
    "params = {'learning_rate': np.logspace(-1, -4, 4),'n_estimators':[50, 100, 150,200],'min_samples_leaf': np.logspace(-1, -4, 4), 'max_features': ['sqrt', 0.2, 0.3, 0.5, 0.7, 0.8], 'max_depth':[2,3,4,5],'subsample': [0.4, 0.8, 1]}\n",
    "cv_tuning = RandomizedSearchCV(gb, params, random_state=123, scoring='accuracy', return_train_score=True, cv=5, n_iter=20)\n",
    "\n",
    "cv_tuning.fit(x_train, y_train)\n",
    "\n",
    "print(cv_tuning.best_params_,' ',cv_tuning.best_score_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:39:58.144204Z",
     "start_time": "2020-03-04T12:39:44.315239Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gb = GradientBoostingClassifier(random_state=123,subsample=0.8,n_estimators=200,min_samples_leaf=0.001,max_features='sqrt',max_depth=5,learning_rate=0.01)\n",
    "gb.fit(x_train,y_train)\n",
    "gb_pred=gb.predict(x_test)\n",
    "gb_proba = [y for (x, y) in gb.predict_proba(x_test)]\n",
    "\n",
    "fig=plt.subplots(figsize=(10,5))\n",
    "fpr, tpr, threshold = roc_curve(y_test,gb_proba)\n",
    "roc_auc = auc(fpr, tpr)\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": {},
   "source": [
    "The overall accuracy is 0.80."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:39:58.267932Z",
     "start_time": "2020-03-04T12:39:58.164624Z"
    }
   },
   "outputs": [
    {
     "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>GB</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUC</th>\n",
       "      <td>0.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prec</th>\n",
       "      <td>0.81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rec</th>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1</th>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Acc</th>\n",
       "      <td>0.75</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        GB\n",
       "AUC   0.80\n",
       "Prec  0.81\n",
       "Rec   0.67\n",
       "F1    0.74\n",
       "Acc   0.75"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "GB=report(y_test,gb_pred,gb_proba,'GB')\n",
    "GB"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With a threshold of 0.5:\n",
    "\n",
    "- Our model has a precision of 0.80 in the classification of yes;\n",
    "- Our model has a recall of 0.67 in the classification of yes;\n",
    "- Our model has a F1 (the harmonic average of precision an recall) of 0.73 in the classification of yes;\n",
    "- Our model accuracy is 75%;\n",
    "\n",
    "We can decide to change the threshold:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:39:59.095964Z",
     "start_time": "2020-03-04T12:39:58.274188Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig=plt.subplots(figsize=(10,5))\n",
    "prec, rec, tre = precision_recall_curve(y_test,gb_proba)\n",
    "F1 =(2 * (prec* rec)) / (prec+ rec)\n",
    "plot_prec_recall_vs_tresh(prec, rec, tre,F1)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The threshold will be set to 0.5, because also in this case it permits to follow our goal, without worsening the overall performance."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2.5 Considerations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:39:59.134156Z",
     "start_time": "2020-03-04T12:39:59.101713Z"
    }
   },
   "outputs": [
    {
     "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>GLM</th>\n",
       "      <th>DT</th>\n",
       "      <th>RF</th>\n",
       "      <th>GB</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AUC</th>\n",
       "      <td>0.79</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.79</td>\n",
       "      <td>0.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Prec</th>\n",
       "      <td>0.83</td>\n",
       "      <td>0.82</td>\n",
       "      <td>0.79</td>\n",
       "      <td>0.81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rec</th>\n",
       "      <td>0.61</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.67</td>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1</th>\n",
       "      <td>0.71</td>\n",
       "      <td>0.70</td>\n",
       "      <td>0.73</td>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Acc</th>\n",
       "      <td>0.74</td>\n",
       "      <td>0.73</td>\n",
       "      <td>0.74</td>\n",
       "      <td>0.75</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       GLM    DT    RF    GB\n",
       "AUC   0.79  0.78  0.79  0.80\n",
       "Prec  0.83  0.82  0.79  0.81\n",
       "Rec   0.61  0.61  0.67  0.67\n",
       "F1    0.71  0.70  0.73  0.74\n",
       "Acc   0.74  0.73  0.74  0.75"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SCORE=GLM.join([DT,RF,GB])\n",
    "SCORE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:01.949208Z",
     "start_time": "2020-03-04T12:39:59.140857Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ((ax1,ax2),(ax3,ax4))=plt.subplots(nrows=2,ncols=2,figsize=(20,10))\n",
    "\n",
    "SCORE.loc['AUC',:].plot.bar(ax=ax1,title='AUC',color='g')\n",
    "SCORE.loc['Prec',:].plot.bar(ax=ax2,title='Prec',color='r')\n",
    "SCORE.loc['Rec',:].plot.bar(ax=ax3,title='Rec',color='b')\n",
    "SCORE.loc['F1',:].plot.bar(ax=ax4,title='F1',color='y')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Observing the plot, it is possible to see that the best precision for the 1's is obtained by the logistic model (83%). Despite this fact, the logistic model is not the best for the other three indeces, that show the overall quality of the model. \n",
    "The model that overperforms all the other models in AUC, Recall and F1 is the Gradient Boosting, reaching also a satisfying level for the precision:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:02.101488Z",
     "start_time": "2020-03-04T12:40:01.968312Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The indeces for the GB model are:\n",
      "        GB\n",
      "AUC   0.80\n",
      "Prec  0.81\n",
      "Rec   0.67\n",
      "F1    0.74\n",
      "Acc   0.75\n"
     ]
    }
   ],
   "source": [
    "print('The indeces for the GB model are:',GB,sep='\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.3 Deeper evaluation: Lift analysis "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:02.305369Z",
     "start_time": "2020-03-04T12:40:02.165857Z"
    }
   },
   "outputs": [],
   "source": [
    "def cum_lift(proba,y_t):\n",
    "    quantiles = list(np.quantile(proba, q=np.linspace(0,1,5)))\n",
    "    d_perc = pd.DataFrame(columns=['P_inf', 'P_sup', 'P_avg', 'Attempts', 'Success','Success_perc']) \n",
    "    for i in range(len(quantiles) - 1):\n",
    "        p = [x for x in proba if quantiles[i+1] > x >= quantiles[i]]\n",
    "        q = y_t.loc[[quantiles[i+1] > x >= quantiles[i] for x in proba]]\n",
    "        d_perc.loc[i+1] = [quantiles[i], quantiles[i+1], np.mean(p), q.count(), q.sum(), q.mean()]\n",
    "    d_perc.sort_values('P_inf', ascending=False, inplace=True)\n",
    "    d_perc['Attempts_cum'] = d_perc['Attempts'].cumsum()\n",
    "    d_perc['Success_cum'] = d_perc['Success'].cumsum()\n",
    "    d_perc['Captured_resp'] = d_perc['Success'] / d_perc.loc[1, 'Success_cum']\n",
    "    d_perc['Captured_resp_cum'] = d_perc['Success_cum'] / d_perc.loc[1, 'Success_cum']\n",
    "    d_perc['Lift'] = d_perc['Success'] / (d_perc['Attempts'] * y_t.mean())\n",
    "    d_perc['Lift_cum'] = d_perc['Success_cum'] / (d_perc['Attempts_cum'] * y_t.mean())\n",
    "    return(d_perc[['Success_perc','Captured_resp_cum','Lift_cum']])\n",
    "\n",
    "def gain_plot(model):\n",
    "    quantile=[0.0,0.25,0.5,0.75,1]\n",
    "    v=0\n",
    "    l=np.array(model['Captured_resp_cum'])\n",
    "    X = np.insert(l,0,v, axis=0)\n",
    "    plt.xlim([0, 1])\n",
    "    plt.title('Cumulative gains chart')\n",
    "    plt.xlabel('% customers contacted')\n",
    "    plt.ylabel('% customers reached')\n",
    "    plt.ylim([0, 1])\n",
    "    plt.plot([0, 1], [0, 1],'r--',label='Baseline')\n",
    "    plt.plot(quantile,X,label='Gain')\n",
    "    plt.legend(loc=\"upper left\")\n",
    "    plt.show()\n",
    "    \n",
    "def lift_plot(model):\n",
    "    quantile=[0.25,0.5,0.75,1]\n",
    "    l=np.array(model['Lift_cum'])\n",
    "    plt.xlim([0.25, 1])\n",
    "    plt.title('Cumulative lift chart')\n",
    "    plt.xlabel('% customers contacted')\n",
    "    plt.ylabel('Lift')\n",
    "    plt.axhline(y=1,c='r',ls='--',label='Baseline')\n",
    "    plt.plot(quantile,l,label='Lift')\n",
    "    plt.legend(loc=\"upper right\")\n",
    "    plt.show()\n",
    "    \n",
    "def wtf_plot(model,color):\n",
    "    plt.bar(['1st','2nd','3rd','4th'],np.array(model['Success_perc'])*100,color=color)\n",
    "    a=y_test.mean()*100\n",
    "    plt.axhline(y=a,c='k',ls='--',label='Overall response rate')\n",
    "    plt.legend(loc=\"upper right\")\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We decided to split the client population into four groups (based on quartiles, sorted from more probable to the less probable that clients purchase banking products): Loyal clients (4), Clients (3), Possible clients (2), Newer (1). <br>\n",
    "Thanks to this ripartition, we will be able to build three indeces that represent the ability of the model to catch people that purchase banking products.\n",
    "The indeces are: \n",
    "- Percentage of success: it is computed for each class and gives the proportion of people that will become a client;\n",
    "- Cumulative gain score: it is the percentage of the overall number of cases in a given category \"gained\" by targeting a percentage of the total number of cases;\n",
    "- Cumulative lift score: it is the cumulative score of how much better one can expect to perform with the predictive model with respect to not using the model;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:02.963109Z",
     "start_time": "2020-03-04T12:40:02.316582Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic evaluation:\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>Success_perc</th>\n",
       "      <th>Captured_resp_cum</th>\n",
       "      <th>Lift_cum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.879464</td>\n",
       "      <td>0.425486</td>\n",
       "      <td>1.703903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.594655</td>\n",
       "      <td>0.713823</td>\n",
       "      <td>1.427696</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.347439</td>\n",
       "      <td>0.882289</td>\n",
       "      <td>1.175990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.242762</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999478</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Success_perc  Captured_resp_cum  Lift_cum\n",
       "4      0.879464           0.425486  1.703903\n",
       "3      0.594655           0.713823  1.427696\n",
       "2      0.347439           0.882289  1.175990\n",
       "1      0.242762           1.000000  0.999478"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Decision tree evaluation:\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>Success_perc</th>\n",
       "      <th>Captured_resp_cum</th>\n",
       "      <th>Lift_cum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.868217</td>\n",
       "      <td>0.395760</td>\n",
       "      <td>1.682112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.544355</td>\n",
       "      <td>0.713781</td>\n",
       "      <td>1.329654</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.321951</td>\n",
       "      <td>0.869258</td>\n",
       "      <td>1.105820</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.264916</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.960795</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Success_perc  Captured_resp_cum  Lift_cum\n",
       "4      0.868217           0.395760  1.682112\n",
       "3      0.544355           0.713781  1.329654\n",
       "2      0.321951           0.869258  1.105820\n",
       "1      0.264916           1.000000  0.960795"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random forest evaluation:\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>Success_perc</th>\n",
       "      <th>Captured_resp_cum</th>\n",
       "      <th>Lift_cum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.883929</td>\n",
       "      <td>0.427646</td>\n",
       "      <td>1.712552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.601336</td>\n",
       "      <td>0.719222</td>\n",
       "      <td>1.438495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.325167</td>\n",
       "      <td>0.876890</td>\n",
       "      <td>1.168793</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.253898</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999478</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Success_perc  Captured_resp_cum  Lift_cum\n",
       "4      0.883929           0.427646  1.712552\n",
       "3      0.601336           0.719222  1.438495\n",
       "2      0.325167           0.876890  1.168793\n",
       "1      0.253898           1.000000  0.999478"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gradient Boosting evaluation:\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>Success_perc</th>\n",
       "      <th>Captured_resp_cum</th>\n",
       "      <th>Lift_cum</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.890625</td>\n",
       "      <td>0.430886</td>\n",
       "      <td>1.725526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.612472</td>\n",
       "      <td>0.727862</td>\n",
       "      <td>1.455774</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.318486</td>\n",
       "      <td>0.882289</td>\n",
       "      <td>1.175990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.242762</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999478</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Success_perc  Captured_resp_cum  Lift_cum\n",
       "4      0.890625           0.430886  1.725526\n",
       "3      0.612472           0.727862  1.455774\n",
       "2      0.318486           0.882289  1.175990\n",
       "1      0.242762           1.000000  0.999478"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print('Logistic evaluation:')\n",
    "logit_report=cum_lift(glm_proba,y_test)\n",
    "display(logit_report)\n",
    "print('Decision tree evaluation:')\n",
    "dt_report=cum_lift(dt_proba,y_test)\n",
    "display(dt_report)\n",
    "print('Random forest evaluation:')\n",
    "rf_report=cum_lift(rf_proba,y_test)\n",
    "display(rf_report)\n",
    "print('Gradient Boosting evaluation:')\n",
    "gb_report=cum_lift(gb_proba,y_test)\n",
    "display(gb_report)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the tables above, we decided to select the two best models, according to the cumulative of the lift score : Logistic and Gradient boosting. <br>\n",
    "Although the models seem to be very similar,this choice is made in terms of 1's prediction ability."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:07.266822Z",
     "start_time": "2020-03-04T12:40:02.983241Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic:\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAESCAYAAAD5d3KwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdd1yV5fvA8Q8gQ8QtIQ4cuHAQoqlpam5zZA4ENdxbv5o50vKHpmWOysrKMmfmLlNJG6KVqYmKggMUU2Q4AJEt+zy/P55ESeGgcs6Bc6736+WrM59zeYdc517XbaYoioIQQgjxL3NDByCEEKJ4kcQghBAiD0kMQggh8pDEIIQQIg9JDEIIIfKQxCCEECIPSQyiSOXk5LBhwwYGDBhAv3796NWrFytWrCAzM1Pnnx0VFUXz5s21vm7Xrl1s2bIFgG3btrFmzRpdh5YvfXx+586dOX/+/DNd49y5c/j4+BRRRKK4K2XoAIRxWbhwIYmJiWzatImyZcty7949Zs2axTvvvMOKFSsMHR4AAQEB1K9fH4AhQ4YYNBZDf35h/fPPP0RHRxs6DKEnkhhEkYmKisLX15ejR49iZ2cHgK2tLe+++y5nzpwBYO7cudSvX58xY8Y8cr9z58706dOHEydOkJiYyNixYzlz5gwXL16kVKlSrF69GgcHBzp37synn35Ks2bNAHLvV6xYMTeWO3fu4OPjQ1xcHLGxsVSvXp1PPvmEM2fOcPjwYY4dO4aNjQ13794lPj6ezp07s2zZMnx9fQFISkqiS5cu+Pn5kZ6ezqJFi7h16xZZWVn07t2biRMnPvL3Dw8P5+233yYxMRF7e3sUReHVV19lwIABfPXVVxw6dIj09HTS0tJ466236NatG6tWrSI+Ph4fHx86d+5M//79+fvvv7l16xb9+vXjjTfeIDU1lXnz5hEeHo65uTlNmjRh0aJFmJvn7fCHhYXh4+PD3bt3MTc3Z9KkSfTq1QuAHTt2sGDBAu7evUu/fv2YMWMGGo2GJUuWEBQURGpqKoqi8N5779GiRQvmzp1LQkICkZGRPP/88xw/fpzk5GTmzZvHBx98UMQ/OaK4kaEkUWQuXrxIvXr1cpPCffb29vTo0aNQ18jIyGDnzp1Mnz4dHx8fRowYwb59+3B0dOTHH38sdCz79+/Hzc2NHTt2cOjQIWxsbNi7dy/dunWjc+fOjBw5kmHDhuW+vl27dqSmpuYOufz000907NiR8uXLM3v2bAYOHMju3bv5/vvvOX78OAcOHHjkM+fMmUPv3r356aefmD9/PoGBgQDcuHGD48ePs3nzZnx9fZkxYwafffbZY+O+d+8eW7duZfv27axfv57IyEgOHjxIamoqe/fu5fvvvwcgMjLykfe++eab9OzZk/3797NmzRo+/vhjUlJSALC2tmb37t3s2rWL9evXc+vWLYKCgoiJiWHHjh0cOHCA/v3788033+ReLz09nf3797NkyRKmTZtGy5YtJSmYCOkxiCJjbm6ORqN5pmt0794dgJo1a1KlShUaNWoEgJOTE4mJiYW+zogRIzh9+jQbNmzg+vXrXLlyheeffz7f15uZmTFw4EB+/PFHmjVrxu7du5kzZw737t3j1KlTJCYm8umnnwLqL+9Lly7lfhsHSExM5Ny5c3z33XcAODs706ZNGwCqV6/O8uXL8fX1JTw8PPcb+uN06dIFAAcHBypXrkxiYiItWrRg5cqVeHt707ZtW0aMGEGtWrXyvC8hIYFLly7h4eEBgKOjI35+frnP9+nTB1CTdJUqVYiLi6N58+aUL1+e7du3ExkZib+/P2XKlMl9T4sWLQrX2MLoSI9BFBlXV1euXbuW+y31vujoaMaPH096ejpmZmY8XJ4rKysrz2utrKxyb1taWub7WQ9f43ET2ytWrMgdXvL09KRdu3ZoKws2aNAgfvnlF0JCQkhOTqZVq1ZoNBoURWH79u3s3buXvXv3smPHDiZMmJDnvRYWFo/Edf+xixcv4unpSUpKCu3atWPs2LH5xmBtbZ17+35b1axZk4MHDzJ+/HhSUlIYNWoUhw8fzvO+UqVK5b7nvmvXrpGenp7n+Yev+8cff+T+Pbp06fLIfIetrW2B7SWMlyQGUWQcHBzo27cvb7/9dm5ySElJYeHChVSoUAEbGxsqVqzIhQsXADVhnDx58ok/p1KlSrnX8Pf3JzY29pHXHD16lBEjRvDaa69RuXJljh8/Tk5ODqD+ws7Ozn5s/K6urvj4+DBo0CAA7OzscHNzY8OGDYA69zBkyBAOHTqU5712dna4u7uze/duQB3q+fvvvzEzM+PUqVM0bdqUUaNG0apVKw4dOpQbS2Fs3bqVefPm8dJLLzF79mxeeuklgoODH/n8Jk2asGfPHgBu3brFkCFDSE5Ozve6x44do1OnTgwdOpSmTZvi5+eXb1z5tZkwTpIYRJFasGAB9erVw8vLi379+uHh4UG9evV47733APD29iY2NpYePXrw9ttv5w63PIlZs2bx7bff0q9fP/bu3UuTJk0eec2UKVNYvnw5ffv2ZdKkSbi7uxMREQFAhw4d2L59O19//fUj7/Pw8CAkJIT+/fvnPvbhhx8SFBRE37598fDwoE+fPrz66quPvHfZsmX8/PPPvPrqqyxatIgaNWpgY2NDnz59iI+P55VXXqFXr17Y2tqSmJj4SM8qP6+99ho5OTn06tWLAQMGkJycjLe39yOv++ijj3I/f+LEibz//vvY29vne10vLy9OnjxJ37596d+/PzVr1iQqKuqxw4Fubm5ERkYyderUQsUsSjYzKbstRNFYvXo13bt3x9nZmeTkZF599VW++eYb6tWrZ+jQhHgiMvksRBGpXbs2M2bMwNzcnJycHMaNGydJQZRI0mMQQgiRh87mGIKCgh47Dnr48GEGDhyIp6cnO3fu1NXHCyGEeEo6GUr65ptv2LdvH6VLl87zeFZWFh988AHff/89pUuXZsiQIXTq1KnACTIhhBD6pZPE4OTkxKpVq5gzZ06ex69evYqTkxPly5cH1A00p0+f5pVXXinwegEBAY9s/zdVGo1G2uJf0hYPSFs8YKptka1RSM3UkJKpIS0zBw1mNLS3wc3N7YmvpZPE0KNHD6Kioh55PCUlhbJly+beL1OmTKGW7JmbmxeqaqYpCAkJwcXFxdBhFAvSFg9IWzxgKm2hKAqh0Sn4hUTzW3A0QZEJANSoWJpu2dF0y7iF9ZCCv3TnR6+rkuzs7PKUAkhNTc2TKIQQQuQvK0fDqet38QuO4WDIbSLvpgHwvIMts2JO0rVdIxqO7pW7Az4kJOSpPkevicHZ2Znw8HASEhKwtbXl9OnTuVU2hRBCPCo5PYs/Q2M5GBzN75diSErPxqqUOS/Vq8Kkjs50PePHc3PfhKwsaL0SHiqL8rT0khh8fX25d+8enp6ezJ07lzFjxqAoCgMHDsTBwUEfIQghRIlxIyGNQyHRHAyO5sS1OLJyFCqVsaJ7k6p0a+xA+/pVsI0Mh3Gj4PffoVMn+OYbcHYuks/XWWKoUaNG7nLUvn375j7euXNnOnfu/MzXz8rKIioqKrdImKnIysp6ou6hjY0NNWrUKLAgnRDCsBRF4eLNJA4Gq8kg+FYSAHXtyzC6XR26NXaguVNFLMwf6g2cPw8BAbBmDYwdWyQ9hftK7M7nqKgoypYtS+3atfNUlDR2aWlpjywDzo+iKMTFxREVFUWdOnV0HJkQ4klkZOdw4tpdDgbf5lBIDLcS0zE3gxa1KvJ2r0Z0cXHA2T7v2SZcuABnzsDw4fDaa3DtGlSuXOSxldjEkJ6ebnJJ4UmZmZlRuXLlx1YfFULoX8K9TA5fisEvJJo/L8eSmplDaUsLOjSowszuDenU0J7KdtaPvjEzE5YsUf84OMDgwWBjo5OkACU4MQCSFApB2kgIw7p+JxW/f+cLTofHk6NReK6sNa+6Vad7YwdedK6MjaVF/hfw94cxY+DiRXj9dVi5Uk0KOlSiE4MQQhQ3Go3C2cgE/EKi8QuO5kqMulerUdWyTH7Zma4uDjSrXh5z80J8abtxA9q3V3sJP/0EvXvrOHqVJIZn4O/vzxtvvEG9evVQFIXs7Gzef/99nJ9xZcCqVauoUqUKbm5uHDp0SGrgC1HMpWXmcPSfO/gFR3PoUjR3UjKxMDejdZ1KDG3tRFcXB2pWeoIT8UJDoUEDqF4dduyALl2gXDnd/QX+QxLDM2rTpg0rV64E1FPDli9f/tgDYJ6Gi4uLSezgFKIkik3O4PClaA4Gx3D0n1jSszSUtS5Fx4b2dGvswMsNnqO87ROuBkxIgDlzYO1a+OMP6NABHjo0Sl+MJzG8/PKjjw0eDJMnw7178NDB7blGjlT/3LkD/x7lmOuPP544hKSkJKpXr87Jkyf5/PPPAXWSfNmyZVSrVo3p06eTkpJCeno6s2fPpnXr1vz8889s3LgRc3NzWrRowaxZs3Kv5+/vz/bt21m5ciXdu3fH3d2dq1evYm9vz6pVq9BoNCxYsIDw8HA0Gg1vvPEGrVu3fuK4hRDaKYrCPzEpHPx3iOhsZAKKAtUrlMbrBbVX0KpOJaxKPWWdpn37YNIkuH0bZs+GF14o2r/AEzCexGAgJ06cwNvbm8zMTC5fvszXX3/NlStXWLFiBQ4ODnz11Vf88ssvdO3alTt37rBx40bi4uK4fv06CQkJrFq1ih9++IHSpUsze/Zsjh079tjPiYyMZNOmTVSoUIFRo0Zx/vx5goODqVixIkuWLCE+Pp7XX3+d/fv367kFhDBe2TkaTofH4xccjV9INNfj7gHgWqM8M7o2oKuLAy6OZZ99kcfYsbBuHTRrBnv3QsuWRRD90zOexFDQN3xb24Kfr1LlqXoIkHco6dq1a3h5ebFkyRLef/99bG1tiY6Oxt3dnfr16zNs2DDefPNNsrOz8fb2JiIigrt37zJ+/HhArR0VGRn52M+pWLEijo6OpKWl4ejoSEZGBqGhoQQEBHDu3DkAsrOziY+Pp2LFik/1dxFCQEpGNkdCY/ELjubw5RgS7mVhZWFO23qVGdu+Ll1cnsOxfOH2EhXo/hlpZmZqIqhVC956C6ysnv3az8h4EkMxUKVKFQDmz5+Pn58fdnZ2vPXWWyiKwuXLl0lNTWXNmjXExMTg5eXF999/j6OjI+vXr8fS0pLdu3fj4uKCn5/fI9d+3DeSunXrUrVqVSZOnEh6ejqrV6/OLWkuhCi8W4lp+IXEqCUorsaRmaOhgq0lnRs9RzcXB9o3sMfOugh/XUZGwsSJ4OUF3t7q7WJEEsMzuj+UZG5uTmpqKnPnzuXy5csMHjyYcuXKUaVKFWJiYqhduzZffPEFe/bswdLSkmnTplGpUiVGjhyJt7c3OTk5VK9eXevZFA/z8vJi/vz5vP7666SkpDB06FCTrEMvxJNSFIXgW0m5VUov3FBLUNSubMuItrXo6uJAi1oVKWVRxP+eNBr4+mu1Z5CTY5CJ5cIoEWc+nz179pHzGEyl5vp/PUlJjPuMta2M9e/1NKQtHsivLTKzNfiHxf07XxDDjYQ0zMzA3akiXV0c6Nb4OZzt7XS3KfTKFXUu4cgR6NpVrXGk41I1T/tzIT0GIYTRSryXxR+hMfwWHM2Ry7EkZ2RjY2lO+/r2TO9Sn06NnsO+7GNKUOhCcDCcOwfr16urIYtxVQJJDEIIo3I7OYu/j4ZxMDiak9fvkqNRqGJnTW9XR7q6OPBS/SoFl6AoSkFBEBgII0ZAv35q0bsSsDhEEoMQokTTaBTO3UjE79+S1ZejkwFo4GDHhA516drYAbcaFQpXgqKoZGTAe+/B0qXg6Aienmp9oxKQFEASgxCiBErPyuH41Tsc/He+IDY5AwtzM16oXZHxLSsxrJMrtSqXMUxwf/+tFr0LCVHLY3/8sc6L3hU1SQxCiBIhLiWDQ5di8AuO5q8rd0jLysHOuhQdG9jTtfFzdGr4HBVsrQgJCTFcUrhxAzp2hKpV4cABeIJVhsWJJAYhRLF1NTZF7RUERxMQEY+igGN5Gwa1qEG3xg60rlsJ61J6mi8oSEgIuLioRe927lSL3pUta+ionpokhmcUGRnJihUruH37NjY2NtjY2DB79mzq16//2NdPnTo1t46SECKvHI1CQHh8bsnqa3dSAWhSrRzTu9Snq4sDTaqVKz7njMTHw8yZsGGDugy1fXv1ZLUSThLDM0hLS2PSpEksXrw4d5/FuXPnWLRoEZs3b37seyQpCPGo63dS2Xj8OnsDbxB/LwtLCzPa1K3MqHa16eLiQLUKRVCCoqj9+KNapDM2FubNM2jRu6JmFInhh4Aodp5+fI2hpzW4ZU0GtqhR4Gt+//132rRpk2fznaurK99++y2hoaEsXboUjUZDUlIS8+fPx93dnXbt2nHs2DG8vb1p1KgRV65cISUlhU8//ZTq1asX6d9BiOJMURT+vhrH+mNhHLoUQylzM3o2daRnk6p0aFCFsjZPWLJan0aPVnsJbm6wfz+4uxs6oiJlFInBUKKionBycsq9P2nSJFJSUoiJiWHixIm89dZbNGzYEF9fX3bv3o37f354XF1deeedd1i5ciX79+/PLaYnhDFLz8phX+BN1h8L49LtZCqXseJ/nevzemsnnitXjFfvPFz0rk0bqF8fZs0Cy2KcwJ6SUSSGgS1qaP12rwtVq1blwoULufdXr14NwODBg6lZsyZffvklNjY2pKamYmdn98j7GzdunHudO3fu6CdoIQwkJimd706Es8U/grjUTBpVLcvyga686lZNfxvOnlZ4OEyYAEOHqktQjfxLnFEkBkPp0qUL33zzDYGBgbi5uQEQHh7O7du3mTNnDt988w3Ozs589tln3Lhxw8DRCmEYF24ksv5oGL7nbpKtUejS6DlGt6vDi86Vi88kcn40Gli9GubOVXsMHh6GjkgvJDE8gzJlyrB69Wo++ugjPvzwQ7KzsylVqhSLFy/m2rVrTJ48mcqVK1O1alXi4+MNHa4QepOjUTgYHM36Y2GcDLuLrZUFw1rXYkTb2tSpYqA9Bk/q8mW16N3Ro9C9u1oVtXZtQ0elF5IYnlGNGjVyD+p5WMeOHRk1atQjj98/oe3hVUtDhgzRXYBC6FFSehY7T0Wy8fh1ouLTqF6hNPN7u+DRsiblS5ewsfjLl+HiRdi4UR0+Ku69myIkiUEI8czuLzfddTqS1MwcWtWuxPzeLnR1cSj6Mw106exZtejdqFHw6qtq0bsKFQwdld5JYhBCPBVFUfj7Whzrj17n0KVoSpmb0de1GqPa1aFZjRJ2kmB6OixaBMuXq7uXhwxR6xuZYFKAEp4YFEUp/pNXBlYCzmESJUx6Vg77gm6y/qi63LRSGSv+16ker7epVbyXm+bn2DG16N3ly2pP4aOPSlzRu6JWYhODjY0NcXFxVK5cAlY2GIiiKMTFxWFj4j/komjEJKfz3YkItpwIL3nLTfNz4wZ06qT2En79VZ1kFiU3MdSoUYOoqChiY2MNHYpeZWVlYfkEG2psbGyoUUP/ezyE8bhwI5H1x8LwDSqBy03zExwMjRurCeGHH9Tk8Ji9RqaqxCYGS0tL6uj4vNTiSM72FfpgFMtNH+fuXXjzTdi0Cf78Ezp0gL59DR1VsVNiE4MQougZ1XLT//rhB5gyBeLi4J13oFUrQ0dUbEliEEIQHnd/uWkUKRnZJXe5aX5GjlR7Ce7u8MsvavE7kS9JDEKYKEVROHHtLuuPheEXUsKXmz7Ow0Xv2rZVD9KZORNKya89bXTSQhqNhoULF3L58mWsrKx47733qFWrVu7z69atY//+/ZiZmTFx4kS6deumizCEEI+RnpWDb9BN1h+7TsitpJK/3PRxwsLUQnevvw4jRhh90buippPE4OfnR2ZmJjt27CAwMJClS5fmVh5NSkpi8+bN/Pbbb6SlpfHaa69JYhBCD2KS09lyIoIt/uHcSTGS5ab/lZNDxc2b4dNPwdwchg0zdEQlkk4SQ0BAAO3btwfAzc0tT2nq0qVLU61aNdLS0khLSyu5y92EKCEu3Ehkw7Hr+AbdJEujMY7lpo8TEgJjxlD177/hlVfgq6/gofNSROHpJDGkpKTkOX/AwsIit/IogKOjI7179yYnJ4cJEyZovZ5GoyEkJEQXoZY46enp0hb/krZ44L9tkaNR8I+6x57gRM5Hp2NTyowe9cvSz6U81ctZQlYsly4Z1x4gu99/xzE4mKjFi0kbMABSU9VkIZ6YThKDnZ0dqampufc1Gk1uUjhy5AgxMTEcOnQIgDFjxuDu7o6rq2u+1zM3N5e1+/+SfQwPSFs8cL8tktOz2Hk6io3Hw4i8qy43faeXC4NfMILlpo8TEABBQepRmy4u8PrrpN24IT8X/3raL046SQzu7u78/vvv9OrVi8DAQBo0aJD7XPny5bGxscHKygozMzPKli1LUlKSLsIQwmTcTMpip+/F3OWmL9SuyNuvuNCtsZEsN/2vtDR491348EOoWVM9Wc3GBsqVU8tciGeik8TQrVs3jh07hpeXF4qisGTJEjZs2ICTkxNdunTh+PHjDB48GHNzc9zd3WnXrp0uwhDCqOVZbhocTSkLM/q4VmNUu9q41jDiqqBHjqgH6Fy5oha/+/BDky96V9R0khjMzc1ZtGhRnsecnZ1zb0+bNo1p06bp4qOFMHqPW27q5VqBN/q0wMFYlpvm58YN6NJF7SX4+am3RZGTnR5ClBCxyRl8dyI8d7lpQ4eyLBvYjH5u1Qn7J9S4k8L589CsmVr07scf1aJ3ZUpwzaZiThKDEMXcxZuJrD+qLjfNzPl3uelLdWhrbMtNH+fOHZgxA7777kHRuz59DB2V0ZPEIEQxlKNR8AuJZv3RMPz/rW46pFVNRrStTV17EygPrSiwaxdMnQrx8bBgAbRubeioTIYkBiGKEZNbbpqfESNg82Zo2RIOHVKHkYTeSGIQohj4b3XTlrUqMu8VF7ob63LTx3m46F3HjuDqCm+8IUXvDEBaXAgD+W91UwszM/q4OjKqXR2er2nEy00f59o1GDdOLXo3apS6DFUYjCQGIfQsIzsH36BbrD8aRvCtJCraWjLl5Xp4v1jLuFcWPU5ODqxapR6cY2EBw4cbOiKBJAYh9CY2OYMt/uF8d0JdbtrAwY6lA5rxWvPqxlPd9EkEB6ulLPz9oXdvteidnE9eLEhiEELHLt5Uq5vuC1SXm3b+t7ppu3omsNy0IGFhcPUqbN0KXl7q3IIoFiQxCKEDORqFQyHRrD8WxolrdyltaYFXq5qMNJXlpvk5dQoCA9X5hN691bmFsmUNHZX4D0kMQhSh5PQsdp2OYuPx60TcvUf1CqV5u1cjPFs6Ud7WhJab/te9e+DjAytXQq1a4O2t1jeSpFAsSWIQoghExN1j4/Hr7DwdmbvcdO4rjUxruWl+/vhDLXp39SpMmADLlknRu2JOEoMQz+DEtTjWHw3joKkvN81PVBR066b2Eg4fVmsciWJPEoMQT+FeZjbz91xg95kbpr3cND9BQfD88+oqo7174eWXwdbW0FGJQso3MezZsyffN7322ms6CUaIkuDy7WSmbD3D1dgUpnWux+RO9UxzuenjxMbC9OmwbZs6hNSxI/TqZeioxBPKNzFcvXoVgMDAQEqXLk3z5s05f/482dnZkhiESVIUhV0BUfjsvYCdtSXfjWlNu3pVDB1W8aAosH07TJsGiYnq6WovvmjoqMRTyjcxzJw5E1DPZF6zZk3u46NHj9Z9VEIUM6kZ2fzfngvsPnuDF+tW5tMhbjxXVoaNcnl7w5YtagXUdeugSRNDRySegdY5hrt375KUlES5cuWIj48nISFBH3EJUWxcvp3M5C0BXLuTyvQu9ZnWpT4W5rIZC41G3ZRmZqZOKrdoofYYLGRYraTTmhgmTpzIwIEDsbOzIyUlhSVLlugjLiEMTlEUdp2OwmefDB094p9/1E1q3t5qWQspemdUtCaGHj160KNHD+Li4ihXrhyWlia8SUeYjIeHjto6V+YTLxk6AiA7Gz75BP7v/8DaWhKCkdKaGE6dOsW7775LTk4OPXv2pFq1anh4eOgjNiEM4tLtJKZsOcO1O6nM6NqAqZ3rydARwIULakns06ehXz/48kuoVs3QUQkd0Lol85NPPuG7776jSpUqTJw4kW3btukjLiH0TlEUdpyKoN/nx0hMy2bLmNZM7yrzCbkiIiA8XF199OOPkhSMmNYeg7m5ORUqVMDMzAxra2vKlCmjj7iE0KvUDHXD2o9nb9CuXmVWesrQEaCWxA4KgvHj1f0I166BnQkXATQRWhODk5MTH330EQkJCaxZs4Zq8i1BGJlLt5OYvOUM1++k8ma3BkzpJENHpKaq8wiffAJ166pnMFtbS1IwEVqHkt59912qVatGixYtsLW15b333tNHXELonKIobD+pDh0lp2fz3djWshQV1JpGrq5qJdSJE+HMGTUpCJOhNTGkpaVRsWJFnn/+eezs7Ni/f78+4hJCp1IzspmxI5C5u8/zQu1KHJjWnrbOshSVqCjo0UPdi/Dnn+oEc7lyho5K6JnWoaQpU6ZQvXp1qlRR/9GY9IlTwiiE3EpiylYZOsrj7Flo3lwteufrq9Y4Kl3a0FEJA9GaGBRF4YMPPtBHLELolKIobD8VycJ9FylX2pItY9vwonNlQ4dlWNHR6m7lnTsfFL3r2dPQUQkDyzcxZGZmAlCzZk3Onj1Lk4dqn1hZWek+MiGKUEpGNu/8eJ69gTdpX78KHw92w76sCY+bK4pa22j6dEhJgffeg7ZtDR2VKCbyTQw9e/bEzMwMRVE4ceJE7m0zMzMOHTqkzxiFeCYht9QNa9fjUpnVvQGTX66HuakPHQ0dqu5HePFFteidi4uhIxLFSL6J4fDhw4Da/b59+zaOjo6cO3cOV1dXvQUnxLNQFIVtJyN51/ci5UtbsnVcG9rUNeGho4eL3nXvriaFKVOk6J14hNZVSQsWLODHH38EYN++fbz//vs6D0qIZ5WSkc307YG8/eN5WtWpxIHp7U07KYSGqhVQ169X748aJZVQRb60JoaQkBAmT54MwPz58wkODtZ5UEI8i+CbSfRddZSfzt1kdo+GbBrViip2JjqfkHgpThQAACAASURBVJ0Ny5erx2yeOycrjUShFGpVUnx8PBUrViQpKYmcnBx9xCXEE1MUha0nI3jXN5gKpS3ZNq4NrU25l3DunFoSOyAA+veHL74AR0dDRyVKgELtYxg4cCDly5cnOTkZHx8ffcQlxBNJTs/i7R8v4Bukrjpa6elmur2E+6KiIDISdu2CgQPVuQUhCkFrYujUqRMdOnQgPj6eypUrywY3UexcvJnI1K1nCY9LZXaPhkzq6Gy6q46OH1d7ChMnPih6J4UvxRPSmhgOHTrE1q1bycrKQlEUEhIS8PX1LfA9Go2GhQsXcvnyZaysrHjvvfeoVatW7vN//vknX3zxBQCNGzdmwYIFknDEE1MUhe9OhLPop2Aq2pr20JFZaqq6J2HVKnB2VieXra0lKYinonXy+YsvvmDq1Kk4OjrSv39/GjZsqPWifn5+ZGZmsmPHDmbOnMnSpUtzn0tJSWHFihV89dVX7Ny5k+rVqxMfH/9sfwthcpLTs1h6JIb5ey7Qpm5lDkxrb7JJgd9+o26/fmpSmDJFit6JZ6a1x1CxYkWaN2/O9u3bGTBgALt379Z60YCAANq3bw+Am5sbFy5cyH3u7NmzNGjQgGXLlhEZGYmHhweVKlUq8HoajYaQkBCtn2sK0tPTTb4t/onL4IM/o7mdks1I94p4NLUjJvIaMYYOzABK3bpFvd690dSowfVvvyWtRQt1bsGEyb+RZ6c1MVhaWnLq1Cmys7P566+/iI2N1XrRlJQU7B6q225hYUF2djalSpUiPj4ef39/9uzZg62tLcOGDcPNzY06derkez1zc3NcZGcmoC4fNtW2UBSF7/wjWPzLdSrZWrGshz0eL7sbOizDCAiAFi3UHcsHDnDd3p5Gbm6GjqpYMOV/I//1tAmyUOcxZGdnM2nSJHbu3Mm0adO0XtTOzo7U1NTc+xqNhlKl1BxUoUIFmjVrhr29PWXKlKFly5aS3YVWyelZTN12lv/bc4EX61Zm/7SXaOpggmvyb98GDw9o2VItiw3QrRuKDB2JIqQ1MTg4OABw5swZpkyZQteuXbVe1N3dnSNHjgAQGBhIgwYNcp9r2rQpoaGh3L17l+zsbIKCgqhXr97Txi9MwIUbifRZdZRfLtxmTs+GbBj5ApVNbSmqosCmTdC4sVoWe8kSKXondEbrUNLHH3/M7du3uXr1KpaWlqxZs4aPP/64wPd069aNY8eO4eXlhaIoLFmyhA0bNuDk5ESXLl2YOXMmY8eOBdRifQ8nDiHuyx068g2mUhkrto9vwwu1C56PMlpeXmpp7HbtYO1aaNTI0BEJI6Y1MQQEBLBlyxa8vb3p378/27Zt03pRc3NzFi1alOcxZ2fn3Nu9e/emd+/eTxGuMBVJ6VnM++E8+8/f4uWG9nw82I1KZUys3PvDRe969YL27WHyZDDX2tEX4ploTQw5OTlkZGRgZmZGTk4O5vJDKXTswo1Epmw9Q1R8Gm/1bMSEDnVNb8PapUswdiyMHKn+d8QIQ0ckTIjWxDBy5EgGDBjA3bt38fDwYNSoUfqIS5ig+xvWFv8UQmU7K3aMb0NLUxs6ysqCFSvg3XfVzWkPre4TQl+0JoYKFSqwdetWwsPDqVGjhtY9B0I8jaT0LOb+cI4D52/TqaE9H5ni0FFgoLpjOTAQBg1SN6xVrWroqIQJ0poYVq1axZYtW+SAHqEz56PUoaMbCWnMfaUR49ub4NARqEtRb9+GH36AAQMMHY0wYVoTg5mZGVOmTKFOnTq58wtvvvmmzgMTxk9RFL79O5z395vw0NHRo2rRu8mToWdPuHoVbG0NHZUwcVoTw8CBA/URhzAxSelZvPX9OX6+cJvOjZ7jI4/nqWhKQ0fJyTBvnnpGQv36MGaMWt9IkoIoBrQmhv79++sjDmFCHh46mvdKI8aZ2tDRr7/C+PHqWQnTp8N770nRO1GsaE0MQhSVh4eOqthZsXNCG1rUMrGho8hI6NMH6tVTh5Fk97IohiQxCL1ITFOHjn65aIJDR4oCp05Bq1ZQsyb8/DO89BLY2Bg6MiEeS+tutStXrnD27FmCgoIYMWIEf//9tz7iEkbkXFQCfVb9hV9ING/3asTa4S1NJyncuqUeq9m69YOid127SlIQxZrWxLBgwQKsrKxYvXo1M2bM4PPPP9dHXMIIKIrCxmNhDFx9nJwchR0TXmR8BxM5dlNRYMMGtejdzz/DsmVqnSMhSgCtQ0mlSpWifv36ZGVl4ebmRk5Ojj7iEiXcw0NHXRo9x4emNHQEMHgwfP+9Wt9o7VqQQpGiBCnUPoaZM2fSoUMHDhw4QOnSJlgDXzyRoMgEpm47w62EdN7p5cLY9nVM40zvnBy14J25OfTtC507w4QJUvROlDhaE8PKlSs5f/48HTp0wN/fn5UrV+ojLlECKYrCxuPXWXIghOfK2rBz4ou4O1U0dFj6ERKi7kUYNQrGjYPhww0dkRBPTWtimDJlSm6p7TZt2ug8IFEyJaZlMef7IH69GE1XF3XoqIKtCQwdZWWp8weLF6sF78qXN3REQjwzrYmhfPnybNq0KU9JjJdeeknngYmSIzAygalbz3A7MZ35vV0Y85KJDB2dPauWxT53Djw94bPP4LnnDB2VEM9Ma2KoWLEily5d4tKlS7mPSWIQoA4dbTh2nQ9+NsGhI4DoaLhzB/bsgX79DB2NEEVGa2L44IMPCAsLIyIigoYNG/KcfCMSQOK9LGZ/H8RvwdF0dXHgQw9X0xg6OnIEzp+HKVPUonf//AOyIEMYGa2J4bvvvuPgwYMkJibSv39/wsPD8fHx0UdsopgyyaGjpCSYOxdWr1aXno4dq9Y3kqQgjJDWdXT79+9n48aNlC1blhEjRhAUFKSPuEQxpCgK646G4fHVcRQFdk18kbHt6xp/UjhwAJo0ga+/hjffhDNnpOidMGpaewyKogDk/uO3sjKB4QLxiMR7Wcz6PoiDwdF0a+zAh4Oep7ytpaHD0r3ISHX+oGFDdcNa69aGjkgIndOaGPr06cOwYcO4efMm48aNo2vXrvqISxQjgZEJTNlyhpjkdP6vT2NGt6tt3L0ERQF/f2jTRi1699tvajkL+VIkTITWxPD666/z4osvEhoaSt26dWnYsKE+4hLFwP2ho6U/X6JqeRt2TWyLW80Khg5Lt27ehEmTYN8++OMP6NgROnUydFRC6JXWxHDu3Dn2799PRkYG/v7+ACxcuFDXcQkDS7iXyaxd5/ALiaZ7YwdWGPvQkaLAunUwaxZkZMCHH0rRO2GytCaGt956i3HjxlGuXDl9xCOKgbMR8UzdepaY5HR8+jRmlLEPHQEMGgS7d6s9hLVr1YN0hDBRWhNDrVq1GDBggD5iEQb236Gj7ye25XljHjp6uOjda69B9+5qnSMpeidMnNbE0KNHD2bMmIGzs3PuY1OnTtVpUEL/Hh466tHEgeWDnqd8aSMeOrpwQd2LMGaMmgy8vQ0dkRDFhtbEsHXrVrp16yZDSUbsTEQ8//t36GhB38aMbGvEQ0eZmfDBB/D++2rBu4omVMJDiEIqVBG98ePH6yMWoWeKorD2rzCW/WIiQ0cBAWrRuwsXYOhQ+OQTsLc3dFRCFDuFKqLn4+ND48aNc79Fenp66jwwoVvq0FEQfiEx9GxSlWWDXI176AggLg4SEsDXF/r0MXQ0QhRbhZp8Brhz547OgxH6ERAez7Rt6tDRwr6NGWHMQ0e//64WvZs2TZ1cvnIFbGwMHZUQxZrW5RdTp06ladOmWFtb06hRI5l4LuHWHQ3D8+u/MTeHHya1ZWQ7Iy2Al5ioHqvZubNa+C4jQ31ckoIQWmlNDB999BG7d+/G0tKSPXv2sGzZMn3EJXRg7V/XWPxTMF1cnuOn/7XHtYaRzif4+kLjxup+hFmz1LkFKXonRKFpHUo6deoU27dvB2DEiBEMHjxY50GJorc38Abv7Q/hlaZV+XyoOxbmRthLALXo3cCB0KiReoDOCy8YOiIhShytPYbs7Gw0Gg2grmIxymEHI3ckNJZZu4JoXacSKz3djC8pKAocP67evl/07vRpSQpCPCWtiaFXr14MGTKEJUuWMHToUHr16qX1ohqNBh8fHzw9PfH29iY8PPyxrxk7dizbtm17ushFoZyLSmDidwE429vxzYiW2FhaGDqkohUVBa++qtY1+vNP9bGXX5ZKqEI8A61DScOHD+ell17i2rVrDBo0iKpVq2q9qJ+fH5mZmezYsYPAwECWLl3K6tWr87zmk08+ITEx8ekjF1qF3Ull1IZTVCpjxabRrShnY0TLUTUaKuzYAR9/DNnZ6n/lLHIhikS+PYbY2FjCwsIYOnQoVlZWNGzYEAsLC0aPHq31ogEBAbRv3x4ANzc3Lly4kOf5X375BTMzMzp06PCM4Yv8xCSlM3y9Pwrw7ehWOJQzstU4Awfi+O676nDRhQswYwZYGFlvSAgDybfHEBQUxKZNmwgLC8PHxwdFUTA3N+elQnwrS0lJwc7OLve+hYUF2dnZlCpVitDQUH766Sc+++wzvvjii0IFqdFoCAkJKdRrjV16errWtkjN1DDnl5vEJmextIcjGXciCTGGbSjZ2WqBO3NzyrVpQ06zZqR6ealLUU3856MwPxemQtri2eWbGLp27UrXrl35888/6dix4xNd1M7OjtTU1Nz7Go2GUqXUj9qzZw/R0dGMGDGCGzduYGlpSfXq1QvsPZibm+Pi4vJEMRirkJCQAtsiPSuHkRtOEpGYxbqRL9CxgZGUfDh3Ti14N3asuj/BxUVrW5gSaYsHpC0eeNoEqXWOwdbWliNHjqAoCosXL2b69On07du3wPe4u7vz+++/06tXLwIDA2nQoEHuc3PmzMm9vWrVKqpUqSJDSkUkR6Pw5s5ATly7yyeebsaRFDIyYMkS9U/FilLbSAg90LoqacWKFdSuXZtvv/2Wbdu25e5pKEi3bt2wsrLCy8uLDz74gHnz5rFhwwYOHTpUJEGLRymKwru+Fzlw/jbze7vwWvPqhg7p2Z06Be7usGgRDBmiDhfJ2SBC6JzWHoO1tTWVK1emVKlS2Nvbk5mZqfWi5ubmLFq0KM9jD5/ncN///ve/JwhVFOTzw//w7d/hTOhQl7Ht6xo6nKIRHw8pKXDgALzyiqGjEcJkaO0xlClThlGjRvHKK6+wZcsWHB0d9RGXeALbTkbw0cFQBjSvzls9Gxk6nGdz+DB8+ql6u3t3CA2VpCCEnmntMXz22WdERERQr149QkND8fDw0EdcopB+u3ibd348z8sN7Vk2yBXzkrqrOSEBZs9W6xu5uMDEiWp9I6lxJITeaU0Ma9aseeQxqbBaPJy6fpf/bTtLsxoV+HKYO5YWJfSs4r17YdIkiI6GOXNg4UJJCEIYkNbEUKVKFUCd3AwODs6tmyQM6/LtZMZsPEX1iqXZMPIFbK20/q8sniIiwMND7SXs2wctWxo6IiFMntbfJl5eXnnujx07VmfBiMKJir/H8PX+lLay4NvRrahUpoTVBVIUOHoU2rcHJyfw84M2baS+kRDFhNbEEBYWlns7NjaWW7du6TQgUbCk9BymrD/Jvcwcdk18kRoVbQ0d0pOJiFDnD37+Gf74Azp2BNnHIkSxojUx+Pj4YGZmhqIo2NjY5NmgJvTrXmY2PoduE5WQxebRrWhUtZyhQyo8jQa++greekvtMXz2mRS9E6KY0poY1q5dy9WrV2ncuDF+fn60bdtWH3GJ/8jK0TBlyxmuxGXw5bAWtK5b2dAhPZkBA9RJ5m7dYM0aqF3b0BEJIfKhdRnL7NmzCQoKAtRhpblz5+o8KJGXoijM/eE8v1+OZUrrKvRsqr30ebGQna32FAA8PWH9evj1V0kKQhRzWhNDdHQ0Q4YMAWDcuHHExMToPCiR17JfLvPDmShmdG1Ar4YlZPgoKAhat1Z7B6CWtBg1CuQEQCGKvUItfL8/AR0RESHLVfVs3dEwvvrzKsNaOzGtSz1Dh6NdejrMn68uO42KgkIc7CSEKF60zjG88847vPHGG8TFxfHcc889UgNJ6M7ewBss/imYnk2qsqhf0+J/3vbJkzBiBFy6pP7344+hUiVDRyWEeEJaE4Orqyt79+7VRyziIX9diWXWriBa16nEJ15uWJSEUhdJSZCWBr/8Aj16GDoaIcRTKqHbZY3buagEJm4OwNnejjXDW2JjWYyPrPztN7h4UT1as2tXuHxZylkIUcKV0OI6xivsTiqjNpyiYhkrNo1uRfnSloYO6fHi49XJ5B49YN069UAdkKQghBEodGIIDw/n0qVLuozF5MUkpzN8vT8K8O3oVjiUszF0SI+3ezc0bgybN8O8eXD6tCQEIYxIoYaS1q1bx82bNzEzMyM2NpZP79fLF0UmKT2LEetPEZeSydZxbahrb2fokB4vIgK8vKBpU/UAnebNDR2REKKI5dtjWLt2be5pbeHh4fzvf/9j2rRp3Lx5U2/BmYqM7BzGf3uaK9HJrH69BW41Kxg6pLwUBf78U73t5KQepuPvL0lBCCOVb4+hefPmzJ49m+7duzN8+HAWL15Meno606dP12d8Ri9HozBjRyAnrt1lpefzdGxQzA67Dw+HCRPUHcv3i95JjSMhjFq+iaFFixa0aNECX19fPv/8c7y9vWnRooU+YzN6iqLwru9FDpy/zTu9XOjfvIahQ3pAo4Evv4T7JVBWrVLLZAshjF6+iSE0NJRdu3Zha2vL7Nmz2b9/Pz/88AOTJk2iZs2a+ozRaH1++B++/Tuc8R3qMq5DXUOHk9drr4Gvr7rq6OuvoVYtQ0ckhNCTfOcYfHx8GDhwIB06dOCTTz5h/PjxzJw5k02bNukzPqO1/WQEHx0MZUDz6szt2cjQ4aiysh4UvRsyBDZtUs9NkKQghEnJt8dgbW3NyZMnuXfvHuXLlwegcuXKzJ8/X2/BGavfLt7m7R/P07GBPcsGuWJeHHY1nzkDY8bAuHEwebKaGIQQJinfxLB69WqOHTuGra2tnMFQhE5dv8v/tp2lWY0KfDnMHUsLA+8xTEuDRYtgxQqwtwcZJhTC5OWbGGxtbenWrZs+YzF6l28nM2bjKapXKM2GkS9QxtrAFUlOnFCL3YWGwujR8OGHULGiYWMSQhic1ErSkxsJaYxYfxIbSws2jW5FpTLF4OD71FR1XuHgQbXOkRBCIIlBL+JTMxm+zp/UzGx2TniRmpVsDRfML7+oRe9mzoQuXdQS2VbFIEkJIYoNKaKnY/cysxm96RSR8WmsHd4SF0cDncAWF6cOG73yirra6N9d7ZIUhBD/JYlBh7JyNEzdepagyAQ+83Kjdd3K+g9CUeD779Wid1u3qqernTolCUEIkS8ZStIRRVGY+8N5Dl+K4f3+TenZ1NEwgUREwNCh4Oqqnp3w/POGiUMIUWJIj0FHlv96mR/ORPFG1/oMa63nDWKKoha6A3Vz2h9/qCuQJCkIIQpBEoMOrDsaxuo/rjKstRPTu9TX74eHhUH37urE8v2KqG3bQinpHAohCkcSQxHbG3iDxT8F07NJVRb1a4qZmZ52NefkwKefquck+PvD6tVS9E4I8VTka2QR+utKLLN2BdGqTiU+8XLDQp+lLvr1g/37oVcv+Oor2cEshHhqkhiKyPmoRCZuDsDZ3o5vhrfExtJC9x+alQUWFmBuDt7ean2joUNBX70UIYRRkqGkIhB2J5WRG05SwdaKTaNbUb60pe4/9PRpaNlSHTIC8PSEYcMkKQghnplOEoNGo8HHxwdPT0+8vb0JDw/P8/zGjRvx8PDAw8ODzz//XBch6E1McjrD1/ujAJvHtMKhnI1OP88sPR3eegtat4bYWCmJLYQocjpJDH5+fmRmZrJjxw5mzpzJ0qVLc5+LjIxk3759bN++nR07dnD06FEuXbqkizB0Ljk9i5HrT3EnOZP1I1+grr2dbj/w77+p078/LF+uFr0LDoY+fXT7mUIIk6OTOYaAgADa/7sixs3NjQsXLuQ+V7VqVdauXYuFhToGn52djbW1dYHX02g0hISE6CLUp5aZo+Djd4vL0eks7FIV65RbhITc0uln2l66RNWcHMLXrePeiy/CrVvqHxOVnp5e7H4uDEXa4gFpi2enk8SQkpKCnd2Db88WFhZkZ2dTqlQpLC0tqVSpEoqisHz5cho3bkydOnUKvJ65uTkuLi66CPWp5GgUpm07S9DtdFZ6Pq/bs5oPHFCL3s2eDS4uhLRogYurq+4+rwQJCQkpVj8XhiRt8YC0xQNPmyB1MpRkZ2dHampq7n2NRkOphzZYZWRkMGvWLFJTU1mwYIEuQtAZRVF41/ci+8/f4p1eLrpLCnfuwOuvQ+/esGXLg6J3lnqY2BZCmDSdJAZ3d3eOHDkCQGBgIA0aNMh9TlEUJk+eTMOGDVm0aFHukFJJ8cXv//Dt3+GM71CXcR3qFv0HKAps3w4uLrBzJyxYACdPStE7IYTe6GQoqVu3bhw7dgwvLy8URWHJkiVs2LABJycnNBoNJ0+eJDMzk7/++guAN998k+bNm+silCK1/WQEH/4WSv/m1Znbs5FuPiQiQi2P/fzzsG4dNGumm88RQoh86CQxmJubs2jRojyPOTs7594+f/68Lj5Wpw4GR/P2j+fp2MCe5YNcMS/KXc2KAocOqaeo1aql1jh64QV185oQQuiZbHArhNPX7zJ16xmaVS/Pl8PcsbQowma7elUteNet24Oid23aSFIQQhiMJAYtQqOTGb3xFNUrlGb9yBcoY11EnaycHPj4Y3WoKCAAvv5ait4JIYoFqZVUgBsJaQxfdxIbSws2jW5FZbuC91s8kb594eef1Q1qq1dDDR0ueRVCiCcgiSEf8amZDF/nT2pmNjsnvEjNSrbPftHMTPVcBHNzGDlSLXzn5SX1jYQQxYoMJT3GvcxsRm86RWR8Gt8Mb4mLY7lnv+jJk9CiBXz5pXp/8GC1GqokBSFEMSOJ4T+ycjRM3XqWoMgEPvNyo03dys92wXv3YOZMePFFiI+Hh1ZnCSFEcSRDSQ9RFIV5u89z+FIM773WlJ5NHZ/tgkePqnsSrl2DCRNg2TIoX75oghVCCB2RxPCQ5b9e5vuAKN7oWp/X2xRBOev7B+n8/ju8/PKzX08IIfRAEsO/1h8NY/UfVxna2onpXeo//YV8fSEkBObMgU6d1NLYpaSZhRAlh8wxAPuCbrLop2B6NqnK4n5NMXuaCeHYWPVYzVdfhW3bHhS9k6QghChhTD4x/HUllpk7A2lVpxKfeLlh8aSlLhQFtm5Vi959/z0sWgT+/lL0TghRYpn019nzUYlM3ByAs70d3wxviY3lU5ShiIiAUaOgeXO16F2TJkUfqBBC6JHJ9hjC7qQycsNJKthasWl0K8qXfoJzDjQa+PVX9XatWvDXX3DsmCQFIYRRMMnEEJOczvD1/mgUhW/HtMKhnE3h33zlCnTuDD17wr9nTtCqlRS9E0IYDZNLDMnpWYxcf4o7yZlsGNUKZ3s77W8CyM6GFSvA1RUCA9VhIyl6J4QwQiY1x5CRncOEzQGERiezdkRL3GpWKPyb+/RRh4/69VPLWlSrprtAhRDCgEwmMeRoFN7cEcTxq3F8PPh5Xm74nPY3ZWSoZyybm8PYsTB6NHh4SH0jIYRRM4mhJEVRWOR7kf3nb/FOLxcGuBeixPWJE+DuDl98od4fNEgtfCdJQQhh5EwiMXz5x1U2/R3OuPZ1GNehbsEvTk2FGTOgbVtITob6z7ALWgghSiCjH0rafjKCFb9epn/z6sx7xaXgF//1l1r0LiwMJk+GDz6AckVQclsIIUoQo04MB4OjefvH83RoYM/yQa6Ya9vVnJ2tzin8+Sd06KCfIIUQopgx2sRw+vpdpm49Q7Pq5Vk9zB1Li3xGzfbsUYvezZunFr27eFHqGwkhTJpRzjGERiczeuMpqlcozfqRL1DG+jG/6KOj1cnk/v3VGkdS9E4IIQAjTAw3E9IYsf4kNpYWbBrdisp21nlfoCiweTM0bgx798L776srkKTonRBCAEY2lBSfmsnw9SdJSc9m58QXqVnJ9tEXRUSoexJatlR3LzdqpP9AhRCiGDOaHkNaZg6jN50i4u49vhnREhfHh1YTaTTw88/q7Vq11IJ3R45IUhBCiMcwisSQlaNhytYzBEYm8JmXG23qVn7wZGioeqxmr17qaiNQewtS9E4IIR6rxCcGRVF4e/d5Dl+KYXG/pvRs6qg+kZ0Ny5apRe/On4cNG2QJqhBCFEKJn2NY8etldgVEMb1LfV5vU+vBE717w2+/wYABalmLqlUNF6QQQpQgJToxrD8axpd/XGVoayfe6Fof0tPVDWoWFjB+vPpn4EBDhymEECVKiR1K2hd0k0U/BdOjiQOL+zXF7PhxcHN7UPRu4EBJCkII8RRKZGI4euUOM3cG0qpOJT7t2wCLN6arh+akp4OLlnpIQgghClTihpLORyUyYfNpnO3t+MY5Axs3V3VvwtSpsGQJ2BXyRDYhhBCPVaISw/U7qYzaeJIKtlZsGt2K8mf9wdZWrYrarp2hwxNCCKNQYhJDTHI6w9efJCc9g2+tL+NQrjN07KguRZU9CUIIUWR0Mseg0Wjw8fHB09MTb29vwsPD8zy/c+dOBgwYwODBg/n999+1Xi9HozByzXFi7ySyfv1MnPdtf1D0TpKCEEIUKZ0kBj8/PzIzM9mxYwczZ85k6dKluc/FxsayefNmtm/fzrp16/j444/JvP9LPh+349MIvZ3M6r1LaT5tFBw/LkXvhBBCR3QylBQQEED79u0BcHNz48KFC7nPnTt3jubNm2NlZYWVlRVOTk5cunQJV1fXfK+Xqliw/OrPvLxvEzRsqIuQhRBC/EsniSElJQW7h1YHWVhYkJ2dTalSpUhJSaFs2bK5z5UpU4aUlJQCr9fQ3gbr96cTotGoh+qYuBBpg1zSFg9IWzwgbaHKyMh4qvfpJDHY2dmRnqKNxgAACw5JREFUmpqae1+j0VDq3wNw/vtcampqnkTxOG5ubroIUwghxGPoZI7B3d2dI0eOABAYGEiDBg1yn3N1dSUgIICMjAySk5O5evVqnueFEEIYlpmiKEpRX1Sj0bBw4UJCQ0NRFIUlS5Zw5MgRnJyc6NKlCzt37mTHjh0oisKECRPo0aNHUYcghBDiKekkMQghhCi5SmStJCGEELojiUEIIUQekhiEEELkUawSQ1GX0iiptLXDxo0b8fDwwMPDg88//9xAUeqHtra4/5qxY8eybds2A0SoP9ra4s8//2Tw4MEMHjyYhQsXYszTh9raYt26dQwYMICBAwdy8OBBA0WpX0FBQXh7ez/y+OHDhxk4cCCenp7s3LmzcBdTipFff/1VeeuttxRFUZSzZ88qEydOzH0uJiZG6dOnj5KRkaEkJSXl3jZGBbVDRESE0r9/fyU7O1vJyclRPD09lZCQEEOFqnMFtcV9H330kTJo0CBl69at+g5Prwpqi+TkZKV3795KXFycoiiKsmbNmtzbxqigtkhMTFQ6duyoZGRkKAkJCcrLL79sqDD1Zs2aNUqfPn0UDw+PPI9nZmYqXbt2VRISEpSMjAxlwIABSkxMjNbrFaseQ2FLaZQtWza3lIYxKqgdqlatytq1a7GwsMDc3Jzs7Gysra0NFarOFdQWAL/88gtmZmZ06NDBEOHpVUFtcfbsWRo0aMCyZcsYOnQoVapUoVKlSoYKVecKaovSpUtTrVo10tLSSEtLw8zMzFBh6o2TkxOrVq165PGrV6/i5ORE+fLlsbKyokWLFpw+fVrr9YpV2e2iLqVRUhXUDpaWllSqVAlFUVi+fDmNGzemTp06BoxWtwpqi9DQUH766Sc+++wzvrh/pKsRK6gt4uPj8ff3Z8+ePdja2jJs2DDc3NyM9mejoLYAcHR0pHfv3uTk5DBhwgRDhak3PXr0ICoq6pHHn/b3ZrFKDEVdSqOkKqgdQK1/8vbbb1OmTBkWLFhgiBD1pqC22LNnD9HR0YwYMYIbN25gaWlJ9erVjbb3UFBbVKhQgWbNmmFvbw9Ay5YtCQkJMdrEUFBbHDlyhJiYGA4dOgTAmDFjcHd3L7BQp7F62t+bxWooSUppqApqB0VRmDx5Mg0bNmTRokVYGPl5FAW1xZw5c9i1axebN2+mf//+jBw50miTAhTcFk2bNiU0NJS7d++SnZ1NUFAQ9erVM1SoOldQW5QvXx4bGxusrKywtrambNmyJCUlGSpUg3J2diY8PJyEhAQyMzM5ffo0zZs31/q+YtVj6NatG8eOHcPLyyu3lMaGDRtyS2l4e3szdOhQFEVhxowZRju2XlA7aDQaTp48SWZmJn/99RcAb775ZqH+Z5dE2n4mTIm2tpg5cyZjx44FoGfPnkb7xQm0t8Xx48cZPHgw5ubmuLu7087Ejv719fXl3r17eHp6MnfuXMaMGYOiKAwcOBAHBwet75eSGEIIIfIoVkNJQgghDE8SgxBCiDwkMQghhMhDEoMQQog8JDEI8f/t3V1IVPsax/GvpmKS1CayFKN0SiWjkiSoC6PIi4RuKsn3rEh8KVNDLVPT1EhMIa3EuVByxEhMwvImwYoojDQqo9LJnBqSTEUlbUZHfc5FZw8M0tl7H3btc+r/uZz5v61Zwzyz1pr1G0VRbKjCoPxQExMTxMTEsG/fPmukSUdHB1qt9rvM19raysDAwHcZ+0e6du0aFovlL/Xp7++nra3tT7f/1X7SqXybKgzKD/XgwQO2b9/O6dOnaWxsRESora1l//7932W+2tranyI6paqqitnZ2b/Up729nSdPnnynFSk/s/+pG9yUn5+Liwsmk4kvX77g4uLCzZs3CQ4O/ubNis+ePaOoqAgRYenSpZw/f57Dhw+Tl5eHRqPh6tWrDA0NERcXx7FjxxgfH8dsNpOeno7JZOLVq1dkZmZSX19PXV0dLS0tODg4EBgYSHp6OhUVFbx7946RkRHGxsaIiIjg9u3b9PX1UVxczIYNG9DpdNy6dQs7OztCQkKIiYnhxIkTjI6OMjo6yuXLl0lJSUFEsFgs5Ofn4+vra90Gs9nMyZMn6e/vx2KxkJOTw9q1a8nKysJoNDIzM8OBAwcICQkhOjoaPz8/9Ho94+PjXLhwgYcPHzI4OEhqaioVFRXk5uby8eNHRkZGCAoKIiUlBYPBQHZ2NhaLBWdnZ0pLS9FqtZjNZgICAvD09KSwsBD4Gp9x9uxZXFxcyMnJ4c2bNyxfvpypqakf8h5Q/g/8fcGvivLHZmZmpKCgQNLS0sRoNMrRo0elp6dHcnJyRKvVzmm/a9cuefPmjYiI1NXVyYsXLyQqKsr6WH19vZSXl0tPT4/s2bNHPn/+LAaDQe7evSsiYm37+vVr2bt3r0xNTcns7KwkJSVJW1ublJeXy6lTp0REpKqqSpKTk0VEpLGxUQoLC0Wv10tYWJg15jw6Olp6e3slMzNTampqRETkzp07kpiYKCaTSbq6uqSjo8NmG2pqaqSkpERERLq7u6WmpkZ0Op0UFRWJyNfI7ODgYBkeHpaoqChpbm4WEZGysjKpqqoSEZFt27aJ2WwWo9EoDQ0NIiJiNptl06ZNIiISHx8v9+7dExGRlpYWuX//vly/ft06b2hoqOj1ehERaWhokLKyMmlra5O0tDQREfnw4YP4+/v/l3tV+dmoIwblh7K3tyc7Oxv4enokJiaGyspKsrOzuXjxIn19fTbBb8PDw2g0GgAiIyPnjCf/vnF/9erVREZGkpaWxvT09Jw/LHn79i3r16/H0dER+Boyp9frAVizZg0Arq6u1nyhhQsXMjk5SU9PD/39/cTGxgIwNjbG+/fvAazrDAoKwmAwkJiYiIODAwkJCXPm/j3DycfHBx8fH/Lz89myZQvwNehMo9FgNBpt1rNs2TKGhoZsxlq0aBFdXV20t7ezYMEC67f8vr4+ayxKSEgIAE1NTdZ+vb295OfnA2CxWPDy8kKv11uD5Tw8PHB3d5/z+iq/JnWNQflHDA8PYzAYCAwMxGQyMW/ePOzs7DCZTDbt3NzcMBgMAGi1WlpbW3FycmJwcBCAly9fAtDd3c3ExARarZZz585RUFAAgJ2dHSKCt7c3z58/Z3p6GhHh8ePH1g/2/5TX7+3tzapVq6itrUWn07F7925rBtHv/R49eoSbmxvV1dUkJCRQVlZmM4ZGo6GrqwsAo9HI8ePH0Wg01lz88fFxenp68PT0/OY67OzsmJ2dpampCVdXV0pLSzl48CBmsxkRsZmjubkZnU6Hvb299bqEl5cXxcXF6HQ60tPT2bp1K97e3jx9+hSAgYGBn+IivfL3UEcMyj+isrKS+Ph4ACIiIjh06BAeHh74+fnZtMvPzycrKwt7e3uWLFlCbGwsTk5OnDlzBnd3d9zc3ABYuXIlly5d4saNGzg6OpKcnAxAQEAAGRkZVFdXs3PnTsLDw5mdnWXjxo3s2LHjD//syc/Pj82bNxMeHs7U1BTr1q2bE0Lm5+dHamoqV65cwd7enqSkJJvnw8LCyMrKIioqipmZGbKysvD19SUnJ4fw8HAmJyc5cuQIixcv/uY6AgMDiYuLIzc3l7S0NDo7O5k/fz4rVqzg06dPZGRkkJubS2VlJc7OzpSUlNDf309lZSX+/v7k5eWRmZnJzMwMAEVFRXh5edHZ2UloaCgeHh789ttvf2LPKb8CFaKnKIqi2FCnkhRFURQbqjAoiqIoNlRhUBRFUWyowqAoiqLYUIVBURRFsaEKg6IoimJDFQZFURTFxr8A5v320p2D/AAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gradient Boosting:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print('Logistic:')\n",
    "gain_plot(logit_report)\n",
    "print('Gradient Boosting:')\n",
    "gain_plot(gb_report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:12.384682Z",
     "start_time": "2020-03-04T12:40:07.285688Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gradient Boosting:\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAESCAYAAAD5d3KwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3dd1gU19vG8e/SFhBEUMSCqGDD3qJijb1iw4IFNdEYjcZC7FFjEkuMsWJi1CTWWLGisUUTe0MFG4qKKFgQEAu9zfuHyb4/gmJjd1l4PtflFZize+beycLDOTtzRqUoioIQQgjxDyN9BxBCCJGzSGEQQgiRgRQGIYQQGUhhEEIIkYEUBiGEEBlIYRBCCJGBFAahc2lpaaxYsYKuXbvSqVMn2rVrx5w5c0hOTtb6vsPDw6lRo8ZrH7d582Z+//13ANavX8+yZcuyZf9eXl7s3buXiIgIPD09AYiNjcXT05P27duzf/9+Jk+ezOXLl9+4z61bt/Lpp5++d7a33a/IvaQwCJ2bNm0aFy5cYNWqVezYsQNfX19u377Nl19+qe9oGufOnSMxMRGAXr16MXjw4Gzt38HBgQ0bNgAQFBREdHQ0u3fvplWrVpw4cQJ9XF6kr/2KnMdE3wFE3hIeHo6fnx/Hjh3DysoKAEtLS77++mvOnz8PwIQJEyhbtiwDBw7M9H2zZs3o0KEDp06d4unTpwwaNIjz589z5coVTExMWLJkCQ4ODjRr1oyFCxdSpUoVAM33tra2mixRUVFMnTqV6OhoIiMjKV68OAsWLOD8+fMcOnSI48ePY25uzuPHj4mJiaFZs2bMnj0bPz8/AJ49e0bz5s35888/SUxM5JtvvuHBgwekpKTQvn17hgwZkuVxcHd3Z8uWLUyaNImIiAg6deqEm5sbjx49YsyYMXz//fdUq1Ytw/OWLl3Ktm3bMDExoWTJknz33XcAREZGMnjwYB48eICxsTFz587FxcWFgIAAzWgsMjKS+vXrM3PmTMLDw+nTpw8uLi7cu3ePWrVqZblfkbfIiEHo1JUrVyhTpoymKPzL3t6e1q1bv1EfSUlJbNq0iZEjRzJ16lT69+/Pzp07KVq0KNu2bXvjLLt376Z69eps3LiRgwcPYm5uzo4dO2jZsiXNmjVjwIAB9OnTR/P4Bg0aEBcXx6VLlwDYtWsXTZo0wcbGhrFjx+Lh4cHWrVvx9fXlxIkT/PHHH6/N4OzszPTp03FycmLHjh1MmDCBwoUL88MPP2T65Xzw4EG2bt3Kxo0b2bVrF46OjqxduxaAsLAwvvzyS/z8/Khduza//vorAKtXr2bEiBFs3ryZ3bt3c+jQIc100cOHD/nss8/Yt28fM2fOfOV+Rd4jIwahU0ZGRqSnp79XH61atQKgRIkSFCpUiAoVKgDg5OTE06dP37if/v374+/vz4oVKwgNDeXGjRtZ/lJUqVR4eHiwbds2qlSpwtatWxk3bhzx8fGcPXuWp0+fsnDhQgDi4+O5du0a7dq1e49XmtHJkydp06YNNjY2AEycOBF48RlD1apVKVmyJACurq4cOHAAgO+++44jR47w888/ExISQlJSEvHx8RQoUAATExOqV6+ebflE7iGFQehU1apVCQkJITY2NsOoISIigilTprBo0SJUKlWGue6UlJQMfZiZmWm+NjU1feW+/rePl32wPWfOHC5evIiHhwd169YlNTX1tXPs3bp1o0uXLnTv3p3nz59Tp04dYmNjURSFDRs2YGFhAcDjx49Rq9VZ9vW2jI2NUalUmu+fPXvGs2fPADAx+f8f5f89fn379qV8+fI0atSItm3bEhgYqGkzMzPL8Dwh/iVTSUKnHBwccHd3Z9KkScTGxgIvzsqZNm0aBQoUwNzcHFtbW810R0REBGfOnHnr/djZ2Wn6OH36NJGRkZkec+zYMfr370/nzp0pWLAgJ06cIC0tDXjxSzg1NfWl+atWrcrUqVPp1q0bAFZWVlSvXp0VK1YAL35h9+rVi4MHD7517qz2Xb9+fQ4cOKA5bj4+PqxcufKV/Tx79oxLly4xZswYWrVqxcOHD7l79+4rR2yv2q/Ie6QwCJ376quvKFOmDJ6ennTq1Inu3btTpkwZpk+fDrw4pTMyMpLWrVszadIk6tWr99b7GDNmDKtXr6ZTp07s2LGDSpUqZXrMsGHD+P7773F3d2fo0KHUrFmTu3fvAtC4cWM2bNjA0qVLMz2ve/fuBAUF0aVLF822H374gcDAQNzd3enevTsdOnSgY8eOb50boGXLlowdO5Zjx45l2N6kSRO6du1Kr169cHd3JyoqilGjRr2yn/z58zN48GC6dOlChw4dWLZsGTVr1uTOnTtvtV+R96hk2W0hhBD/S0YMQgghMpDCIIQQIgMpDEIIITKQwiCEECIDgziJOSAgINvPCf9XUlKS1vrWBkPLC4aX2dDygmTWBUPLCy8yv8tFjAZRGNRqNa6urlrpOygoSGt9a4Oh5QXDy2xoeUEy64Kh5YUXmd+FTCUJIYTIQAqDEEKIDKQwCCGEyMAgPmMQQuRdKSkphIeHa26cpM8c7zpnr23m5uY4Ojpmuajk25DCIITI0cLDw7G2tqZUqVIZVpfVtYSEBM3quTmJoihER0cTHh5O6dKls6VPmUoSQuRoiYmJFCxYUK9FISdTqVQULFgwW0dUUhiEEDmeFIWsZffxMYjC8CQ+mdS097vrlxBCiDdjEIXh7uMEPJacIDjiub6jCCHymNOnT+Pm5sbAgQPp27cvnp6e3Lp167379fHxYf369QQFBbF48eJsSJp9DKIwONlZEhaTQPtFR/E5eIMUGT0IIXSoXr16/Prrr6xdu5bhw4fz/fffZ1vfrq6uDB8+PNv6yw4GcVZSAUtTDoxuzFc7rzD3QDB7Lj9kTveqVCpmo+9oQghd+/DDzNt69IDPPoP4eGjXLnP7gAEv/kVFwT+3ZNX4+++32v2zZ88oXrw4Z86c0fyln5iYyOzZsylWrBgjR44kNjaWxMRExo4dS926ddmzZw8rV67EyMiIWrVqMWbMGE1/p0+fZsOGDcyfP59WrVpRs2ZNbt++TcGCBfHx8SE9PZ2vvvqKO3fukJ6ezqhRo6hbt+5bZX5bBlEYAApaqVncuyYdqj5k8vbLdFp8nM8+dGFYszKoTYz1HU8IkYudOnWKgQMHkpaWxvXr11m6dCk3btxgzpw5ODg48PPPP7N3715atGhBVFQUK1euJDo6mtDQUJ48eYKPjw9btmzBwsKCsWPHcvz48ZfuJywsjFWrVlG0aFE8PT25dOkSV69exdbWlpkzZxITE0Pfvn3ZvXu3Vl+vwRSGf7WpXIR6znZ843eVRYdusvfKQ+Z0q0a1EgX0HU0IoQtZ/YVvaZl1e6FCbz1CgBdTSTNnzsTCwoKQkBA8PT2ZOXMmM2bMwNLSkoiICGrWrEnZsmXp06cP3t7epKam4uXlxd27d3n8+DGDBw8GIC4ujrCwsJfux9bWlqJFiwJQtGhRkpKSCA4O5ty5c1y8eBGA1NRUYmJisLW1fevX8aa0VhgCAwP54YcfWLNmjWZbZGQk3t7emu+DgoL44osv6NWr11v1XcDSjHk9q9OhWlEmbb1Ml5+O80ljZ0a3KIe5qYwehBDaU6hQIQAmT57Mn3/+iZWVFePHj0dRFK5fv05cXBzLli3j0aNHeHp64uvrS9GiRfntt98wNTVl69atuLq68ueff2bq+2WnnTo7O1OkSBGGDBlCYmIiS5YswcZGu9PoWikMy5cvZ+fOnZmuErS3t9cUigsXLjB//nx69OjxzvtpVsGB/d52zNwdxNLDIRy4GsGcblWpVdLuvfILIcT/+ncqydTUlLi4OCZMmMD169fp0aMH+fPnp1ChQjx69IhSpUrx448/sn37dkxNTRkxYgR2dnYMGDAALy8v0tLSKF68OG3btn3jfXt6ejJ58mT69u1LbGwsvXv3xshIu+cNqRRFUbK703379lG+fHnGjRvHpk2bMrUrioKHhwc//PADzs7Or+3vTW7Uc/5+PAtPRBEZl0pH1/wMqGGHuenrD15iYiLm5uavfVxOYWh5wfAyG1peyN2ZU1JSKFu2rA4SZU1RlBx9od2NGzdeulbSu9xDQisjhtatWxMeHv7K9kOHDlG2bNk3KgrwZjfqcXWFTg1T+X7vNVafvMOFiFS+86hCfZdCWT7P0G6+YWh5wfAyG1peyN2Zg4KCcsQaRTl1raR/mZqaZjqeBnWjnp07d77XFNKrWKlN+KZTZTYMrodKBb2Xn+bLbZeITUrN9n0JIURupZfCcOXKFWrWrKm1/us5F2TvyMYMbFiadWfu0nr+EQ4HR2ptf0IIkZvopDD4+fmxceNGAB4/fky+fPm0PldnYWbMlA4V8R1SH3NTI/r/doZxvoE8TUjR6n6FEMLQae10VUdHR80Hz+7u7prtdnZ27NixQ1u7zaRWSVt2j2jEwoM3WHr4FoeDI5nZpQrNXR10lkEIIQyJQayV9L7MTY0Z36YC24c1oICFGQNX+TNqwwVi4pL1HU0IIXKcPFEY/lXVsQB+nzdkRPOy7Lr4gJbzD3PsTqy+YwkhcrDTp08zevToDNtGjx5NcnIyYWFhdOrUifHjx3P9+nXOnj2rp5TZK08VBgAzEyO8W5Zj5/CGOOQ3Z8bfj/js93NExSbpO5oQwkDMnz8fMzMzzp8/j5ubG7Nnz2b//v3cvHlT39GyhcGtlZRdKhbLz/ZhDZjue4r1Fx9x8tZhpnWsRMdqxXL0RSxC5GVbzoWzyf/l6wy9qx61S+BRy/GtntOsWTPWrl3LkiVLSExMxMbGhm3btmFqakqlSpWoWrVqtmbUtTxbGABMjY3wrGpLnw+rMNb3IiM3BOAX+IAZXSrjkN+wriIVQuhWwYIFGTx4MCEhIQwdOpTU1FQKFSpk8EUB8nhh+Fc5B2u2Dq3Pr8dCmLs/mJbzDjOlQ0W61XKU0YMQOYhHLce3/utevL089xnDqxgbqRjc2IU9IxtRvog1Y30vMmDFWe49SdB3NCGEAVCpVKSn5467S8qI4T+c7a3YONiN1SdDmb33Oq3nH2Fiuwr0ruMkowch8qjjx4/Tu3dvze+A5OTMp7pXrlyZ77//HhcXF+rVq6friNlKCsNLGBmpGNCgNM0qODBh60W+3HaZ3Rcf8F3XqjgVtNR3PCGEDtWtW5czZ868dBG9rl27ar7+8MMP+fBltx01QDKVlAWngpb8PqguM7tU4WL4U1ovOMLK47dJT8/2lcqFECLHkMLwGiqVit51ndg3ujEflLZjmt9Vei47SUikXBgnhMidpDC8oeIFLFj10QfM6VaV6w+f03bhUZYduUWajB6E0Dot3E8sV8nu4yOF4S2oVCq61y7BAe8mNCprz8w/rtF1yQluRDzXdzQhci1zc3Oio6OlOLyCoihER0dn6x385MPnd+CQ35zl/WqxM/A+03Zeof2iY4xsUZbBjZ0xNZZaK0R2cnR0JDw8nMhI/d5TJSUl5aW3zswJzM3NcXTMvus7pDC8I5VKRafqxanvUohpO68wZ991/rj0gDndqlGxWH59xxMi1zA1NaV06dL6jmGQt099V/Ln7Xuyt1bzY5+aLOlTk4hniXRcfIx5B4JJTs0dF7oIIfIeKQzZpG2VohwY3QT3asVYdPAG7j7HuBj+RN+xhBDirUlhyEa2+cyY37M6v/avzZOEZDr/eJzv9lwjMSVN39GEEOKNSWHQguauDuwf3YTutUrw8+FbtFt0lHN3Hus7lhBCvBGtFYbAwEC8vLwybb948SK9e/emV69ejBgxgqSk3HmDHBsLU2Z3q8rqj+uQlJJOt59P8u2uqyQky+hBCJGzaaUwLF++nMmTJ2f6pa8oClOmTGHWrFmsX7+eRo0ace/ePW1EyDEal7Nn3+jG9KnrxK/HbtNm4RFOhUTrO5YQQrySStHCVSP79u2jfPnyjBs3jk2bNmm2h4SE8PXXX+Pi4kJwcDBNmjThk08+eW1/AQEBqNXq7I4JQGJiYrZeGJKViw8TmH88koexqXQon5+Patlhafp2tVmXebOLoWU2tLwgmXXB0PL+611OsdXKdQytW7cmPDw80/aYmBguXLjAlClTKFmyJEOGDKFy5cq4ubll2Z9ardba+cO6PDfZ1RXcG6Tyw75gVpy4zYWIFL7zqEKjsvZv3IchnkttaJkNLS9IZl0wtLzwIvO70OmHzwUKFKBkyZKUKVMGU1NTGjVqxOXLl3UZQe8szUyY6l6RzZ+6oTYxwuvXM0zYcpFniSn6jiaEEICOC0OJEiWIi4vjzp07APj7+1O2bFldRsgxapey44+Rjfi0iTOb/MNoNe8Ih65F6DuWEELopjD4+fmxceNGzMzMmDFjBl988QUeHh4UKVIk19zY4l2Ymxozsa0r2z5rQH4LEz5e6Y/3xgCexGe+O5QQQuiK1tZKcnR01Hzw7O7urtnu5uaGr6+vtnZrkKqVKIDf5w358dBNfvr7FkduRDG9c2XaVC6i72hCiDxILnDLIdQmxni3Ks+O4Q0obK1myNpzDFt3nujY3HmdhxAi55LCkMNUKmbDjuENGNOqHPuvPKTl/CPsDLwva9ELIXRGCkMOZGpsxPBmZdk9ohElbC0Ysf4Cn645x6NnifqOJoTIA6Qw5GDlHKzZMrQ+E9tW4O/gSFrOP8Kft57L6EEIoVVSGHI4E2MjPm3iwp6RjShb2Iq5xyL5eOVZHjxN0Hc0IUQuJYXBQLjYW7HxUzeG1CnIqZDHtJp3hPVn7sroQQiR7aQwGBBjIxWdXG3YO6oRlYrnZ+LWS/T99TRhj+P1HU0IkYtIYTBAJQvmY92gekzvXJmAu09oveAIq0+Gkp4uowchxPuTwmCgjIxU9K1Xkv3eTahV0papO67gufwUoVFx+o4mhDBwUhgMXPECFqz+uA7fd6tK0INntFl4hF+OhpAmowchxDuSwpALqFQqetQuwYHRTWjgUojpu4Po9vMJbj56ru9oQggDJIUhFyliY84v/WuzoGd1bkfF0W7hMX786yapaen6jiaEMCBSGHIZlUpF5xrFOTC6Cc1dCzNn33W6/HSCoAfP9B1NCGEgpDDkUvbWapb0rcVPfWpy/0kCHRcfY8GfwSSnyuhBCJE1KQy5XLsqRTng3YR2VYqy4M8bdFx8jEvhT/UdSwiRg0lhyAPs8pmx0LMGy/vV5nFcMp1/Os73e6+RmJKm72hCiBxICkMe0rKiAwdGN6FrjeL89PctOvgc4/zdGH3HEkLkMFIY8hgbS1PmdK/Gyo8+ID4plW5LTjBj91USkmX0IIR4QQpDHvVh+cLsG90YzzpOLD96m7YLj3Dm9mN9xxJC5ABaKwyBgYF4eXll2r5ixQrat2+Pl5cXXl5ehISEaCuCeA1rc1NmdqnCukF1SVMUeiw9yVc7LhOXlKrvaEIIPTLRRqfLly9n586dWFhYZGq7cuUKs2fPpnLlytrYtXgH9csUYu/IxszZd51VJ0M5eO0Rsz2q0qBMIX1HE0LogVZGDE5OTvj4+Ly07cqVKyxbtoxevXqxdOlSbexevIN8ahOmdazEpk/dMDU2os8vp5m49RLPElP0HU0IoWMqRUt3egkPD8fb25tNmzZl2L548WJ69+6NlZUVw4cPp1evXjRt2jTLvgICAlCr1dqISWJiIubm5lrpWxt0kTcpNZ01ATFsu/oUOwtjRrjZ84Gj5Tv3J8dY+ySz9hla3n+5urq+9XO0MpX0Koqi0L9/f6ytrQFo0qQJV69efW1hUKvV7/Ti3kRQUJDW+tYGXeWtXgX63o1hnO9Fph58iEdNR6Z2qIiNpelb9yXHWPsks/YZWl54kfld6PSspNjYWDp06EBcXByKonD69Gn5rCEHq+Fky64RDRnetAzbA+7RYv5h9l95qO9YQggt00lh8PPzY+PGjVhbWzN69Gj69etH7969KVOmDE2aNNFFBPGO1CbGjGldnh3DGlDISs3gNef4fP0FHscl6zuaEEJLtDaV5OjoqPl8wd3dXbO9c+fOdO7cWVu7FVpSubgNO4Y14OfDt/A5dIMTN6P4ulMl2lcpikql0nc8IUQ2kgvcxBszMzFiRPOy+H3ekOK2Fgxfd4Gha8/z6HmivqMJIbKRFAbx1ioUyc/WofUZ36YCh64/otX8I2y7EI6WTnATQuiYFAbxTkyMjRj6oQt/jGiEc6F8jN4YyKBV/jx8KqMHIQydFAbxXsoUtmLzkPpM6VCR47eiaDnvMBvP3pXRgxAGTAqDeG/GRioGNizN3pGNqVgsP+O3XKLfb2cIj4nXdzQhxDuQwiCyTalC+Vj/ST2+7VSJc3diaD3/CGtO3SFdRg9CGBQpDCJbGRmp8HIrxb5RjalZ0pYp2y8zaf8DGT0IYUCkMAitKGFnyeqP6/Bd1yoERyXRZsFRNp0Nk88ehDAAUhiE1qhUKjzrOLGkkyOViuVn3JaLDFrlL9c9CJHDSWEQWudgZcr6T+oxpUNFjt2MotX8I+y6eF/fsYQQryCFQeiE0T9nLu0e0YiSdpYMX3eBz9df4Em8rLkkRE4jhUHoVJnCVmwZWp8vWpZjz6UHtJp/hL+uPdJ3LCHE/5DCIHTOxNiIz5uXZfuwBthamvHRyrNM3HqRWLnXtBA5ghQGoTeVi9uw8/MGDGniwsazYbRZcIRTIdH6jiVEnieFQeiV2sSYCW0rsHmIG8ZGKnotP8W3u66SmJKm72hC5FlSGESOUKukHXtGNqJv3ZL8euw27RcdJTDsib5jCZEnSWEQOYalmQnfdq7MmoF1iE9Oo+uSE8zdf53k1HR9RxMiT5HCIHKcRmXt2TuqMZ2qF8Pn0E26/HSc6w+f6zuWEHmGFAaRI9lYmDKvR3WWetXi4dNE3H2O8fPhW6Sly5IaQmib1gpDYGAgXl5er2yfMmUKP/zwg7Z2L3KJ1pWKsG90Y5pWsOe7PdfosfQkoVFx+o4lRK6mlcKwfPlyJk+eTFJS0kvbN2zYQHBwsDZ2LXKhQlZqfu5bi/k9qxEc8Zy2C4+y5mSoLMgnhJZopTA4OTnh4+Pz0rYLFy4QGBhIz549tbFrkUupVCq61HBk/+jG1C5ly5QdV+j32xnuP0nQdzQhch2VoqU/u8LDw/H29mbTpk2abY8ePWLixIksXryYPXv2EBISwpgxY17bV0BAAGq1WhsxSUxMxNzcXCt9a4Oh5YXsz6woCn8EP+cX/2iMVSqG1C1Ic2crVCpVtvQvx1g3DC2zoeX9l6ur61s/x0QLOV5p7969xMTEMHjwYCIjI0lMTMTZ2ZmuXbtm+Ty1Wv1OL+5NBAUFaa1vbTC0vKCdzBUrQvdGcYzZHMjcY5FcemzEzK5VKGT1/n9AyDHWDUPLbGh54UXmd6HTwtCvXz/69esHwNatWwkJCXltURDiVUoWzMeGwW78eiyEH/YF03r+EWZ0qUKbykX0HU0Ig6aT01X9/PzYuHGjLnYl8hhjIxWDG7uwa0RDihYwZ8jac4zeGMDThBR9RxPCYGltxODo6Kj5fMHd3T1Tu4wURHYq52DNts8asPjQTRb/dZOTt6L5vltVGpez13c0IQyOXOAmcg1TYyNGtyzHts/qY2VuQr/fzjB5+yXiZDlvId7KGxWGS5cuZfj+zJkzWgkjRHao6liAXZ83ZFDD0vx++i7tFh3lbOhjfccSwmBkOZXk7+/PzZs3WblyJR999BEAaWlprFu3jl27dukkoBDvwtzUmMkdKtKyogNjfAPpsfQkgxs5M7plOcxNjfUdT4gcLcsRQ/78+YmKiiI5OZnIyEgiIyOJiYlh7NixusonxHup61yQPSMb4/mBE0uPhNBx8TEu33uq71hC5GhZjhhWrFjBrFmzMDExYciQIbrKJES2slKbMKtrFVpVcmC870U6/3icz5uV5bOmLpgay8dsQvxXloUhKCiI2bNns2/fPuLj4zO0eXt7azWYENmtafnC7B/dmK92XmH+n8EcvBbBvB7VKFPYWt/RhMhRsvxzafHixbi6uqJWqyldunSGf0IYogKWZiz0rMFPfWoS9jiedouO8cvRENJlOW8hNLIcMRQpUoSOHTtSp04d7OzsdJVJCK1rV6UotUvZMmnrJabvDmL/1Qjmdq9GCTtLfUcTQu+yLAzjx49n7ty59O7dW7NAmaIopKWlcfjwYZ0EFEJbClubs7xfbXzPhfON31XaLDjC5A4VqWYloweRt2VZGObOnQvAoUOHMmz38PDQXiIhdEilUtG9dgnqlynE2M2BTNx6idrFLfjR0RmH/Ia3kqYQ2eGdTsnIruWNhcgpihewYO3AunzdsRKXHibSav4RdgTck5sBiTxJztUT4h9GRir61y/FYvfiONvnY+SGAIatO8/juGR9RxNCp7KcSvL29s40OlAUhbCwMK2GEkKfHG3M2PypG0uPhLDgz2DO3I7hu65VaFHRQd/RhNCJLAuDp6fnW20XIrcwMTZiWNMyNKtQmNEbAxi02p9utRyZ6l6R/Oam+o4nhFZlWRjq1KmjqxxC5EiuRfOzc3hDFh4MZsnftzh5K5o53apSv0whfUcTQmvkMwYhXsPMxIixrSvgO7Q+ahMjev9ymmk7r5CQnKbvaEJohRQGId5QTSdbdo9oxID6pVh5IpR2i45y/m6MvmMJke2kMAjxFizMjJnWsRLrBtUlOTWdbktO8P3eaySlyuhB5B5SGIR4B/XLFGLvqEZ0q+XIT3/fotPi41y9/0zfsYTIFlorDIGBgXh5eWXavm/fPjw8POjWrRubN2/W1u6F0Dprc1O+71aNX/rVJio2mU4/HuPHv26Smpau72hCvJcsz0p6V8uXL2fnzp1YWFhk2J6WlsbcuXPZsmULlpaWtGvXjubNm8sCfcKgtajowIGStkzecZk5+67zZ9CLBfmc7a30HU2Id6KVEYOTkxM+Pj6ZthsbG/PHH39gbW3NkydPAMiXL582IgihU7b5zPixd00W9apBSGQc7RYdZeXx27KctzBIKkVLi8GEh4fj7e3Npk2bMrXt37+fb775hiZNmvDNN99gbJz1PXgDAgJQq9XaiEliYiLm5oazWJqh5QXDy/y+eaPjU1lwIhL/ewlUK2LO6Ab2OFhp9/jiL48AAB8mSURBVKI4QzvGYHiZDS3vv1xdXd/6OVqZSnqdVq1a0aJFCyZMmMD27dtfu1qrWq1+pxf3JoKCgrTWtzYYWl4wvMzZkbdBTYWNZ8P4dtdVhu96wFT3inSv5ai1BSgN7RiD4WU2tLzwIvO70OlZSbGxsfTt25fk5GSMjIywsLDAyEhOjBK5j0qlwrOOE3tHNaZSsfyM873IJ6v9efQ8Ud/RhHgtnfxW9vPzY+PGjVhZWeHu7k6fPn3o1asXKpWKjh076iKCEHpRws6S9Z/UY3J7V47ciKL1/CPsvvhA37GEyJLWppIcHR01ny+4u7trtvfs2ZOePXtqa7dC5DhGRioGNXLmw/L2fLEpkGHrzrPvSjG+6VSJApZm+o4nRCYyjyOEjpQpbM2WofXxblmOPy49oNX8I/x1/ZG+YwmRiRQGIXTIxNiIEc3Lsn1YAwpYmvLRirNM3HqR2KRUfUcTQkMKgxB6ULm4DX6fN+TTJs5sOBtGmwVHOBUSre9YQgBSGITQG7WJMRPburL5UzeMjVT0Wn6Kb3ddJTFFFuQT+iWFQQg9q13Kjj9GNKJv3ZL8euw27RcdJTDsib5jiTxMCoMQOUA+tQnfdq7M6o/rEJeURtclJ5i3/zrJqbIgn9A9KQxC5CCNy9mzb3RjOlUrxqJDN+ny03GuP3yu71gij5HCIEQOY2Nhyrye1fm5by0ePk3E3ecYPx++RZosyCd0RAqDEDlUm8pF2De6MU0r2PPdnmv0XHqS0Kg4fccSeYAUBiFysEJWan7uW4t5PapxPeI5bRceZc2pO2hpUWQhACkMQuR4KpWKrjUd2T+6MbVL2TJl+2X6/XaGB08T9B1N5FJSGIQwEEVtLFj9cR2+7VwZ/9AYWs0/wtbz4TJ6ENlOCoMQBkSlUuFVryR7RzWivIM13psCGbL2HE8S5KI4kX2kMAhhgEoWzMfGT92Y2LYCf12LZMjOMPZefqjvWCKXkMIghIEyNlLxaRMXdo1oiL2lCUPWnsN7YwBPE1L0HU0YOCkMQhi4cg7WzG9fnBHNy7Ij8D5tFhzh6I1IfccSBkwKgxC5gImRCu+W5dg6tD6WZsZ4/XqGydsvEZ8sy3mLtyeFQYhcpFqJAuwe0YhBDUvz++m7tF14FP/Qx/qOJQyMFAYhchlzU2Mmd6jI+k/qkZau0H3pSWbtCZLlvMUbk8IgRC5Vz7kge0c1xvMDJ5YeDqHj4mNcvvdU37GEAdBaYQgMDMTLyyvT9l27dtG9e3c8PT2ZOnUq6emyrLAQ2mKlNmFW1yqs+OgDnsSn0PnH4yw6eIPUNPm5E6+mlcKwfPlyJk+eTFJSUobtiYmJLFiwgNWrV7NhwwZiY2P566+/tBFBCPE/mpYvzP7RjWlXpSjzDgTjseQENx/Jct7i5VSKFq6n37dvH+XLl2fcuHFs2rRJsz09PZ3Hjx9TqFAhAEaMGEGPHj1o2LBhlv0FBASgVquzOybwoliZm5trpW9tMLS8YHiZDS0vvF3mo6GxLD4VRWKqwoAatnSqaIORSqXlhJkZ2nE2tLz/cnV1fevnmGghB61btyY8PDzTdiMjI01RWLNmDfHx8TRo0OC1/anV6nd6cW8iKChIa31rg6HlBcPLbGh54e0yu7pC54aJTNp6iWX+jwiIhrndq1HCzlLLKTMytONsaHnhReZ3ofMPn9PT05k9ezbHjx/Hx8cHlR7+UhEirytsbc7yfrX5vltVrt5/RpsFR1h/5q4syCcAPRSGqVOnkpSUxE8//YSFhYWudy+E+IdKpaJH7RLsHdWIqo4FmLj1Eh+tPEvEs0R9RxN6ppPC4Ofnx8aNG7ly5Qq+vr4EBwfTv39/vLy8OHDggC4iCCFewdHWkt8H1WWae0VOhUTTav4RdgTck9FDHqaVzxgAHB0dNR88u7u7a7Zfu3ZNW7sUQrwjIyMVAxqUpnE5e7w3BTJyQwD7r0TwbefK2OUz03c8oWNygZsQQsPZ3grfIW6MbV2e/Vcf0nLeYXwO3iAqNun1Txa5hhQGIUQGJsZGDGtahh3DGlKpuA1zDwRTf9YhvtgUKFdO5xFam0oSQhi2isXys/rjOtx8FMvqk6H4ngtny/lwPihly4D6pWldyQETY/nbMjeSwiCEyFKZwlZ806kyX7Qqz2b/MFafvMOwdecpamOOl1tJPD9wks8hchkp90KIN2JjYcqgRs78NeZDfulXG2f7fHy/9zpusw4y3vciQQ+e6TuiyCYyYhBCvBVjIxUtKjrQoqIDwRHPWXkilK3nw9noH0Y9ZzsG1C9Ny4oOGBvJxauGSkYMQoh3Vs7BmpldqnBqYnMmtatA2OMEhqw9R+Pv/2Lp4Vs8jZf7TxsiKQxCiPdWwNKMwY1dODz2Q37uW4sSdhbM2nONerMOMmnbJYIjZCVXQyJTSUKIbGNibESbykVoU7kIQQ+esfJ4KFvOhbPu9F0alCnIR/VL07RCYZlmyuGkMAghtMK1aH5md6vK+LYV2HD2LmtO3mHQan+c7Czp51aSavnlVqM5lRQGIYRW2eUz47MPyzC4kTP7rkSw8sRtpu8OwtxERY9Q6OdWijKFrfQdU/wPKQxCCJ0wMTaifdWitK9alMv3nrLwjwA2nHlxXUTjcvZ8VL8UTcrZYyTTTHonHz4LIXSucnEbvBsW5sTEZnzRshzXHjzjo5VnaT7vMCuP3+Z5opzNpE9SGIQQelPISs3nzctybHwzFvWqga2lKdP8ruI26xDTdl7hdlScviPmSTKVJITQOzMTIzpWK0bHasUIDHvCyhOh/H76DqtOhtK0fGEG1C9Fo7KF5I6POiIjBiFEjlKtRAHm96zO8QnNGNm8LBfDn9LvtzO0mHeYNSdDiUtK1XfEXE8KgxAiRypsbc6oFuU4PqEp83tWI5/ahCk7rlBv1kGm77rK3eh4fUfMtWQqSQiRo6lNjOlSw5HO1YtzIewJK46HsvJEKL8ev03zCg583KAUbi4FZZopG0lhEEIYBJVKRU0nW2o62fKwnSu/n77DutN36R0UQTkHKwbUL02XGsWxMDPWd1SDp7WppMDAQLy8vF7alpCQgKenJ7du3dLW7oUQuVgRG3O+aFWe4xOaMadbVUyNjZi07RL1Zh1k1p4gwmNkmul9aGXEsHz5cnbu3ImFhUWmtkuXLvHVV18RERGhjV0LIfIQc1NjutcuQbdajvjfiWHl8VB+OXqb5UdCaFWxCAMalKJuaTuZZnpLWhkxODk54ePj89K25ORkfvzxR5ydnbWxayFEHqRSqfiglB0/9qnJkXFN+bSJC6duR+O57BTtFh1j09kwElNkbaY3pVIURdFGx+Hh4Xh7e7Np06aXtnt5eTFt2jRcXFxe21dAQABqtTq7IwKQmJiIubm5VvrWBkPLC4aX2dDygmR+maTUdP6+Hcv2oGeExiSTX21E23L5aV8+P/b53n6yxBCPMYCrq+tbP8cgPnxWq9Xv9OLeRFBQkNb61gZDywuGl9nQ8oJkfpXqVWCku8KpkMesPHGbzZcj8L3ylDaVi/BR/VLUKmn7xtNMhnqM34VBFAYhhHhXKpUKN5eCuLkUJOxxPGtO3WHDmbvsvviAysXzM6B+adyrFUVtImcz/UsnF7j5+fmxceNGXexKCCFeqYSdJZPauXJqUnNmdKlMUko6YzYH0uC7Q8zbf52IZ4n6jpgjaG3E4OjoqPl8wd3dPVP7mjVrtLVrIYTIkqWZCX3qlqR3HSeO34xm5Ynb+Px1k5/+vkW7KkX5qEEpajjZ6jum3shUkhAiz1KpVDQsW4iGZQtxJzqO1SfvsOlsGDsD71OtRAE+ql+KdlWKYmaSt1YPyluvVgghXqFkwXxM6VCRk5Oa802nSjxPTGHUxgAazD7Egj+DiUnIO4v3yYhBCCH+h5XahH5upehbtyRHb0ax4vhtFvx5AxMj6HgjjQENSlHVsYC+Y2qVFAYhhHgJIyMVTcrZ06ScPSGRsSzYfYF9Vx6y9cI9apW0ZUD9UrSpXART49w38ZL7XpEQQmQzZ3srhtYtxKlJzfnKvSLRsUl8vv4CDWcfYvGhG0THJuk7YraSEYMQQrwha3NTPmpQmv5upfg7+BErjofyw/5gFh26SadqxRjQoBSVitnoO+Z7k8IghBBvychIRbMKDjSr4MDNR89ZeSKULefusflcOHVK2fFRg1K0rOiAiYFOM0lhEEKI91CmsDXTO1dhbOsKbPYPY9XJUIb+fp5iNuZ4uZXC84MS2OYz03fMt2KY5UwIIXIYGwtTBjVy5u8xTVnmVYtShfIxe+816s06yIQtF7n28Jm+I74xGTEIIUQ2MjZS0apSEVpVKsL1hy+mmbZdCGfD2TDcnAsyoEEpWrg6YGyUc+8RISMGIYTQkvJFrJnVtQqnJjZnQtsK3H0cz6drztFkzl8sPxLC0/gUfUd8KSkMQgihZQUszRjSxIXDYz/k5741KV7Aghl/BFFv1kG+3HaJGxHP9R0xA5lKEkIIHTExNqJN5aK0qVyUK/efsupEKJvPhfP76bs0KluIAfVL0bR8YYz0PM0kIwYhhNCDSsVs+L5bNU5NbM7Y1uW5ERHLwFX+NJ37N78eu82zRP1NM0lhEEIIPbLLZ8awpmU4Or4pi3vXwN5Kzbe7ruI28yBf7bjMrchYnWeSqSQhhMgBTI2N6FC1GB2qFuNS+FNWnghl/ZkwVp28Q5Ny9nzUoBSNy9rrZJpJRgxCCJHDVHG0YW6Pahyf0AzvluW4+uAZA1acpcW8w6w6EUpsknaXAJfCIIQQOZS9tZoRzctyfHwzFnpWJ7+FKV/tvILbzIN843eV0Kg4rexXppKEECKHMzMxolP14nSqXpwLd2NYdSKUNadCWXHiNs3KF2ZAg1I0LFMIlSp7ppm0NmIIDAzEy8sr0/ZDhw7h4eFBz549NfeEFkII8WZqONmywLMGx8c34/NmZQkMf4LXr2doOf8Ia0/dIT75/aeZtDJiWL58OTt37sTCwiLD9pSUFGbNmoWvry8WFhb06tWLpk2bYm9vr40YQgiRaxXOb453y3IMa+rC7osPWHE8lMnbL/P93mv0/KAE/dxKvXPfWikMTk5O+Pj4MG7cuAzbb926hZOTEzY2L9Yrr1WrFv7+/rRt2zbrDq9fh6FDM27r0QM++wzi46Fdu8zPGTDgxb+oKOjWLXP70KHQsycmDx5k7hvgiy/A3f3Fvj/9NHP75MnQogUEBMCoUZnbZ86E+vXhxAmYNClz+4IFUL06/PknTJ+euX3pUihfHvz8YO5czWan+HiwtIQ1a6BECdi4EZYsyfx8X18oVAhWrnzx77/++ONFPz/9BC8buf3994v//vAD7NqVsc3CAvbsefH1t9/CwYMZ2wsWhC1bXnw9cSJOBw++2Ne/HB1h7doXX48a9eIY/q9y5WDZshdfDx4MwcEZ26tXf3H8APr2hfDwjO1ubjBr1ouvPTwgOjpje/PmMGXKi6/btoWEhAzNdh98AHPmvPjmww/JJJvee4SFwUtG1e/y3tO8L0Br7z2NbHrv2a5fD0eOZG7PxvceJ09mbH+P955TfPyL46rF9x4dOsCYMS++fsP3nhroCnQBzvcYxIoCFfnt2G1+PXKL3QPKZO7jDWilMLRu3Zrw/x4wIDY2Fmtra833+fLlIzb29efopqenExcfn2Hb84cPiQkKQpWQQIn/tAE8vX+fp0FBGMfEUPwl7TH37vE8KIi0pKRMfQM8DgsjNigIs9u3KfKS9qg7d4gPCkIdEoLDS9ojQ0NJsLXFIjQU+5e0R4SEkKRWY3nnDoVe0v7w1i2S09OxCgvD7n/a/z0W92/cIDU2Fut797B9yfPvBQeTFhmJzf372LykPezaNRQLC2wfPsT6Je13g4IAsIuIwOo/7UpaGmH/tBeKjMTyP+1pZmbc+6fdPioK9X/+/6U+fcr9f9odHj9G/Z/nJz95wsN/2os8eYLZf9qTHj8m4p/2Yk+fYvKf9oSoKCL/aS/+/DnG/2mPj4wk6p/2ErGxqJIy3n0rNTWVoH/anV5ybLLrvWfy4AHFsum9978/I9p67/0ru957VikpL/3Zy873nsV/2t/nvZeens5jLb/3YiMiePwe772iT+4yrGktejnYcGLpDuDdCgOKloSFhSndu3fPsC0oKEgZNGiQ5vsZM2Yoe/bseW1fV69ezfZ8uuhbGwwtr6IYXmZDy6soklkXDC2vorx7Zp2eruri4sKdO3d48uQJycnJ+Pv7U6NGDV1GEEII8Ro6OV3Vz8+P+Ph4evbsyYQJExg4cCCKouDh4YGDg4MuIgghhHhDWisMjo6OmtNR3d3dNdubNWtGs2bNtLVbIYQQ70mufBZCCJGBFAYhhBAZSGEQQgiRgRQGIYQQGUhhEEIIkYFKURRF3yFeJyAgALVare8YQghhUJKSkqhevfpbP88gCoMQQgjdkakkIYQQGUhhEEIIkYEUBiGEEBlIYRBCCJGBFAYhhBAZSGEQQgiRQZ4oDOnp6UydOpWePXvi5eXFnTt3MrT//vvveHh40K1bN/766y89pczodZn/fcygQYNYv369HhJmzpJV3unTp9O1a1e8vLzw8vLi+fPnekr6/16X+fDhw/To0YMePXowbdo0csKZ3VllDgoK0hxfLy8vqlSpwpGX3TpTh153jH/99Ve6du2Kh4cHBw4c0FPKjF6XedmyZXTq1Ik+ffrkmN8XAIGBgXi95Faxhw4dwsPDg549e2pWvH6tbLtVUA62b98+Zfz48YqiKMqFCxeUIUOGaNqio6OVdu3aKcnJycrz58+Vxo0bK+np6fqKqpFV5n/NnTtX6datm7Ju3Tpdx8vkdXk9PT2V6OhofUR7pawyP3/+XGnfvr0m87Jly3JE/jd5XyiKovzxxx+Kt7e3LqO9VFZ5nz59qjRp0kRJSkpSnjx5onz44Yf6iplBVpmvXbumuLu7K4mJiUpiYqLSuXNnJT4+Xl9RNZYtW6Z06NAh010zk5OTlRYtWihPnjxRkpKSlK5duyqPHj16bX95YsRw7tw5GjVqBED16tW5fPmyps3Ozo4dO3ZgampKVFQU+fPnR6VS6SuqRlaZAfbu3YtKpaJx48b6iJdJVnnT09O5c+cOU6dOxdPTE19fX33FzCCrzBcuXKBcuXLMnj2b3r17U6hQIezs7PQVVeN17wuA+Ph4fHx8+PLLL3UdL5Os8lpYWFCsWDESEhJISEjIET93kHXmW7duUadOHdRqNWq1mpIlS3L9+nV9RdVwcnLCx8cn0/Zbt27h5OSEjY0NZmZm1KpVC39//9f2lycKQ2xsLFZWVprvjY2NSU1N1XxvYmLC2rVr6dmzJ61bt9ZHxEyyyhwcHMyuXbsYOXKkvuJlklXe+Ph4+vbty5w5c/jll19Yt24d165d01dUjawyx8TEcPr0acaMGcPy5ctZtWoVt2/f1ldUjde9lwF8fX1p06ZNjihkr8tbtGhR2rdvT5cuXejXr58+ImaSVeby5cvj7+9PbGwsMTExXLhwgYSEBH1F1WjdujUmJpnvuxYbG4u1tbXm+3z58hEbG/va/vJEYbCysiIuLk7zfXp6eqaD2LdvX44ePcrZs2c5deqUriNmklXm7du3ExERQf/+/dm2bRsrV67U+1xyVnktLCzo168fFhYWWFlZUa9evRxRGLLKXKBAAapUqYK9vT358uWjdu3aBAUF6Suqxpu8l/38/Ojevbuuo71UVnmPHDnCo0ePOHjwIH///Td//vknFy9e1FdUjawyu7i40KdPHz755BNmz55NtWrVsLW11VfU1/rva4mLi8tQKF4lTxSGmjVran5xBgQEUK5cOU1bSEgIw4cPR1EUTE1NMTMzw8hI/4clq8zjxo1j8+bNrFmzhi5dujBgwAC9TylllTc0NJTevXuTlpZGSkoK58+fp1KlSvqKqpFV5sqVKxMcHMzjx49JTU0lMDCQMmXK6CuqRlaZAZ4/f05ycjJFixbVR7xMssprY2ODubk5ZmZmqNVqrK2tefbsmb6iamSV+fHjx8TExLB+/Xq+/PJLHjx4QNmyZfUV9bVcXFy4c+cOT548ITk5GX9/f2rUqPHa52ntns85ScuWLTl+/Dienp4oisLMmTNZsWIFTk5ONG/enAoVKtCzZ09UKhWNGjWiTp06+o782sw5zevyuru706NHD0xNTenUqVOO+GF6XeYvvviCQYMGAdCmTZtMv4T14XWZb9++TfHixfUdU+N1eU+cOEGPHj0wMjKiZs2aNGjQQN+Rs8zcrFkzwsPD8fDwwNTUlHHjxmFsbKzvyJn4+fkRHx9Pz549mTBhAgMHDkRRFDw8PHBwcHjt82V1VSGEEBnof85ECCFEjiKFQQghRAZSGIQQQmQghUEIIUQGUhiEEEJkIIVB6FRcXBz9+vWjZ8+emovc/P39WbZsmVb2d+DAASIiIrTSty5t3LiRlJSUt3rO/fv3OXTo0Bs/PiecKipyBikMQqeOHz9Os2bN+Oqrr/D19UVRFFavXk3//v21sr/Vq1e/0RIAOd3SpUtJT09/q+ecOnWK8+fPaymRyM3yxAVuIuewtLQkISGB+Ph4LC0t8fPzo2XLlqjV6pc+PjAwkBkzZqAoCg4ODvzwww988sknTJs2DRcXF9avX09UVBSDBw9m5MiRxMbGkpiYyNixY0lISCAoKIjx48ezbt061q5dy+7duzExMaF27dqMHTsWHx8f7ty5Q0xMDE+fPqV3797s37+f27dvM3v2bKpXr86aNWvYtWsXKpWKdu3a0a9fPyZMmMCTJ0948uQJP/30E6NGjUJRFFJSUvj6668pX7685jUkJiYyceJE7t+/T0pKClOmTKFy5cpMmjSJsLAw0tLS+Oijj2jXrh1eXl5UqFCBGzduEBsby8KFCzlx4gSRkZGMHj0aHx8fpk6dysOHD4mJiaFx48aMGjWK0NBQJk+eTEpKCubm5sydO5dly5aRmJhIjRo1cHR0ZPr06cCL5T5mzpyJpaUlU6ZM4ebNm5QoUYLk5GSdvAeEAci+hV+FeL20tDTl22+/Vby9vZWwsDDl888/V4KDg5UpU6Yoy5Yty/R4d3d35ebNm4qiKMratWuVy5cvK3379tVsW7dunbJo0SIlODhY8fDwUJ4/f66EhoYqf//9t6Ioiuax165dU7p166YkJycr6enpyrBhw5RDhw4pixYtUr788ktFURRl6dKlyogRIxRFURRfX19l+vTpyo0bNxRPT08lNTVVSUtLU7y8vJRbt24p48ePV1asWKEoiqL89ddfymeffaYkJCQoly5dUvz9/TO8hhUrVihz5sxRFEVRrl+/rqxYsUJZs2aNMmPGDEVRXizx3bJlSyU6Olrp27evsnPnTkVRFGXevHnK0qVLFUVRlKZNmyqJiYlKWFiYsmnTJkVRFCUxMVGpU6eOoiiKMmTIEOXw4cOKoijK7t27laNHjypbtmzR7Ld79+7KjRs3FEVRlE2bNinz5s1TDh06pFma+969e0qlSpXe8f+qyG1kxCB0ysjIiMmTJwMvpkf69evHkiVLmDx5MosXL+b27duULl1a8/jo6GhcXFwA6NOnT6b+lH8u3C9btix9+vTB29ub1NTUTDcsCQkJoVq1apiamgJQu3Ztbty4AUDFihUBsLa21qyHZGNjQ1JSEsHBwdy/f58BAwYA8PTpU+7evQugydm4cWNCQ0P57LPPMDExYejQoZn2/e9aVuXKlaNcuXJ8/fXX1K9fH3ix0JmLiwthYWEZ8hQpUoSoqKgMfRUoUIBLly5x6tQprKysNH/l3759W7MGTrt27QDYunWr5nm3bt3i66+/BiAlJYXSpUtz48YNqlatCkCxYsVyzPpKQv/kMwahF9HR0YSGhlK7dm0SEhIwNjZGpVJlWsK4cOHChIaGAi/unHXgwAHMzMyIjIwE4OrVqwBcv36duLg4li1bxnfffce3334LgEqlQlEUnJ2duXjxIqmpqSiKwtmzZzW/2LO6D4CzszNlypRh9erVrFmzhq5du2rWTPr3eadPn6Zw4cL89ttvDB06lHnz5mXow8XFhUuXLgEQFhbGF198gYuLi2Zd/NjYWIKDg3F0dHxlDpVKRXp6Olu3bsXa2pq5c+fy8ccfk5iYiKIoGfaxc+dO1qxZg5GRkeZzidKlSzN79mzWrFnD2LFjadKkCc7OzgQEBAAQERGRKz6kF9lDRgxCL5YsWcKQIUMA6N27NwMHDqRYsWJUqFAhw+O+/vprJk2ahJGREfb29gwYMAAzMzO++eYbihYtSuHChQEoVaoUP/74I9u3b8fU1JQRI0YAUKNGDcaNG8dvv/1G27Zt6dWrF+np6dSqVYsWLVq8dvnvChUq4ObmRq9evUhOTqZq1aqZFiGrUKECo0ePZtWqVRgZGTFs2LAM7Z6enkyaNIm+ffuSlpbGpEmTKF++PFOmTKFXr14kJSUxfPhwChYs+MoctWvXZvDgwUydOhVvb2/OnTuHhYUFJUuW5NGjR4wbN46pU6eyZMkSzM3NmTNnDvfv32fJkiVUqlSJadOmMX78eNLS0gCYMWMGpUuX5ty5c3Tv3p1ixYrl6OWjhW7JInpCCCEykKkkIYQQGUhhEEIIkYEUBiGEEBlIYRBCCJGBFAYhhBAZSGEQQgiRgRQGIYQQGfwfeLzNtuofgV4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print('Logistic:')\n",
    "lift_plot(logit_report)\n",
    "print('Gradient Boosting:')\n",
    "lift_plot(gb_report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:13.625728Z",
     "start_time": "2020-03-04T12:40:12.404759Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic waterfall:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print('Logistic waterfall:')\n",
    "wtf_plot(logit_report,'g')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-03-04T12:40:17.392761Z",
     "start_time": "2020-03-04T12:40:13.678687Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gradient Boosting waterfall:\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW8AAAD3CAYAAADSftWOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAYeklEQVR4nO3de3TMd/7H8efkIiQRCSm2whL3S7Hq1nZLUIJKcZq6drpZ2x/Ldm3UJUhCVMuGrsuhGpftSTco6tZ2W10Hq+nG5di21qlm3fZYgiIRJyZkOknm94c22zRMIrfx4fX4azKf+c737Z3x+n7nk+/F4nQ6nYiIiFE83F2AiIjcO4W3iIiBFN4iIgZSeIuIGEjhLSJiIK/qWMnRo0fx8fGpjlVVCbvdbnT97qb+VYz6VzEm989ut9O5c+c7jlVLePv4+NC2bdvqWFWVSE9PN7p+d1P/Kkb9qxiT+5eenn7XMU2biIgYSOEtImIghbeIiIGqZc5bREpyOBxkZGSQl5dX6utczX2Kayb0r2bNmoSEhODt7V3mZRTeIm6SkZFB7dq1adq0KRaL5a6vu3XrFrVq1arGyh4s93v/nE4nWVlZZGRk0KxZszIvp2kTETfJy8ujXr16LoNbHnwWi4V69eqV+g3spxTeIm6k4BYo3+dA4S0iYiCFt8hD7Pz58/z+97/HarUyatQoEhISsNlslb6ejIwMRowYAUDfvn2x2+2Vvo77zYkTJzhy5EiVvb8R4V1QcG9zQZXN3WdnufvfLw+mvLw8Jk2axMsvv0xKSgqbNm2iU6dOTJ061d2lPRB2797N6dOnq+z9jTjaxNOzJvv3P7xzg2FhutnRwyAsLKzEcyNGjODXv/41N2/eZPDgwSXGo6KiiIqKIjMzk8jIyGJj+/fvd7m+/fv3061bNzp16lT03PDhw3nvvff4z3/+w4QJE/jggw/w9fVl3bp1eHl5ER4eTnx8fNH1QubPn09BQQETJ04kMDCQXr160alTJ1auXAnc3kAkJiaWegjckCFDaNq0KTVq1GDevHnExsaSnZ0NQFxcHK1bt2bmzJmcO3cOu93Ob37zGwYPHszgwYPp2rUrp06dok6dOixZsgRvb29mz57N+fPnKSgoYOzYsQwbNgyr1UqbNm04deoUNpuN5cuXExwczB/+8AdsNht5eXlMnz6dHj16sGvXLpKTk/Hw8ODxxx9n2rRpxeq1Wq0EBQWRk5PDihUriIuL48aNG2RnZ/PCCy/Qr18/duzYgbe3N+3btycvL4+lS5fi6elJ48aNee211+7psMA7MSK8RaTynT9/niZNmpR4PiQkhKtXrzJgwAB2797NsGHD+OSTT/jzn//MvHnzsFqt9O7dm4MHD/Lmm28yZcoUrl69yrZt26hRowYbNmxg8eLFNGjQgKSkJD799FMiIiJc1nLz5k0mTZpEu3btWLx4MT179mTMmDGcPXuWWbNmsXbtWg4fPsy2bdsASEtLA25vHCIiIujWrRuLFi1i8+bNeHt7ExQUxOLFi7HZbAwbNoxevXoB0LFjR2JjY1m6dCkff/wxffr0ITMzk+TkZLKysjh79izXr19nxYoVbNu2jVq1ajF9+nTS0tJ46qmnitUcERFB//79OX78OM8++ywDBgzg8uXLWK1WxowZw/DhwwkODuaxxx5j4MCBbNy4kXr16rFs2TJ27NhRNI1UXgpvkfvE3faUb926ha+vr8s96eDg4FL3tH+qQYMGHDt2rMTzZ8+e5dFHH+WFF14gISGB0NBQmjZtSlBQECdPnmT16tWsW7cOp9NZtPcYEhJCjRo1it73jTfewNfXl8uXL9OlS5cy1fPDMc4nT57k0KFD7Nq1C4CcnBz8/f2Jj48nPj4em83Gc889B4CXlxfdunUDoEuXLqSmpuLp6cmTTz4JgL+/P82bN+f8+fMAtGvXDoCGDRuSmZlJy5YtGTt2LK+++ir5+flYrVbOnTvHtWvXGD9+PAC5ublFy9+p3uDgYN599112796Nv78/+fn5xV537do1rly5QnR0NHB7g/PTDUF5KLxFHlL9+vUjKSmJY8eO0bFjRwDef/996tatS+PGjYHbJ5CsW7eO0aNHAxAaGsq4cePo0qULZ86cKfqDnIfH//58FhcXx549e/D39ycmJoay3uP8h/cIDQ3lueeeIyIigqysLN5//32uXLnC8ePHeeutt7Db7fTu3ZuhQ4eSn5/Pv//9b9q0acMXX3xBixYt8PDw4J///Cf9+/fHZrNx6tQpQkJC7rjOEydOkJuby5o1a7hy5QqjRo1i69at/OxnP+Odd97B29ub7du33/HvXj8c3vfOO+/QuXNnxowZw6FDh/jss8+KxgsLCwkKCqJhw4asWrWK2rVrs3fvXnx9fcvUE1cU3iIPKT8/P5KSkliwYAHXr1+noKCA1q1bs2TJkqLXREZGsnz5cnr27AlATEwMCQkJ2O128vLyiI2NLfG+Q4cOZcSIEQQEBBAcHMyVK1fuqa7f/va3xMbGsmXLFmw2G6+88gqPPPIIV69eZdiwYfj6+jJu3Di8vG7H19q1a7l48SKPPvooU6ZMASA+Pp7Ro0djt9uZMGEC9erVu+O6mjZtyltvvcXOnTvx9vZm8uTJ1K1bl6ioKKxWKwUFBTRq1IhBgwbdtd4+ffqQkJDARx99RGBgIJ6ennz33Xd06NCBRYsW0bx5c2JjYxk/fjxOpxM/Pz8WLVp0Tz25E4uzrJvFCqiM6+nqD5bmMvl6ylWprH2530/vdqe+ffuya9culzdbMKV/d/o8uPqMGHGooIiIFKdpExEx1r59+9xdgttoz1vEjaph1lIMUJ7PgcJbxE1q1qxJVlaWAvwh98MlYWvWrHlPy2naRMRNQkJCyMjI4OrVqy5f53A4Knw23sPMhP79cDOGe6HwFnETb2/vMl18X0frVMyD2j9Nm4iIGEjhLSJiIIW3iIiBFN4iIgZSeIuIGKjUo00cDgczZ87kwoULeHh4MH/+fLy8vJg5cyYWi4WWLVsyd+7cYlcVExGRqlVqeH/22Wfk5+ezadMm0tLSWLZsGQ6Hg+joaHr06MGcOXPYu3cv/fv3r456RUSEMkybNGvWjIKCAgoLC7HZbHh5eXH8+HG6d+8OQK9evThw4ECVFyoiIv9T6p63r68vFy5cYNCgQWRnZ5OUlMSRI0eKLkTu5+fHjRs3XL6H3W4nPT293EU+iAfY36uK9M/d8vLyjK7f3dS/inlQ+1dqeCcnJ/PLX/6SqVOncunSJX71q1/hcDiKxnNzcwkICHD5Hj4+PgrgCjK5fw/qGW7VRf2rGJP752qjU+q0SUBAALVr1wagTp065Ofn065dOw4fPgxAamoqXbt2raRSRUSkLErd846KimL27NmMGTMGh8PBlClT6NChA/Hx8SxZsoTQ0FDCw8Oro1YREfleqeHt5+fH8uXLSzy/fv36KilIRERKp4OzRUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwvshUFCQ59b1t23b1q3rd/e/X6QqeJXlRatXr2bfvn04HA5Gjx5N9+7dmTlzJhaLhZYtWzJ37lw8PLQduF95etZk/36Lu8twm7Awp7tLEKl0pSbu4cOH+eqrr3jvvfdISUnh22+/ZeHChURHR7Nx40acTid79+6tjlpFROR7pe55/+Mf/6BVq1b87ne/w2azMWPGDLZs2UL37t0B6NWrF2lpafTv3/+u72G320lPTy93ke7+2n0/UP8qpiL9c7e8vDyj63e3B7V/pYZ3dnY2Fy9eJCkpiYyMDCZOnIjT6cRiuf013M/Pjxs3brh8Dx8fHwVIBal/FWNy/9LT042u391M7p+rjU6p4R0YGEhoaCg1atQgNDQUHx8fvv3226Lx3NxcAgICKqdSEREpk1LnvB9//HE+//xznE4nly9f5tatWzzxxBMcPnwYgNTUVLp27VrlhYqIyP+Uuufdp08fjhw5QmRkJE6nkzlz5hASEkJ8fDxLliwhNDSU8PDw6qhVRES+V6ZDBWfMmFHiufXr11d6MSIiUjY6OFtExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEBluqpgRZ04cYKJEycWe27EiBFMmjSJmzdvMnjw4BLLREVFERUVRWZmJpGRkVy/Xnz8ueegb1+4cgUWLCi5zhEj4Mkn4dw5WLKk5LjVCo8/DqdPw8qVJcdffhk6dICvv4Z160qOv/IKtGgBX3wBKSklx199FZo0gQMHYMuWkuOzZ0P9+rBvH3z4YcnxefOgTh349FNISAgrMf7JJ5/g6+vLqlWr2HKHFezfvx+AN998k7/+9a/F+ufjA4mJtx//5S/w5ZfFlw0IgNdeu/147Vo4frz4+COPQGzs7ccrV97u4Y+FhMC0aXy/fsjIKD7eosXt/gG88QZcvVp8vH17+L//u/14zhzIySk+3qULvPTS7ccxMWC3Fx9/4gkYOfL24+hoCAwMKzZ+r5+9n5o4cSIjR47k/PnzWK3WEuNTp04lIiKCEydOMGHChBLjcXFxPPPMMxw9epTo6OgS4wsWLODJJ5/kwIEDzJ49m5s3b+Lr61s0vmzZMjp37syePXt4/fXXSyy/evVqWrduzUcffcSf/vSnEuMpKSk0btyYzZs38/bbb5cY37p1K8HBwSQnJ5OcnFxi/F4/ez9Wq1Ytdu3aBcD8+fNL3P+2Xr16bNu2DYBZs2Zx8ODBYuMhISFFVzSNjo7m6NGjxcZbtWrFmjVrABg/fjwnT54s1r/OnTuzbNkyAF588UUyfvLhfOKJJ1i4cCEAzz//PFlZWcXG+/XrR3x8PACDBg3i1q1bxcaHDBnCtO8//GFhYSV6c6+fvTv9fn6gPW8REQNZnE6ns6pXUhn3kNu/31JJ1ZgnLKzivyL1z1wm34PxfmBy/1zVrj1vEREDKbxFRAyk8BYRMZDCW0TEQApvEREDKbxFRAyk8BYRMZDCW0TEQApvEREDKbxFRAyk8BYRMZDCW0TEQApvEREDKbxFRAyk8BYRMZDCW0TEQApvEREDKbxFRAyk8BYRMZDCW0TEQApvEREDKbxFRAxUpvDOysqid+/enDlzhv/+97+MHj2aMWPGMHfuXAoLC6u6RhER+YlSw9vhcDBnzhxq1qwJwMKFC4mOjmbjxo04nU727t1b5UWKiEhxpYZ3YmIio0aNon79+gAcP36c7t27A9CrVy8OHDhQtRWKiEgJXq4Gt2/fTt26dXn66adZs2YNAE6nE4vFAoCfnx83btwodSV2u5309PRyF9m2bdtyL/ugUP8qpiL9c7e8vDyj63e3B7V/LsN727ZtWCwWDh48SHp6OjExMVy7dq1oPDc3l4CAgFJX4uPjowCpIPWvYkzuX3p6utH1u5vJ/XO10XEZ3hs2bCh6bLVaSUhIYPHixRw+fJgePXqQmppKz549K69SEREpk3s+VDAmJoYVK1YwcuRIHA4H4eHhVVGXiIi44HLP+8dSUlKKHq9fv75KihERkbLRSToiIgZSeIuIGEjhLSJiIIW3iIiBFN4iIgZSeIuIGEjhLSJiIIW3iIiBFN4iIgZSeIuIGEjhLSJiIIW3iIiBFN4iIgZSeIuIGEjhLVKKgoI8t67f3XeBcfe/X+6szNfzFnlYeXrWZP9+i7vLcJuwMKe7S5A70J63iIiBFN4iIgZSeIuIGEjhLSJiIIW3iIiBFN4iIgZSeIuIGEjhLSJiIIW3iIiBFN4iIgZSeIuIGEjhLSJiIIW3iIiBFN4iIgZSeIuIGEjhLSJVyt03c3hQb2ahmzGISJXSzSyq5mYW2vMWETGQwltExEAKbxERAym8RUQMpPAWETGQwltExEAuDxV0OBzMnj2bCxcu8N133zFx4kRatGjBzJkzsVgstGzZkrlz5+LhoW2AiEh1chneH374IYGBgSxevJjs7GyGDx9OmzZtiI6OpkePHsyZM4e9e/fSv3//6qpXREQoJbwHDhxIeHh40c+enp4cP36c7t27A9CrVy/S0tJKDW+73U56enq5i3T3GVL3A/WvYtS/ilH/KqYi/bsbl+Ht5+cHgM1mY/LkyURHR5OYmIjFYikav3HjRqkr8fHx0S+wgtS/ilH/Kkb9q5jy9s9V6Jc6WX3p0iVeeuklhg4dSkRERLH57dzcXAICAspVlIiIlJ/L8M7MzGTcuHFMnz6dyMhIANq1a8fhw4cBSE1NpWvXrlVfpYiIFOMyvJOSksjJyWHVqlVYrVasVivR0dGsWLGCkSNH4nA4is2Ji4hI9XA55x0XF0dcXFyJ59evX19lBYmISOl0gLaIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBlJ4i4gYSOEtImIghbeIiIEU3iIiBvIqz0KFhYUkJCRw4sQJatSoweuvv87Pf/7zyq5NRETuolx73nv27OG7775j8+bNTJ06lT/+8Y+VXZeIiLhQrvD+4osvePrppwHo3LkzX3/9daUWJSIirpVr2sRms+Hv71/0s6enJ/n5+Xh53fnt7HY76enp5avwew0afFOh5U1W0d6B+ldR6l/FqH/lY7fb7zpWrvD29/cnNze36OfCwsK7Bjfc3jsXEZHKU65pky5dupCamgrA0aNHadWqVaUWJSIirlmcTqfzXhf64WiTkydP4nQ6WbBgAc2bN6+K+kRE5A7KFd4iIuJeOklHRMRACm8REQMpvEVEDKTw/ol//etfWK3Wu45fvHiRffv2VWNF9z+Hw8H06dMZM2YMkZGR7N27t0zLjRgxgoyMjCquzgwFBQXMmjWLUaNGMXbsWM6dO3fX19rtdvr27VuN1ZkjKyuL3r17c+bMGU6cOMGRI0cA6Nu3r8tjpk2k8P6RtWvXEhcX5/KXfOjQIb788stqrOr+9+GHHxIYGMjGjRtZu3Yt8+fPd3dJxvn73/8OwKZNm5g8eTILFy50c0XmcTgczJkzh5o1awKwe/duTp8+7eaqqk65TtJ5UDVp0oQVK1YwY8YMADZs2MDOnTvx8PCgS5cuTJs2jTVr1pCXl8cvfvEL+vXr5+aK7w8DBw4kPDy86GdPT0+sVitt2rTh1KlT2Gw2li9fTqNGjVi6dCmff/45DRs2JDs7241V31+eeeYZwsLCgNvf7oKDg7FarQQFBZGTk1P0uczJyaFJkybuLfY+lZiYyKhRo1izZg3Xr19nx44deHt70759ewASEhKKvumtXLmSOnXquLPcCtOe94+Eh4cXO1N0+/btxMbGsnnzZho3bozT6WT8+PEMGTJEwf0jfn5++Pv7Y7PZmDx5MtHR0QB07NiR5ORknnrqKT7++GNOnjzJkSNH2Lp1K4sWLSp2lq6Al5cXMTExzJ8/v2hjGBERQXJyMh988AGtWrViw4YNjBo1ys2V3n+2b99O3bp1i665FBgYyPDhw4mKiqJjx44APP/886SkpNCoUSPS0tLcWW6lUHi7sHDhQjZt2sSLL77IxYsX0SHxd3fp0iVeeuklhg4dSkREBADt2rUDoGHDhtjtdk6fPk2HDh3w8PDA399fZ+beQWJiIn/729+Ij4/n5s2bNGvWDIBTp07x2GOPAdCpUyeXl6N4GG3bto0DBw5gtVpJT08nJiaGq1evFntNhw4dAAgODiYvL88dZVYqhbcLW7ZsYd68eaxfv5709HS++uorPDw8KCwsdHdp95XMzEzGjRvH9OnTiYyMvOvrmjVrxrFjxygsLOTmzZsP9Hzkvdq5cyerV68GoFatWlgsFjw9PbFYLACEhoZy9OhRAL755hvy8/PdVuv9aMOGDaxfv56UlBTatm1LYmIi9evXL/Z/9YdePii0+XahdevWREZGEhQURIMGDejUqRP+/v68/fbbtG/fnmeffdbdJd4XkpKSyMnJYdWqVaxatQrgjns2bdu2ZeDAgURGRlK/fn3q1atX3aXetwYMGMCsWbMYO3Ys+fn5zJ49m3fffbdofOzYscyaNYvRo0cTGhqKt7e3G6s1Q4cOHVi0aNEDe+kOnR4vImIgTZuIiBhI4S0iYiCFt4iIgRTeIiIGUniLiBhI4S0iYiCFt4iIgf4fL+GpWFVrEPsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print('Gradient Boosting waterfall:')\n",
    "wtf_plot(gb_report,'y')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The comparison between Logistic and GB demonstrate again the similarity between the two models, even if we display all the three indeces. <br>\n",
    "The first type of graph describe how many 1's we are able to predict w.r.t the total 1's in the test set. For example if the bank contacts the first 20% of people that belongs to the 'loyal clients' group, it is able to reach more than 30% of customer with a purchase.\n",
    "The second plot shows how better is the model prediction respect to the prediction made randomly.\n",
    "The third and last plot shows the response rate for each group of client, compared with the overall response rate (51%).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.4 Conclusions\n",
    "\n",
    "At the end of the analysis, it is possible to say that:\n",
    "\n",
    "- The threshold provides a tool that guarantees to establish to which category the customer belongs to. The choice of the cutoff is driven to a good classification of 1 in order to don't waste financial resources during the implementation of future marketing campaigns. The best models, following this approach, are: logistic and gradient boosting;\n",
    "\n",
    "- The use of lift score and cumulative gains, applied to the same model of the previous section, identifies the ability of the classifier to allocate resources in a good manner. They allow to measure the model performance by using the actual response rates to see how much the predictive model would have helped in this situation. The information can be used to determine whether we should use this model or one similar to it in the future.\n",
    "\n",
    "- Furthermore, we need to take into account that we have lost an important part of information due to data cleaning( for example, we have reparametrize pdays into 4 catgories and we have removed default variable). For this reason accuracy might not reach optimal levels (all models under 80% of accuracy).\n",
    "\n",
    "- For future implementation in business context, in order to best exploit the predictive capabilities of our model, it would be strongly suggested to enrich data features. "
   ]
  }
 ],
 "metadata": {
  "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": "3.7.3"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autoclose": false,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}