Classification_Models.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Project 3 Modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def warn(*args, **kwargs):\n",
    "    pass\n",
    "import warnings\n",
    "warnings.warn = warn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "import pickle\n",
    "import statistics\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from collections import defaultdict\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.pyplot import figure\n",
    "import seaborn as sns\n",
    "import itertools\n",
    "from sklearn.decomposition import PCA\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import StackingClassifier \n",
    "\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.preprocessing import StandardScaler, label_binarize, LabelEncoder, OneHotEncoder\n",
    "from sklearn.model_selection import train_test_split, KFold, cross_val_score, StratifiedKFold\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import GridSearchCV, RandomizedSearchCV\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.tree import DecisionTreeClassifier, ExtraTreeClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier, VotingClassifier\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.metrics import precision_score, recall_score, accuracy_score, f1_score, log_loss, make_scorer\n",
    "from sklearn.metrics import precision_recall_fscore_support as score\n",
    "from sklearn.metrics import roc_auc_score, roc_curve, classification_report, confusion_matrix, auc\n",
    "from sklearn.naive_bayes import GaussianNB, BernoulliNB, MultinomialNB\n",
    "from imblearn.over_sampling import SMOTE, ADASYN, RandomOverSampler\n",
    "from xgboost import XGBClassifier\n",
    "from catboost import CatBoostRegressor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df=pd.read_pickle(\"Data/df.pkl\")\n",
    "df_drop=pd.read_pickle(\"Data/df_drop.pkl\")\n",
    "df_no_na=pd.read_pickle(\"Data/df_no_na.pkl\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_drop = df_drop.reset_index(drop = True)\n",
    "df_no_na = df_no_na.reset_index(drop = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_save = df.iloc[:,1:] # Without the ID"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_save.to_csv('df_save.csv') "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train=df_drop.values\n",
    "\n",
    "X = df_train[:, 2:].astype(str)\n",
    "y = df_train[:, 1].astype(str)\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.30, random_state=4444)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Setting up a columns name translator \n",
    "X_dummies = df_drop.iloc[:, 2:].astype(str)\n",
    "X_dummies = pd.get_dummies(X_dummies)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "175"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(X_dummies.columns) # Verify the length from \"get_dummies\" is the same as \"one_hot_encode\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "36"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(X_train[0]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "names = list(X_dummies.columns) # Feature reference list for interpretation "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# one-hot encode input variables\n",
    "onehot_encoder = OneHotEncoder(sparse=False)\n",
    "onehot_encoder.fit(X_train)\n",
    "X_train = onehot_encoder.transform(X_train)\n",
    "X_test = onehot_encoder.transform(X_test)\n",
    "#ordinal encode target variable\n",
    "label_encoder = LabelEncoder()\n",
    "label_encoder.fit(y_train)\n",
    "y_train = label_encoder.transform(y_train)\n",
    "y_test = label_encoder.transform(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0., 1., 0., ..., 0., 0., 0.],\n",
       "       [0., 1., 0., ..., 0., 0., 0.],\n",
       "       [1., 0., 0., ..., 0., 1., 0.],\n",
       "       ...,\n",
       "       [1., 0., 0., ..., 0., 0., 0.],\n",
       "       [0., 1., 0., ..., 0., 0., 0.],\n",
       "       [0., 1., 0., ..., 0., 0., 0.]])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## RandomForest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training: 100.00%\n",
      "Test set:  85.42%\n",
      "Accuracy_: 85.42\n",
      "roc_auc_score: 79.56\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.87      0.94      0.90      1561\n",
      "           1       0.81      0.66      0.72       641\n",
      "\n",
      "    accuracy                           0.85      2202\n",
      "   macro avg       0.84      0.80      0.81      2202\n",
      "weighted avg       0.85      0.85      0.85      2202\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model = RandomForestClassifier()\n",
    "# fit on the training set\n",
    "model.fit(X_train, y_train)\n",
    "# predict on test set\n",
    "yhat = model.predict(X_test)\n",
    "# evaluate predictions\n",
    "print(\"Training: {:6.2f}%\".format(100*model.score(X_train, y_train)))\n",
    "print(\"Test set: {:6.2f}%\".format(100*model.score(X_test, y_test)))\n",
    "accuracy = accuracy_score(y_test, yhat)\n",
    "roc_auc = roc_auc_score(y_test, yhat)\n",
    "print('Accuracy_: %.2f' % (accuracy*100))\n",
    "print('roc_auc_score: %.2f' % (roc_auc*100))\n",
    "print(classification_report(y_test, yhat))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# get importance\n",
    "importance = model.feature_importances_\n",
    "# summarize feature importance\n",
    "# for i,v in enumerate(importance):\n",
    "#     print('Feature: %0d, Score: %.5f' % (i,v))\n",
    "# plot feature importance\n",
    "plt.bar([x for x in range(len(importance))], importance);\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 85,  99,   4,  42,  67, 134,  89,  77,   8,  76,  74,  59,  41,\n",
       "        49,  71,  73,  34,  35,   9,  70,  53,  56,  68,  52,  50,  33,\n",
       "        55,  32,   0,   1])"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "indices = np.argsort(importance)[-30:]\n",
    "indices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_raw=[]\n",
    "for i in indices:\n",
    "    feature_raw.append(names[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['age_group_55 - 64 Years',\n",
       " 'income_poverty_> $75,000',\n",
       " 'h1n1_concern_2.0',\n",
       " 'health_worker_1.0',\n",
       " 'opinion_seas_vacc_effective_4.0',\n",
       " 'employment_industry_fcxhlnwr',\n",
       " 'education_College Graduate',\n",
       " 'opinion_seas_sick_from_vacc_2.0',\n",
       " 'h1n1_knowledge_1.0',\n",
       " 'opinion_seas_sick_from_vacc_1.0',\n",
       " 'opinion_seas_risk_5.0',\n",
       " 'opinion_h1n1_sick_from_vacc_2.0',\n",
       " 'health_worker_0.0',\n",
       " 'opinion_h1n1_vacc_effective_4.0',\n",
       " 'opinion_seas_risk_2.0',\n",
       " 'opinion_seas_risk_4.0',\n",
       " 'doctor_recc_seasonal_0.0',\n",
       " 'doctor_recc_seasonal_1.0',\n",
       " 'h1n1_knowledge_2.0',\n",
       " 'opinion_seas_risk_1.0',\n",
       " 'opinion_h1n1_risk_2.0',\n",
       " 'opinion_h1n1_risk_5.0',\n",
       " 'opinion_seas_vacc_effective_5.0',\n",
       " 'opinion_h1n1_risk_1.0',\n",
       " 'opinion_h1n1_vacc_effective_5.0',\n",
       " 'doctor_recc_h1n1_1.0',\n",
       " 'opinion_h1n1_risk_4.0',\n",
       " 'doctor_recc_h1n1_0.0',\n",
       " 'seasonal_vaccine_0',\n",
       " 'seasonal_vaccine_1']"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_raw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_clean=[]\n",
    "for i in feature_raw:\n",
    "    if \".\" in (i.split(\"_\")[-1]):\n",
    "        feature_clean.append(\"_\".join(i.split(\"_\")[:-1]))\n",
    "    else:\n",
    "        feature_clean.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_keep = set(feature_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "17"
      ]
     },
     "execution_count": 139,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(feature_keep)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {},
   "outputs": [],
   "source": [
    "zip_list=list(zip(feature_clean, importance[-30:]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('age_group_55 - 64 Years', 0.0018459710968270911),\n",
       " ('income_poverty_> $75,000', 0.0020599087288576535),\n",
       " ('h1n1_concern', 0.0017853189541767842),\n",
       " ('health_worker', 0.005403747053501599),\n",
       " ('opinion_seas_vacc_effective', 0.0030382179894259797),\n",
       " ('employment_industry_fcxhlnwr', 0.0018126966996551356),\n",
       " ('education_College Graduate', 0.0015145449309786779),\n",
       " ('opinion_seas_sick_from_vacc', 0.0015201488871464974),\n",
       " ('h1n1_knowledge', 0.0072727450524772246),\n",
       " ('opinion_seas_sick_from_vacc', 0.005096474122892275),\n",
       " ('opinion_seas_risk', 0.002261676249823239),\n",
       " ('opinion_h1n1_sick_from_vacc', 0.004790131111084453),\n",
       " ('health_worker', 0.001615459755866395),\n",
       " ('opinion_h1n1_vacc_effective', 0.002508738110761985),\n",
       " ('opinion_seas_risk', 0.0013480392000766598),\n",
       " ('opinion_seas_risk', 0.002533108920855741),\n",
       " ('doctor_recc_seasonal', 0.004382791078920471),\n",
       " ('doctor_recc_seasonal', 0.001585181270511172),\n",
       " ('h1n1_knowledge', 0.0017842324807938192),\n",
       " ('opinion_seas_risk', 0.00035811754769925604),\n",
       " ('opinion_h1n1_risk', 0.0017685203450124659),\n",
       " ('opinion_h1n1_risk', 0.0006441837636814123),\n",
       " ('opinion_seas_vacc_effective', 0.0018830785900443142),\n",
       " ('opinion_h1n1_risk', 0.0021150139811904538),\n",
       " ('opinion_h1n1_vacc_effective', 0.0015392781092618119),\n",
       " ('doctor_recc_h1n1', 0.0013854223939411608),\n",
       " ('opinion_h1n1_risk', 0.004107831501166931),\n",
       " ('doctor_recc_h1n1', 0.0027106803624739823),\n",
       " ('seasonal_vaccine_0', 0.00544740826312538),\n",
       " ('seasonal_vaccine_1', 0.0016819483209400627)]"
      ]
     },
     "execution_count": 169,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "zip_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_dict = {('age_group_55 - 64 Years', 0.0018459710968270911),\n",
    " ('income_poverty_> $75,000', 0.0020599087288576535),\n",
    " ('h1n1_concern', 0.0017853189541767842),\n",
    " ('health_worker', 0.005403747053501599),\n",
    " ('opinion_seas_vacc_effective', 0.0030382179894259797),\n",
    " ('employment_industry_fcxhlnwr', 0.0018126966996551356),\n",
    " ('education_College Graduate', 0.0015145449309786779),\n",
    " ('opinion_seas_sick_from_vacc', 0.0015201488871464974),\n",
    " ('h1n1_knowledge', 0.0072727450524772246),\n",
    " ('opinion_seas_sick_from_vacc', 0.005096474122892275),\n",
    " ('opinion_seas_risk', 0.002261676249823239),\n",
    " ('opinion_h1n1_sick_from_vacc', 0.004790131111084453),\n",
    " ('health_worker', 0.001615459755866395),\n",
    " ('opinion_h1n1_vacc_effective', 0.002508738110761985),\n",
    " ('opinion_seas_risk', 0.0013480392000766598),\n",
    " ('opinion_seas_risk', 0.002533108920855741),\n",
    " ('doctor_recc_seasonal', 0.004382791078920471),\n",
    " ('doctor_recc_seasonal', 0.001585181270511172),\n",
    " ('h1n1_knowledge', 0.0017842324807938192),\n",
    " ('opinion_seas_risk', 0.00035811754769925604),\n",
    " ('opinion_h1n1_risk', 0.0017685203450124659),\n",
    " ('opinion_h1n1_risk', 0.0006441837636814123),\n",
    " ('opinion_seas_vacc_effective', 0.0018830785900443142),\n",
    " ('opinion_h1n1_risk', 0.0021150139811904538),\n",
    " ('opinion_h1n1_vacc_effective', 0.0015392781092618119),\n",
    " ('doctor_recc_h1n1', 0.0013854223939411608),\n",
    " ('opinion_h1n1_risk', 0.004107831501166931),\n",
    " ('doctor_recc_h1n1', 0.0027106803624739823),\n",
    " ('seasonal_vaccine_0', 0.00544740826312538),\n",
    " ('seasonal_vaccine_1', 0.0016819483209400627)}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_dict_df = pd.DataFrame(imp_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_dict_df.columns = [\"Features\", \"Importance\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_dict_df.to_csv(\"Data/imp_features.csv\", index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30"
      ]
     },
     "execution_count": 150,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(zip_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import Counter, defaultdict\n",
    "\n",
    "imp=defaultdict(float)\n",
    "for i in zip_list:\n",
    "    imp[i[0]]+=i[1]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_df=pd.Series(imp).sort_values(ascending = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_df = pd.DataFrame(imp_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_df.reset_index(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [],
   "source": [
    "imp_df.rename(columns = {'index':'features', 0:'importance'}, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {},
   "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>features</th>\n",
       "      <th>importance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>h1n1_knowledge</td>\n",
       "      <td>0.009057</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>opinion_h1n1_risk</td>\n",
       "      <td>0.008636</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>health_worker</td>\n",
       "      <td>0.007019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>opinion_seas_sick_from_vacc</td>\n",
       "      <td>0.006617</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>opinion_seas_risk</td>\n",
       "      <td>0.006501</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>doctor_recc_seasonal</td>\n",
       "      <td>0.005968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>seasonal_vaccine_0</td>\n",
       "      <td>0.005447</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>opinion_seas_vacc_effective</td>\n",
       "      <td>0.004921</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>opinion_h1n1_sick_from_vacc</td>\n",
       "      <td>0.004790</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>doctor_recc_h1n1</td>\n",
       "      <td>0.004096</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>opinion_h1n1_vacc_effective</td>\n",
       "      <td>0.004048</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>income_poverty_&gt; $75,000</td>\n",
       "      <td>0.002060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>age_group_55 - 64 Years</td>\n",
       "      <td>0.001846</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>employment_industry_fcxhlnwr</td>\n",
       "      <td>0.001813</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>h1n1_concern</td>\n",
       "      <td>0.001785</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>seasonal_vaccine_1</td>\n",
       "      <td>0.001682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>education_College Graduate</td>\n",
       "      <td>0.001515</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        features  importance\n",
       "0                 h1n1_knowledge    0.009057\n",
       "1              opinion_h1n1_risk    0.008636\n",
       "2                  health_worker    0.007019\n",
       "3    opinion_seas_sick_from_vacc    0.006617\n",
       "4              opinion_seas_risk    0.006501\n",
       "5           doctor_recc_seasonal    0.005968\n",
       "6             seasonal_vaccine_0    0.005447\n",
       "7    opinion_seas_vacc_effective    0.004921\n",
       "8    opinion_h1n1_sick_from_vacc    0.004790\n",
       "9               doctor_recc_h1n1    0.004096\n",
       "10   opinion_h1n1_vacc_effective    0.004048\n",
       "11      income_poverty_> $75,000    0.002060\n",
       "12       age_group_55 - 64 Years    0.001846\n",
       "13  employment_industry_fcxhlnwr    0.001813\n",
       "14                  h1n1_concern    0.001785\n",
       "15            seasonal_vaccine_1    0.001682\n",
       "16    education_College Graduate    0.001515"
      ]
     },
     "execution_count": 213,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "imp_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "ax = sns.barplot(x='importance', y='features', data=imp_df, palette=sns.color_palette(\"GnBu\", 10))\n",
    "# plt.xticks(rotation='vertical')\n",
    "ax.grid(b=True, which='major', color='#d3d3d3', linewidth=1.0)\n",
    "ax.grid(b=True, which='minor', color='#d3d3d3', linewidth=0.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CatBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Learning rate set to 0.064751\n",
      "0:\tlearn: 0.4080180\ttotal: 104ms\tremaining: 1m 44s\n",
      "1:\tlearn: 0.3998695\ttotal: 112ms\tremaining: 55.9s\n",
      "2:\tlearn: 0.3924745\ttotal: 120ms\tremaining: 39.8s\n",
      "3:\tlearn: 0.3858070\ttotal: 127ms\tremaining: 31.6s\n",
      "4:\tlearn: 0.3799702\ttotal: 135ms\tremaining: 26.8s\n",
      "5:\tlearn: 0.3748260\ttotal: 143ms\tremaining: 23.6s\n",
      "6:\tlearn: 0.3705012\ttotal: 150ms\tremaining: 21.3s\n",
      "7:\tlearn: 0.3662194\ttotal: 158ms\tremaining: 19.6s\n",
      "8:\tlearn: 0.3624458\ttotal: 167ms\tremaining: 18.4s\n",
      "9:\tlearn: 0.3592650\ttotal: 174ms\tremaining: 17.2s\n",
      "10:\tlearn: 0.3562589\ttotal: 181ms\tremaining: 16.3s\n",
      "11:\tlearn: 0.3536671\ttotal: 189ms\tremaining: 15.6s\n",
      "12:\tlearn: 0.3514061\ttotal: 197ms\tremaining: 14.9s\n",
      "13:\tlearn: 0.3494695\ttotal: 205ms\tremaining: 14.4s\n",
      "14:\tlearn: 0.3476687\ttotal: 212ms\tremaining: 14s\n",
      "15:\tlearn: 0.3458318\ttotal: 220ms\tremaining: 13.5s\n",
      "16:\tlearn: 0.3442407\ttotal: 228ms\tremaining: 13.2s\n",
      "17:\tlearn: 0.3428013\ttotal: 237ms\tremaining: 12.9s\n",
      "18:\tlearn: 0.3416165\ttotal: 252ms\tremaining: 13s\n",
      "19:\tlearn: 0.3405303\ttotal: 260ms\tremaining: 12.7s\n",
      "20:\tlearn: 0.3395350\ttotal: 266ms\tremaining: 12.4s\n",
      "21:\tlearn: 0.3386485\ttotal: 273ms\tremaining: 12.1s\n",
      "22:\tlearn: 0.3378009\ttotal: 277ms\tremaining: 11.8s\n",
      "23:\tlearn: 0.3370339\ttotal: 281ms\tremaining: 11.4s\n",
      "24:\tlearn: 0.3364830\ttotal: 285ms\tremaining: 11.1s\n",
      "25:\tlearn: 0.3358187\ttotal: 289ms\tremaining: 10.8s\n",
      "26:\tlearn: 0.3353028\ttotal: 293ms\tremaining: 10.6s\n",
      "27:\tlearn: 0.3348185\ttotal: 301ms\tremaining: 10.4s\n",
      "28:\tlearn: 0.3345108\ttotal: 307ms\tremaining: 10.3s\n",
      "29:\tlearn: 0.3341159\ttotal: 371ms\tremaining: 12s\n",
      "30:\tlearn: 0.3337276\ttotal: 380ms\tremaining: 11.9s\n",
      "31:\tlearn: 0.3333117\ttotal: 390ms\tremaining: 11.8s\n",
      "32:\tlearn: 0.3329602\ttotal: 399ms\tremaining: 11.7s\n",
      "33:\tlearn: 0.3326538\ttotal: 421ms\tremaining: 12s\n",
      "34:\tlearn: 0.3323453\ttotal: 450ms\tremaining: 12.4s\n",
      "35:\tlearn: 0.3320646\ttotal: 470ms\tremaining: 12.6s\n",
      "36:\tlearn: 0.3317324\ttotal: 477ms\tremaining: 12.4s\n",
      "37:\tlearn: 0.3314640\ttotal: 488ms\tremaining: 12.3s\n",
      "38:\tlearn: 0.3311711\ttotal: 503ms\tremaining: 12.4s\n",
      "39:\tlearn: 0.3309580\ttotal: 527ms\tremaining: 12.6s\n",
      "40:\tlearn: 0.3306913\ttotal: 576ms\tremaining: 13.5s\n",
      "41:\tlearn: 0.3304202\ttotal: 586ms\tremaining: 13.4s\n",
      "42:\tlearn: 0.3302091\ttotal: 599ms\tremaining: 13.3s\n",
      "43:\tlearn: 0.3300473\ttotal: 607ms\tremaining: 13.2s\n",
      "44:\tlearn: 0.3298902\ttotal: 617ms\tremaining: 13.1s\n",
      "45:\tlearn: 0.3297433\ttotal: 625ms\tremaining: 13s\n",
      "46:\tlearn: 0.3296323\ttotal: 631ms\tremaining: 12.8s\n",
      "47:\tlearn: 0.3294373\ttotal: 640ms\tremaining: 12.7s\n",
      "48:\tlearn: 0.3292954\ttotal: 647ms\tremaining: 12.6s\n",
      "49:\tlearn: 0.3292131\ttotal: 656ms\tremaining: 12.5s\n",
      "50:\tlearn: 0.3290331\ttotal: 663ms\tremaining: 12.3s\n",
      "51:\tlearn: 0.3289712\ttotal: 670ms\tremaining: 12.2s\n",
      "52:\tlearn: 0.3288015\ttotal: 676ms\tremaining: 12.1s\n",
      "53:\tlearn: 0.3286659\ttotal: 682ms\tremaining: 11.9s\n",
      "54:\tlearn: 0.3285384\ttotal: 689ms\tremaining: 11.8s\n",
      "55:\tlearn: 0.3284085\ttotal: 696ms\tremaining: 11.7s\n",
      "56:\tlearn: 0.3282327\ttotal: 721ms\tremaining: 11.9s\n",
      "57:\tlearn: 0.3280769\ttotal: 745ms\tremaining: 12.1s\n",
      "58:\tlearn: 0.3278781\ttotal: 768ms\tremaining: 12.3s\n",
      "59:\tlearn: 0.3277234\ttotal: 775ms\tremaining: 12.1s\n",
      "60:\tlearn: 0.3276438\ttotal: 783ms\tremaining: 12.1s\n",
      "61:\tlearn: 0.3275337\ttotal: 789ms\tremaining: 11.9s\n",
      "62:\tlearn: 0.3273765\ttotal: 795ms\tremaining: 11.8s\n",
      "63:\tlearn: 0.3272776\ttotal: 812ms\tremaining: 11.9s\n",
      "64:\tlearn: 0.3271579\ttotal: 817ms\tremaining: 11.8s\n",
      "65:\tlearn: 0.3270382\ttotal: 823ms\tremaining: 11.7s\n",
      "66:\tlearn: 0.3268689\ttotal: 829ms\tremaining: 11.5s\n",
      "67:\tlearn: 0.3267778\ttotal: 833ms\tremaining: 11.4s\n",
      "68:\tlearn: 0.3266642\ttotal: 839ms\tremaining: 11.3s\n",
      "69:\tlearn: 0.3265489\ttotal: 844ms\tremaining: 11.2s\n",
      "70:\tlearn: 0.3264135\ttotal: 851ms\tremaining: 11.1s\n",
      "71:\tlearn: 0.3263206\ttotal: 855ms\tremaining: 11s\n",
      "72:\tlearn: 0.3261937\ttotal: 860ms\tremaining: 10.9s\n",
      "73:\tlearn: 0.3260877\ttotal: 864ms\tremaining: 10.8s\n",
      "74:\tlearn: 0.3260167\ttotal: 869ms\tremaining: 10.7s\n",
      "75:\tlearn: 0.3259106\ttotal: 875ms\tremaining: 10.6s\n",
      "76:\tlearn: 0.3258131\ttotal: 880ms\tremaining: 10.5s\n",
      "77:\tlearn: 0.3257371\ttotal: 890ms\tremaining: 10.5s\n",
      "78:\tlearn: 0.3256333\ttotal: 895ms\tremaining: 10.4s\n",
      "79:\tlearn: 0.3254966\ttotal: 901ms\tremaining: 10.4s\n",
      "80:\tlearn: 0.3254417\ttotal: 905ms\tremaining: 10.3s\n",
      "81:\tlearn: 0.3253697\ttotal: 909ms\tremaining: 10.2s\n",
      "82:\tlearn: 0.3253438\ttotal: 923ms\tremaining: 10.2s\n",
      "83:\tlearn: 0.3252532\ttotal: 937ms\tremaining: 10.2s\n",
      "84:\tlearn: 0.3251622\ttotal: 946ms\tremaining: 10.2s\n",
      "85:\tlearn: 0.3250820\ttotal: 956ms\tremaining: 10.2s\n",
      "86:\tlearn: 0.3250215\ttotal: 962ms\tremaining: 10.1s\n",
      "87:\tlearn: 0.3249210\ttotal: 966ms\tremaining: 10s\n",
      "88:\tlearn: 0.3248490\ttotal: 971ms\tremaining: 9.94s\n",
      "89:\tlearn: 0.3247786\ttotal: 976ms\tremaining: 9.87s\n",
      "90:\tlearn: 0.3246907\ttotal: 980ms\tremaining: 9.79s\n",
      "91:\tlearn: 0.3246213\ttotal: 985ms\tremaining: 9.72s\n",
      "92:\tlearn: 0.3245387\ttotal: 992ms\tremaining: 9.67s\n",
      "93:\tlearn: 0.3244498\ttotal: 996ms\tremaining: 9.6s\n",
      "94:\tlearn: 0.3244054\ttotal: 1000ms\tremaining: 9.52s\n",
      "95:\tlearn: 0.3243267\ttotal: 1.01s\tremaining: 9.48s\n",
      "96:\tlearn: 0.3242557\ttotal: 1.01s\tremaining: 9.44s\n",
      "97:\tlearn: 0.3241476\ttotal: 1.02s\tremaining: 9.37s\n",
      "98:\tlearn: 0.3240801\ttotal: 1.02s\tremaining: 9.31s\n",
      "99:\tlearn: 0.3239911\ttotal: 1.03s\tremaining: 9.24s\n",
      "100:\tlearn: 0.3239373\ttotal: 1.04s\tremaining: 9.25s\n",
      "101:\tlearn: 0.3238636\ttotal: 1.04s\tremaining: 9.19s\n",
      "102:\tlearn: 0.3237321\ttotal: 1.05s\tremaining: 9.12s\n",
      "103:\tlearn: 0.3236732\ttotal: 1.05s\tremaining: 9.06s\n",
      "104:\tlearn: 0.3236437\ttotal: 1.06s\tremaining: 9s\n",
      "105:\tlearn: 0.3235715\ttotal: 1.06s\tremaining: 8.94s\n",
      "106:\tlearn: 0.3235227\ttotal: 1.06s\tremaining: 8.88s\n",
      "107:\tlearn: 0.3234776\ttotal: 1.07s\tremaining: 8.82s\n",
      "108:\tlearn: 0.3234392\ttotal: 1.07s\tremaining: 8.77s\n",
      "109:\tlearn: 0.3233907\ttotal: 1.08s\tremaining: 8.71s\n",
      "110:\tlearn: 0.3232994\ttotal: 1.08s\tremaining: 8.65s\n",
      "111:\tlearn: 0.3232165\ttotal: 1.08s\tremaining: 8.6s\n",
      "112:\tlearn: 0.3231244\ttotal: 1.09s\tremaining: 8.54s\n",
      "113:\tlearn: 0.3230481\ttotal: 1.09s\tremaining: 8.49s\n",
      "114:\tlearn: 0.3229685\ttotal: 1.1s\tremaining: 8.44s\n",
      "115:\tlearn: 0.3228846\ttotal: 1.1s\tremaining: 8.39s\n",
      "116:\tlearn: 0.3227829\ttotal: 1.1s\tremaining: 8.34s\n",
      "117:\tlearn: 0.3227055\ttotal: 1.11s\tremaining: 8.29s\n",
      "118:\tlearn: 0.3226154\ttotal: 1.12s\tremaining: 8.29s\n",
      "119:\tlearn: 0.3225809\ttotal: 1.14s\tremaining: 8.33s\n",
      "120:\tlearn: 0.3225087\ttotal: 1.14s\tremaining: 8.31s\n",
      "121:\tlearn: 0.3224332\ttotal: 1.15s\tremaining: 8.29s\n",
      "122:\tlearn: 0.3224123\ttotal: 1.16s\tremaining: 8.26s\n",
      "123:\tlearn: 0.3223887\ttotal: 1.16s\tremaining: 8.21s\n",
      "124:\tlearn: 0.3223630\ttotal: 1.17s\tremaining: 8.18s\n",
      "125:\tlearn: 0.3223043\ttotal: 1.17s\tremaining: 8.14s\n",
      "126:\tlearn: 0.3222440\ttotal: 1.18s\tremaining: 8.1s\n",
      "127:\tlearn: 0.3221823\ttotal: 1.18s\tremaining: 8.05s\n",
      "128:\tlearn: 0.3221418\ttotal: 1.19s\tremaining: 8.01s\n",
      "129:\tlearn: 0.3220604\ttotal: 1.19s\tremaining: 7.98s\n",
      "130:\tlearn: 0.3220249\ttotal: 1.2s\tremaining: 7.95s\n",
      "131:\tlearn: 0.3219925\ttotal: 1.2s\tremaining: 7.92s\n",
      "132:\tlearn: 0.3218933\ttotal: 1.21s\tremaining: 7.89s\n",
      "133:\tlearn: 0.3218287\ttotal: 1.22s\tremaining: 7.9s\n",
      "134:\tlearn: 0.3217308\ttotal: 1.24s\tremaining: 7.92s\n",
      "135:\tlearn: 0.3216520\ttotal: 1.25s\tremaining: 7.94s\n",
      "136:\tlearn: 0.3216237\ttotal: 1.26s\tremaining: 7.94s\n",
      "137:\tlearn: 0.3215626\ttotal: 1.28s\tremaining: 8.01s\n",
      "138:\tlearn: 0.3214797\ttotal: 1.29s\tremaining: 8.01s\n",
      "139:\tlearn: 0.3214616\ttotal: 1.32s\tremaining: 8.12s\n",
      "140:\tlearn: 0.3214021\ttotal: 1.33s\tremaining: 8.12s\n",
      "141:\tlearn: 0.3213586\ttotal: 1.35s\tremaining: 8.15s\n",
      "142:\tlearn: 0.3212842\ttotal: 1.37s\tremaining: 8.22s\n",
      "143:\tlearn: 0.3212160\ttotal: 1.38s\tremaining: 8.22s\n",
      "144:\tlearn: 0.3211477\ttotal: 1.39s\tremaining: 8.19s\n",
      "145:\tlearn: 0.3211121\ttotal: 1.4s\tremaining: 8.18s\n",
      "146:\tlearn: 0.3210058\ttotal: 1.4s\tremaining: 8.15s\n",
      "147:\tlearn: 0.3209328\ttotal: 1.41s\tremaining: 8.14s\n",
      "148:\tlearn: 0.3208493\ttotal: 1.43s\tremaining: 8.15s\n",
      "149:\tlearn: 0.3207955\ttotal: 1.43s\tremaining: 8.11s\n",
      "150:\tlearn: 0.3206945\ttotal: 1.44s\tremaining: 8.11s\n",
      "151:\tlearn: 0.3206310\ttotal: 1.45s\tremaining: 8.09s\n",
      "152:\tlearn: 0.3205483\ttotal: 1.46s\tremaining: 8.08s\n",
      "153:\tlearn: 0.3204822\ttotal: 1.47s\tremaining: 8.05s\n",
      "154:\tlearn: 0.3204151\ttotal: 1.47s\tremaining: 8.04s\n",
      "155:\tlearn: 0.3203324\ttotal: 1.48s\tremaining: 8.01s\n",
      "156:\tlearn: 0.3202795\ttotal: 1.49s\tremaining: 8.01s\n",
      "157:\tlearn: 0.3202273\ttotal: 1.51s\tremaining: 8.03s\n",
      "158:\tlearn: 0.3201620\ttotal: 1.52s\tremaining: 8.07s\n",
      "159:\tlearn: 0.3201140\ttotal: 1.53s\tremaining: 8.06s\n",
      "160:\tlearn: 0.3200604\ttotal: 1.55s\tremaining: 8.1s\n",
      "161:\tlearn: 0.3200255\ttotal: 1.57s\tremaining: 8.14s\n",
      "162:\tlearn: 0.3199728\ttotal: 1.59s\tremaining: 8.18s\n",
      "163:\tlearn: 0.3199459\ttotal: 1.6s\tremaining: 8.17s\n",
      "164:\tlearn: 0.3198820\ttotal: 1.61s\tremaining: 8.14s\n",
      "165:\tlearn: 0.3198135\ttotal: 1.61s\tremaining: 8.12s\n",
      "166:\tlearn: 0.3197203\ttotal: 1.62s\tremaining: 8.1s\n",
      "167:\tlearn: 0.3196937\ttotal: 1.63s\tremaining: 8.09s\n",
      "168:\tlearn: 0.3196541\ttotal: 1.64s\tremaining: 8.09s\n",
      "169:\tlearn: 0.3195948\ttotal: 1.66s\tremaining: 8.1s\n",
      "170:\tlearn: 0.3195318\ttotal: 1.67s\tremaining: 8.07s\n",
      "171:\tlearn: 0.3194540\ttotal: 1.67s\tremaining: 8.05s\n",
      "172:\tlearn: 0.3193878\ttotal: 1.68s\tremaining: 8.01s\n",
      "173:\tlearn: 0.3193451\ttotal: 1.68s\tremaining: 7.98s\n",
      "174:\tlearn: 0.3192646\ttotal: 1.69s\tremaining: 7.96s\n",
      "175:\tlearn: 0.3191876\ttotal: 1.7s\tremaining: 7.95s\n",
      "176:\tlearn: 0.3191135\ttotal: 1.7s\tremaining: 7.92s\n",
      "177:\tlearn: 0.3190324\ttotal: 1.71s\tremaining: 7.89s\n",
      "178:\tlearn: 0.3189666\ttotal: 1.71s\tremaining: 7.85s\n",
      "179:\tlearn: 0.3189378\ttotal: 1.72s\tremaining: 7.84s\n",
      "180:\tlearn: 0.3189156\ttotal: 1.73s\tremaining: 7.82s\n",
      "181:\tlearn: 0.3188388\ttotal: 1.74s\tremaining: 7.8s\n",
      "182:\tlearn: 0.3187457\ttotal: 1.75s\tremaining: 7.79s\n",
      "183:\tlearn: 0.3186802\ttotal: 1.76s\tremaining: 7.8s\n",
      "184:\tlearn: 0.3186253\ttotal: 1.77s\tremaining: 7.8s\n",
      "185:\tlearn: 0.3186019\ttotal: 1.77s\tremaining: 7.77s\n",
      "186:\tlearn: 0.3185570\ttotal: 1.78s\tremaining: 7.75s\n",
      "187:\tlearn: 0.3185107\ttotal: 1.79s\tremaining: 7.74s\n",
      "188:\tlearn: 0.3184691\ttotal: 1.8s\tremaining: 7.72s\n",
      "189:\tlearn: 0.3184351\ttotal: 1.8s\tremaining: 7.7s\n",
      "190:\tlearn: 0.3184162\ttotal: 1.81s\tremaining: 7.67s\n",
      "191:\tlearn: 0.3183782\ttotal: 1.82s\tremaining: 7.65s\n",
      "192:\tlearn: 0.3183014\ttotal: 1.82s\tremaining: 7.62s\n",
      "193:\tlearn: 0.3182960\ttotal: 1.83s\tremaining: 7.61s\n",
      "194:\tlearn: 0.3182286\ttotal: 1.84s\tremaining: 7.59s\n",
      "195:\tlearn: 0.3181769\ttotal: 1.84s\tremaining: 7.56s\n",
      "196:\tlearn: 0.3181253\ttotal: 1.85s\tremaining: 7.53s\n",
      "197:\tlearn: 0.3181188\ttotal: 1.85s\tremaining: 7.51s\n",
      "198:\tlearn: 0.3181154\ttotal: 1.86s\tremaining: 7.48s\n",
      "199:\tlearn: 0.3181128\ttotal: 1.86s\tremaining: 7.44s\n",
      "200:\tlearn: 0.3180838\ttotal: 1.87s\tremaining: 7.42s\n",
      "201:\tlearn: 0.3180075\ttotal: 1.87s\tremaining: 7.39s\n",
      "202:\tlearn: 0.3179622\ttotal: 1.88s\tremaining: 7.37s\n",
      "203:\tlearn: 0.3179084\ttotal: 1.88s\tremaining: 7.35s\n",
      "204:\tlearn: 0.3178519\ttotal: 1.89s\tremaining: 7.32s\n",
      "205:\tlearn: 0.3178493\ttotal: 1.89s\tremaining: 7.29s\n",
      "206:\tlearn: 0.3178298\ttotal: 1.9s\tremaining: 7.27s\n",
      "207:\tlearn: 0.3178070\ttotal: 1.9s\tremaining: 7.24s\n",
      "208:\tlearn: 0.3177844\ttotal: 1.91s\tremaining: 7.23s\n",
      "209:\tlearn: 0.3177613\ttotal: 1.92s\tremaining: 7.23s\n",
      "210:\tlearn: 0.3176924\ttotal: 1.93s\tremaining: 7.23s\n",
      "211:\tlearn: 0.3176220\ttotal: 1.94s\tremaining: 7.22s\n",
      "212:\tlearn: 0.3176186\ttotal: 1.95s\tremaining: 7.2s\n",
      "213:\tlearn: 0.3175469\ttotal: 1.96s\tremaining: 7.19s\n",
      "214:\tlearn: 0.3174757\ttotal: 1.96s\tremaining: 7.17s\n",
      "215:\tlearn: 0.3174368\ttotal: 1.97s\tremaining: 7.15s\n",
      "216:\tlearn: 0.3174319\ttotal: 1.98s\tremaining: 7.13s\n",
      "217:\tlearn: 0.3173851\ttotal: 1.98s\tremaining: 7.1s\n",
      "218:\tlearn: 0.3173812\ttotal: 1.98s\tremaining: 7.08s\n",
      "219:\tlearn: 0.3173679\ttotal: 1.99s\tremaining: 7.05s\n",
      "220:\tlearn: 0.3173047\ttotal: 1.99s\tremaining: 7.02s\n",
      "221:\tlearn: 0.3172898\ttotal: 2s\tremaining: 7s\n",
      "222:\tlearn: 0.3172385\ttotal: 2s\tremaining: 6.97s\n",
      "223:\tlearn: 0.3172000\ttotal: 2s\tremaining: 6.95s\n",
      "224:\tlearn: 0.3171453\ttotal: 2.01s\tremaining: 6.92s\n",
      "225:\tlearn: 0.3171389\ttotal: 2.02s\tremaining: 6.9s\n",
      "226:\tlearn: 0.3171363\ttotal: 2.02s\tremaining: 6.88s\n",
      "227:\tlearn: 0.3171301\ttotal: 2.02s\tremaining: 6.85s\n",
      "228:\tlearn: 0.3170950\ttotal: 2.03s\tremaining: 6.82s\n",
      "229:\tlearn: 0.3170849\ttotal: 2.03s\tremaining: 6.8s\n",
      "230:\tlearn: 0.3170099\ttotal: 2.04s\tremaining: 6.78s\n",
      "231:\tlearn: 0.3169893\ttotal: 2.04s\tremaining: 6.75s\n",
      "232:\tlearn: 0.3169859\ttotal: 2.04s\tremaining: 6.72s\n",
      "233:\tlearn: 0.3169835\ttotal: 2.05s\tremaining: 6.7s\n",
      "234:\tlearn: 0.3169813\ttotal: 2.05s\tremaining: 6.68s\n",
      "235:\tlearn: 0.3169799\ttotal: 2.06s\tremaining: 6.65s\n",
      "236:\tlearn: 0.3169705\ttotal: 2.06s\tremaining: 6.63s\n",
      "237:\tlearn: 0.3169637\ttotal: 2.06s\tremaining: 6.6s\n",
      "238:\tlearn: 0.3169620\ttotal: 2.07s\tremaining: 6.58s\n",
      "239:\tlearn: 0.3169529\ttotal: 2.07s\tremaining: 6.55s\n",
      "240:\tlearn: 0.3169476\ttotal: 2.07s\tremaining: 6.53s\n",
      "241:\tlearn: 0.3169460\ttotal: 2.08s\tremaining: 6.51s\n",
      "242:\tlearn: 0.3168976\ttotal: 2.08s\tremaining: 6.49s\n",
      "243:\tlearn: 0.3168686\ttotal: 2.09s\tremaining: 6.46s\n",
      "244:\tlearn: 0.3168665\ttotal: 2.09s\tremaining: 6.44s\n",
      "245:\tlearn: 0.3168085\ttotal: 2.09s\tremaining: 6.42s\n",
      "246:\tlearn: 0.3167813\ttotal: 2.1s\tremaining: 6.41s\n",
      "247:\tlearn: 0.3167544\ttotal: 2.12s\tremaining: 6.43s\n",
      "248:\tlearn: 0.3167528\ttotal: 2.13s\tremaining: 6.42s\n",
      "249:\tlearn: 0.3167314\ttotal: 2.14s\tremaining: 6.42s\n",
      "250:\tlearn: 0.3166967\ttotal: 2.15s\tremaining: 6.41s\n",
      "251:\tlearn: 0.3166578\ttotal: 2.16s\tremaining: 6.41s\n",
      "252:\tlearn: 0.3166337\ttotal: 2.17s\tremaining: 6.41s\n",
      "253:\tlearn: 0.3166320\ttotal: 2.18s\tremaining: 6.4s\n",
      "254:\tlearn: 0.3166250\ttotal: 2.19s\tremaining: 6.39s\n",
      "255:\tlearn: 0.3165833\ttotal: 2.19s\tremaining: 6.38s\n",
      "256:\tlearn: 0.3165750\ttotal: 2.2s\tremaining: 6.37s\n",
      "257:\tlearn: 0.3165108\ttotal: 2.21s\tremaining: 6.37s\n",
      "258:\tlearn: 0.3165052\ttotal: 2.22s\tremaining: 6.36s\n",
      "259:\tlearn: 0.3165037\ttotal: 2.23s\tremaining: 6.35s\n",
      "260:\tlearn: 0.3164969\ttotal: 2.24s\tremaining: 6.35s\n",
      "261:\tlearn: 0.3164938\ttotal: 2.25s\tremaining: 6.34s\n",
      "262:\tlearn: 0.3164905\ttotal: 2.26s\tremaining: 6.33s\n",
      "263:\tlearn: 0.3164863\ttotal: 2.34s\tremaining: 6.54s\n",
      "264:\tlearn: 0.3164486\ttotal: 2.36s\tremaining: 6.54s\n",
      "265:\tlearn: 0.3164085\ttotal: 2.37s\tremaining: 6.53s\n",
      "266:\tlearn: 0.3164056\ttotal: 2.37s\tremaining: 6.51s\n",
      "267:\tlearn: 0.3163946\ttotal: 2.38s\tremaining: 6.49s\n",
      "268:\tlearn: 0.3163895\ttotal: 2.38s\tremaining: 6.47s\n",
      "269:\tlearn: 0.3163862\ttotal: 2.38s\tremaining: 6.45s\n",
      "270:\tlearn: 0.3163522\ttotal: 2.39s\tremaining: 6.42s\n",
      "271:\tlearn: 0.3163491\ttotal: 2.39s\tremaining: 6.41s\n",
      "272:\tlearn: 0.3163292\ttotal: 2.41s\tremaining: 6.41s\n",
      "273:\tlearn: 0.3163057\ttotal: 2.41s\tremaining: 6.39s\n",
      "274:\tlearn: 0.3163037\ttotal: 2.42s\tremaining: 6.38s\n",
      "275:\tlearn: 0.3162634\ttotal: 2.42s\tremaining: 6.36s\n",
      "276:\tlearn: 0.3161954\ttotal: 2.43s\tremaining: 6.34s\n",
      "277:\tlearn: 0.3161481\ttotal: 2.44s\tremaining: 6.32s\n",
      "278:\tlearn: 0.3161454\ttotal: 2.44s\tremaining: 6.31s\n",
      "279:\tlearn: 0.3160874\ttotal: 2.44s\tremaining: 6.29s\n",
      "280:\tlearn: 0.3160782\ttotal: 2.45s\tremaining: 6.26s\n",
      "281:\tlearn: 0.3160709\ttotal: 2.45s\tremaining: 6.25s\n",
      "282:\tlearn: 0.3160268\ttotal: 2.46s\tremaining: 6.23s\n",
      "283:\tlearn: 0.3160259\ttotal: 2.46s\tremaining: 6.21s\n",
      "284:\tlearn: 0.3160203\ttotal: 2.47s\tremaining: 6.19s\n",
      "285:\tlearn: 0.3160128\ttotal: 2.47s\tremaining: 6.17s\n",
      "286:\tlearn: 0.3160093\ttotal: 2.47s\tremaining: 6.15s\n",
      "287:\tlearn: 0.3159742\ttotal: 2.48s\tremaining: 6.13s\n",
      "288:\tlearn: 0.3159452\ttotal: 2.48s\tremaining: 6.11s\n",
      "289:\tlearn: 0.3159088\ttotal: 2.49s\tremaining: 6.09s\n",
      "290:\tlearn: 0.3159079\ttotal: 2.49s\tremaining: 6.07s\n",
      "291:\tlearn: 0.3159023\ttotal: 2.5s\tremaining: 6.05s\n",
      "292:\tlearn: 0.3158995\ttotal: 2.5s\tremaining: 6.04s\n",
      "293:\tlearn: 0.3158968\ttotal: 2.51s\tremaining: 6.03s\n",
      "294:\tlearn: 0.3158960\ttotal: 2.52s\tremaining: 6.01s\n",
      "295:\tlearn: 0.3158854\ttotal: 2.52s\tremaining: 5.99s\n",
      "296:\tlearn: 0.3158835\ttotal: 2.52s\tremaining: 5.97s\n",
      "297:\tlearn: 0.3158773\ttotal: 2.53s\tremaining: 5.97s\n",
      "298:\tlearn: 0.3158738\ttotal: 2.54s\tremaining: 5.97s\n",
      "299:\tlearn: 0.3158598\ttotal: 2.55s\tremaining: 5.95s\n",
      "300:\tlearn: 0.3158139\ttotal: 2.56s\tremaining: 5.93s\n",
      "301:\tlearn: 0.3158087\ttotal: 2.56s\tremaining: 5.92s\n",
      "302:\tlearn: 0.3158081\ttotal: 2.56s\tremaining: 5.9s\n",
      "303:\tlearn: 0.3157845\ttotal: 2.57s\tremaining: 5.88s\n",
      "304:\tlearn: 0.3157695\ttotal: 2.57s\tremaining: 5.86s\n",
      "305:\tlearn: 0.3157513\ttotal: 2.58s\tremaining: 5.85s\n",
      "306:\tlearn: 0.3157127\ttotal: 2.58s\tremaining: 5.83s\n",
      "307:\tlearn: 0.3156585\ttotal: 2.59s\tremaining: 5.82s\n",
      "308:\tlearn: 0.3156521\ttotal: 2.59s\tremaining: 5.8s\n",
      "309:\tlearn: 0.3156296\ttotal: 2.6s\tremaining: 5.79s\n",
      "310:\tlearn: 0.3156261\ttotal: 2.61s\tremaining: 5.77s\n",
      "311:\tlearn: 0.3156201\ttotal: 2.61s\tremaining: 5.76s\n",
      "312:\tlearn: 0.3156062\ttotal: 2.62s\tremaining: 5.75s\n",
      "313:\tlearn: 0.3156046\ttotal: 2.62s\tremaining: 5.73s\n",
      "314:\tlearn: 0.3155944\ttotal: 2.63s\tremaining: 5.72s\n",
      "315:\tlearn: 0.3155904\ttotal: 2.63s\tremaining: 5.7s\n",
      "316:\tlearn: 0.3155899\ttotal: 2.64s\tremaining: 5.69s\n",
      "317:\tlearn: 0.3155882\ttotal: 2.65s\tremaining: 5.68s\n",
      "318:\tlearn: 0.3155879\ttotal: 2.65s\tremaining: 5.66s\n",
      "319:\tlearn: 0.3155846\ttotal: 2.66s\tremaining: 5.65s\n",
      "320:\tlearn: 0.3155791\ttotal: 2.66s\tremaining: 5.63s\n",
      "321:\tlearn: 0.3155257\ttotal: 2.67s\tremaining: 5.62s\n",
      "322:\tlearn: 0.3155140\ttotal: 2.67s\tremaining: 5.61s\n",
      "323:\tlearn: 0.3155010\ttotal: 2.68s\tremaining: 5.6s\n",
      "324:\tlearn: 0.3154938\ttotal: 2.69s\tremaining: 5.58s\n",
      "325:\tlearn: 0.3154513\ttotal: 2.7s\tremaining: 5.58s\n",
      "326:\tlearn: 0.3154212\ttotal: 2.71s\tremaining: 5.57s\n",
      "327:\tlearn: 0.3153979\ttotal: 2.71s\tremaining: 5.56s\n",
      "328:\tlearn: 0.3153747\ttotal: 2.72s\tremaining: 5.55s\n",
      "329:\tlearn: 0.3153540\ttotal: 2.73s\tremaining: 5.55s\n",
      "330:\tlearn: 0.3152906\ttotal: 2.74s\tremaining: 5.54s\n",
      "331:\tlearn: 0.3152861\ttotal: 2.75s\tremaining: 5.53s\n",
      "332:\tlearn: 0.3152819\ttotal: 2.75s\tremaining: 5.52s\n",
      "333:\tlearn: 0.3152813\ttotal: 2.76s\tremaining: 5.51s\n",
      "334:\tlearn: 0.3152463\ttotal: 2.78s\tremaining: 5.51s\n",
      "335:\tlearn: 0.3152376\ttotal: 2.79s\tremaining: 5.51s\n",
      "336:\tlearn: 0.3152190\ttotal: 2.79s\tremaining: 5.5s\n",
      "337:\tlearn: 0.3152185\ttotal: 2.8s\tremaining: 5.49s\n",
      "338:\tlearn: 0.3152180\ttotal: 2.81s\tremaining: 5.48s\n",
      "339:\tlearn: 0.3151623\ttotal: 2.81s\tremaining: 5.46s\n",
      "340:\tlearn: 0.3151621\ttotal: 2.82s\tremaining: 5.45s\n",
      "341:\tlearn: 0.3151461\ttotal: 2.83s\tremaining: 5.45s\n",
      "342:\tlearn: 0.3151443\ttotal: 2.84s\tremaining: 5.44s\n",
      "343:\tlearn: 0.3151348\ttotal: 2.85s\tremaining: 5.43s\n",
      "344:\tlearn: 0.3151247\ttotal: 2.85s\tremaining: 5.41s\n",
      "345:\tlearn: 0.3151229\ttotal: 2.86s\tremaining: 5.4s\n",
      "346:\tlearn: 0.3151207\ttotal: 2.86s\tremaining: 5.39s\n",
      "347:\tlearn: 0.3151082\ttotal: 2.87s\tremaining: 5.37s\n",
      "348:\tlearn: 0.3151077\ttotal: 2.87s\tremaining: 5.36s\n",
      "349:\tlearn: 0.3151073\ttotal: 2.88s\tremaining: 5.35s\n",
      "350:\tlearn: 0.3151025\ttotal: 2.89s\tremaining: 5.34s\n",
      "351:\tlearn: 0.3150994\ttotal: 2.9s\tremaining: 5.33s\n",
      "352:\tlearn: 0.3150588\ttotal: 2.9s\tremaining: 5.32s\n",
      "353:\tlearn: 0.3150563\ttotal: 2.91s\tremaining: 5.31s\n",
      "354:\tlearn: 0.3150488\ttotal: 2.92s\tremaining: 5.3s\n",
      "355:\tlearn: 0.3150373\ttotal: 2.92s\tremaining: 5.29s\n",
      "356:\tlearn: 0.3150311\ttotal: 2.93s\tremaining: 5.29s\n",
      "357:\tlearn: 0.3149894\ttotal: 2.94s\tremaining: 5.27s\n",
      "358:\tlearn: 0.3149872\ttotal: 2.94s\tremaining: 5.26s\n",
      "359:\tlearn: 0.3149491\ttotal: 2.95s\tremaining: 5.25s\n",
      "360:\tlearn: 0.3149453\ttotal: 2.96s\tremaining: 5.23s\n",
      "361:\tlearn: 0.3149439\ttotal: 2.96s\tremaining: 5.22s\n",
      "362:\tlearn: 0.3149426\ttotal: 2.97s\tremaining: 5.21s\n",
      "363:\tlearn: 0.3149390\ttotal: 2.97s\tremaining: 5.19s\n",
      "364:\tlearn: 0.3148864\ttotal: 2.98s\tremaining: 5.18s\n",
      "365:\tlearn: 0.3148576\ttotal: 2.98s\tremaining: 5.17s\n",
      "366:\tlearn: 0.3148453\ttotal: 2.99s\tremaining: 5.16s\n",
      "367:\tlearn: 0.3148330\ttotal: 3s\tremaining: 5.15s\n",
      "368:\tlearn: 0.3148239\ttotal: 3s\tremaining: 5.13s\n",
      "369:\tlearn: 0.3148208\ttotal: 3.01s\tremaining: 5.12s\n",
      "370:\tlearn: 0.3148073\ttotal: 3.01s\tremaining: 5.11s\n",
      "371:\tlearn: 0.3148047\ttotal: 3.02s\tremaining: 5.1s\n",
      "372:\tlearn: 0.3147723\ttotal: 3.03s\tremaining: 5.09s\n",
      "373:\tlearn: 0.3147643\ttotal: 3.03s\tremaining: 5.07s\n",
      "374:\tlearn: 0.3147627\ttotal: 3.04s\tremaining: 5.06s\n",
      "375:\tlearn: 0.3147611\ttotal: 3.04s\tremaining: 5.05s\n",
      "376:\tlearn: 0.3147552\ttotal: 3.05s\tremaining: 5.04s\n",
      "377:\tlearn: 0.3147373\ttotal: 3.05s\tremaining: 5.03s\n",
      "378:\tlearn: 0.3147369\ttotal: 3.06s\tremaining: 5.01s\n",
      "379:\tlearn: 0.3147033\ttotal: 3.07s\tremaining: 5s\n",
      "380:\tlearn: 0.3146551\ttotal: 3.07s\tremaining: 4.99s\n",
      "381:\tlearn: 0.3146548\ttotal: 3.08s\tremaining: 4.97s\n",
      "382:\tlearn: 0.3146362\ttotal: 3.08s\tremaining: 4.96s\n",
      "383:\tlearn: 0.3145838\ttotal: 3.09s\tremaining: 4.96s\n",
      "384:\tlearn: 0.3145461\ttotal: 3.09s\tremaining: 4.94s\n",
      "385:\tlearn: 0.3145460\ttotal: 3.1s\tremaining: 4.93s\n",
      "386:\tlearn: 0.3145434\ttotal: 3.11s\tremaining: 4.92s\n",
      "387:\tlearn: 0.3145366\ttotal: 3.11s\tremaining: 4.91s\n",
      "388:\tlearn: 0.3145345\ttotal: 3.12s\tremaining: 4.9s\n",
      "389:\tlearn: 0.3145000\ttotal: 3.13s\tremaining: 4.89s\n",
      "390:\tlearn: 0.3144929\ttotal: 3.13s\tremaining: 4.88s\n",
      "391:\tlearn: 0.3144904\ttotal: 3.14s\tremaining: 4.87s\n",
      "392:\tlearn: 0.3144878\ttotal: 3.14s\tremaining: 4.86s\n",
      "393:\tlearn: 0.3144800\ttotal: 3.15s\tremaining: 4.85s\n",
      "394:\tlearn: 0.3144744\ttotal: 3.16s\tremaining: 4.84s\n",
      "395:\tlearn: 0.3144340\ttotal: 3.16s\tremaining: 4.82s\n",
      "396:\tlearn: 0.3144303\ttotal: 3.17s\tremaining: 4.81s\n",
      "397:\tlearn: 0.3144239\ttotal: 3.17s\tremaining: 4.8s\n",
      "398:\tlearn: 0.3144191\ttotal: 3.18s\tremaining: 4.79s\n",
      "399:\tlearn: 0.3144189\ttotal: 3.19s\tremaining: 4.78s\n",
      "400:\tlearn: 0.3144027\ttotal: 3.19s\tremaining: 4.77s\n",
      "401:\tlearn: 0.3143084\ttotal: 3.22s\tremaining: 4.78s\n",
      "402:\tlearn: 0.3143009\ttotal: 3.24s\tremaining: 4.79s\n",
      "403:\tlearn: 0.3142541\ttotal: 3.24s\tremaining: 4.78s\n",
      "404:\tlearn: 0.3142476\ttotal: 3.25s\tremaining: 4.77s\n",
      "405:\tlearn: 0.3142292\ttotal: 3.25s\tremaining: 4.75s\n",
      "406:\tlearn: 0.3142228\ttotal: 3.25s\tremaining: 4.74s\n",
      "407:\tlearn: 0.3142152\ttotal: 3.26s\tremaining: 4.73s\n",
      "408:\tlearn: 0.3142109\ttotal: 3.26s\tremaining: 4.72s\n",
      "409:\tlearn: 0.3142106\ttotal: 3.27s\tremaining: 4.7s\n",
      "410:\tlearn: 0.3142045\ttotal: 3.27s\tremaining: 4.69s\n",
      "411:\tlearn: 0.3141405\ttotal: 3.28s\tremaining: 4.68s\n",
      "412:\tlearn: 0.3140771\ttotal: 3.29s\tremaining: 4.67s\n",
      "413:\tlearn: 0.3140768\ttotal: 3.29s\tremaining: 4.66s\n",
      "414:\tlearn: 0.3140482\ttotal: 3.29s\tremaining: 4.64s\n",
      "415:\tlearn: 0.3140430\ttotal: 3.3s\tremaining: 4.63s\n",
      "416:\tlearn: 0.3140408\ttotal: 3.3s\tremaining: 4.62s\n",
      "417:\tlearn: 0.3140393\ttotal: 3.31s\tremaining: 4.61s\n",
      "418:\tlearn: 0.3140325\ttotal: 3.32s\tremaining: 4.6s\n",
      "419:\tlearn: 0.3140252\ttotal: 3.33s\tremaining: 4.6s\n",
      "420:\tlearn: 0.3140221\ttotal: 3.34s\tremaining: 4.59s\n",
      "421:\tlearn: 0.3140114\ttotal: 3.34s\tremaining: 4.58s\n",
      "422:\tlearn: 0.3139955\ttotal: 3.35s\tremaining: 4.57s\n",
      "423:\tlearn: 0.3139917\ttotal: 3.36s\tremaining: 4.57s\n",
      "424:\tlearn: 0.3139904\ttotal: 3.37s\tremaining: 4.56s\n",
      "425:\tlearn: 0.3139877\ttotal: 3.38s\tremaining: 4.56s\n",
      "426:\tlearn: 0.3139752\ttotal: 3.39s\tremaining: 4.55s\n",
      "427:\tlearn: 0.3139514\ttotal: 3.4s\tremaining: 4.54s\n",
      "428:\tlearn: 0.3139494\ttotal: 3.41s\tremaining: 4.54s\n",
      "429:\tlearn: 0.3139146\ttotal: 3.41s\tremaining: 4.53s\n",
      "430:\tlearn: 0.3138857\ttotal: 3.42s\tremaining: 4.51s\n",
      "431:\tlearn: 0.3138676\ttotal: 3.42s\tremaining: 4.5s\n",
      "432:\tlearn: 0.3138440\ttotal: 3.43s\tremaining: 4.49s\n",
      "433:\tlearn: 0.3138404\ttotal: 3.43s\tremaining: 4.47s\n",
      "434:\tlearn: 0.3137984\ttotal: 3.43s\tremaining: 4.46s\n",
      "435:\tlearn: 0.3137831\ttotal: 3.44s\tremaining: 4.45s\n",
      "436:\tlearn: 0.3137455\ttotal: 3.44s\tremaining: 4.44s\n",
      "437:\tlearn: 0.3137454\ttotal: 3.45s\tremaining: 4.42s\n",
      "438:\tlearn: 0.3137454\ttotal: 3.45s\tremaining: 4.41s\n",
      "439:\tlearn: 0.3137453\ttotal: 3.46s\tremaining: 4.4s\n",
      "440:\tlearn: 0.3137428\ttotal: 3.46s\tremaining: 4.39s\n",
      "441:\tlearn: 0.3137427\ttotal: 3.47s\tremaining: 4.38s\n",
      "442:\tlearn: 0.3137070\ttotal: 3.47s\tremaining: 4.37s\n",
      "443:\tlearn: 0.3137069\ttotal: 3.48s\tremaining: 4.36s\n",
      "444:\tlearn: 0.3137018\ttotal: 3.48s\tremaining: 4.34s\n",
      "445:\tlearn: 0.3136349\ttotal: 3.49s\tremaining: 4.34s\n",
      "446:\tlearn: 0.3135941\ttotal: 3.5s\tremaining: 4.33s\n",
      "447:\tlearn: 0.3135451\ttotal: 3.5s\tremaining: 4.32s\n",
      "448:\tlearn: 0.3135170\ttotal: 3.51s\tremaining: 4.31s\n",
      "449:\tlearn: 0.3134877\ttotal: 3.52s\tremaining: 4.3s\n",
      "450:\tlearn: 0.3134875\ttotal: 3.52s\tremaining: 4.29s\n",
      "451:\tlearn: 0.3134822\ttotal: 3.53s\tremaining: 4.28s\n",
      "452:\tlearn: 0.3134803\ttotal: 3.53s\tremaining: 4.27s\n",
      "453:\tlearn: 0.3134794\ttotal: 3.54s\tremaining: 4.26s\n",
      "454:\tlearn: 0.3134385\ttotal: 3.54s\tremaining: 4.25s\n",
      "455:\tlearn: 0.3134383\ttotal: 3.55s\tremaining: 4.23s\n",
      "456:\tlearn: 0.3134354\ttotal: 3.55s\tremaining: 4.22s\n",
      "457:\tlearn: 0.3134346\ttotal: 3.56s\tremaining: 4.21s\n",
      "458:\tlearn: 0.3134332\ttotal: 3.57s\tremaining: 4.2s\n",
      "459:\tlearn: 0.3133991\ttotal: 3.57s\tremaining: 4.19s\n",
      "460:\tlearn: 0.3133717\ttotal: 3.58s\tremaining: 4.19s\n",
      "461:\tlearn: 0.3133681\ttotal: 3.59s\tremaining: 4.18s\n",
      "462:\tlearn: 0.3133305\ttotal: 3.59s\tremaining: 4.17s\n",
      "463:\tlearn: 0.3133253\ttotal: 3.6s\tremaining: 4.16s\n",
      "464:\tlearn: 0.3133200\ttotal: 3.6s\tremaining: 4.14s\n",
      "465:\tlearn: 0.3133163\ttotal: 3.61s\tremaining: 4.14s\n",
      "466:\tlearn: 0.3132570\ttotal: 3.61s\tremaining: 4.13s\n",
      "467:\tlearn: 0.3132530\ttotal: 3.62s\tremaining: 4.12s\n",
      "468:\tlearn: 0.3132490\ttotal: 3.63s\tremaining: 4.1s\n",
      "469:\tlearn: 0.3132222\ttotal: 3.63s\tremaining: 4.1s\n",
      "470:\tlearn: 0.3132208\ttotal: 3.64s\tremaining: 4.08s\n",
      "471:\tlearn: 0.3131645\ttotal: 3.64s\tremaining: 4.08s\n",
      "472:\tlearn: 0.3131440\ttotal: 3.65s\tremaining: 4.07s\n",
      "473:\tlearn: 0.3130946\ttotal: 3.65s\tremaining: 4.05s\n",
      "474:\tlearn: 0.3130945\ttotal: 3.66s\tremaining: 4.04s\n",
      "475:\tlearn: 0.3130943\ttotal: 3.67s\tremaining: 4.04s\n",
      "476:\tlearn: 0.3130885\ttotal: 3.67s\tremaining: 4.03s\n",
      "477:\tlearn: 0.3130885\ttotal: 3.68s\tremaining: 4.02s\n",
      "478:\tlearn: 0.3130884\ttotal: 3.68s\tremaining: 4s\n",
      "479:\tlearn: 0.3130873\ttotal: 3.69s\tremaining: 4s\n",
      "480:\tlearn: 0.3130840\ttotal: 3.7s\tremaining: 3.99s\n",
      "481:\tlearn: 0.3130643\ttotal: 3.7s\tremaining: 3.98s\n",
      "482:\tlearn: 0.3130620\ttotal: 3.71s\tremaining: 3.97s\n",
      "483:\tlearn: 0.3130312\ttotal: 3.72s\tremaining: 3.96s\n",
      "484:\tlearn: 0.3129715\ttotal: 3.72s\tremaining: 3.95s\n",
      "485:\tlearn: 0.3129714\ttotal: 3.73s\tremaining: 3.94s\n",
      "486:\tlearn: 0.3129443\ttotal: 3.73s\tremaining: 3.93s\n",
      "487:\tlearn: 0.3128831\ttotal: 3.74s\tremaining: 3.92s\n",
      "488:\tlearn: 0.3128808\ttotal: 3.74s\tremaining: 3.91s\n",
      "489:\tlearn: 0.3128570\ttotal: 3.75s\tremaining: 3.9s\n",
      "490:\tlearn: 0.3128357\ttotal: 3.76s\tremaining: 3.89s\n",
      "491:\tlearn: 0.3127958\ttotal: 3.76s\tremaining: 3.88s\n",
      "492:\tlearn: 0.3127951\ttotal: 3.77s\tremaining: 3.87s\n",
      "493:\tlearn: 0.3127777\ttotal: 3.77s\tremaining: 3.87s\n",
      "494:\tlearn: 0.3127316\ttotal: 3.78s\tremaining: 3.85s\n",
      "495:\tlearn: 0.3127028\ttotal: 3.79s\tremaining: 3.85s\n",
      "496:\tlearn: 0.3126584\ttotal: 3.79s\tremaining: 3.84s\n",
      "497:\tlearn: 0.3126526\ttotal: 3.8s\tremaining: 3.83s\n",
      "498:\tlearn: 0.3126229\ttotal: 3.8s\tremaining: 3.82s\n",
      "499:\tlearn: 0.3125930\ttotal: 3.81s\tremaining: 3.81s\n",
      "500:\tlearn: 0.3125769\ttotal: 3.81s\tremaining: 3.8s\n",
      "501:\tlearn: 0.3125554\ttotal: 3.82s\tremaining: 3.79s\n",
      "502:\tlearn: 0.3125507\ttotal: 3.83s\tremaining: 3.78s\n",
      "503:\tlearn: 0.3125318\ttotal: 3.83s\tremaining: 3.77s\n",
      "504:\tlearn: 0.3125075\ttotal: 3.84s\tremaining: 3.76s\n",
      "505:\tlearn: 0.3124726\ttotal: 3.84s\tremaining: 3.75s\n",
      "506:\tlearn: 0.3124501\ttotal: 3.85s\tremaining: 3.74s\n",
      "507:\tlearn: 0.3123984\ttotal: 3.85s\tremaining: 3.73s\n",
      "508:\tlearn: 0.3123445\ttotal: 3.86s\tremaining: 3.72s\n",
      "509:\tlearn: 0.3123403\ttotal: 3.87s\tremaining: 3.71s\n",
      "510:\tlearn: 0.3122893\ttotal: 3.89s\tremaining: 3.72s\n",
      "511:\tlearn: 0.3122180\ttotal: 3.89s\tremaining: 3.71s\n",
      "512:\tlearn: 0.3122156\ttotal: 3.9s\tremaining: 3.7s\n",
      "513:\tlearn: 0.3122127\ttotal: 3.9s\tremaining: 3.69s\n",
      "514:\tlearn: 0.3122094\ttotal: 3.91s\tremaining: 3.68s\n",
      "515:\tlearn: 0.3121893\ttotal: 3.92s\tremaining: 3.67s\n",
      "516:\tlearn: 0.3121506\ttotal: 3.92s\tremaining: 3.67s\n",
      "517:\tlearn: 0.3121481\ttotal: 3.93s\tremaining: 3.66s\n",
      "518:\tlearn: 0.3121122\ttotal: 3.94s\tremaining: 3.65s\n",
      "519:\tlearn: 0.3120609\ttotal: 3.94s\tremaining: 3.64s\n",
      "520:\tlearn: 0.3120058\ttotal: 3.95s\tremaining: 3.63s\n",
      "521:\tlearn: 0.3119964\ttotal: 3.95s\tremaining: 3.62s\n",
      "522:\tlearn: 0.3119524\ttotal: 3.96s\tremaining: 3.61s\n",
      "523:\tlearn: 0.3119469\ttotal: 3.97s\tremaining: 3.6s\n",
      "524:\tlearn: 0.3119314\ttotal: 3.97s\tremaining: 3.6s\n",
      "525:\tlearn: 0.3119263\ttotal: 3.98s\tremaining: 3.58s\n",
      "526:\tlearn: 0.3119225\ttotal: 3.98s\tremaining: 3.58s\n",
      "527:\tlearn: 0.3119160\ttotal: 3.99s\tremaining: 3.57s\n",
      "528:\tlearn: 0.3119147\ttotal: 4s\tremaining: 3.56s\n",
      "529:\tlearn: 0.3119130\ttotal: 4s\tremaining: 3.55s\n",
      "530:\tlearn: 0.3118900\ttotal: 4.01s\tremaining: 3.54s\n",
      "531:\tlearn: 0.3118622\ttotal: 4.01s\tremaining: 3.53s\n",
      "532:\tlearn: 0.3118439\ttotal: 4.02s\tremaining: 3.52s\n",
      "533:\tlearn: 0.3118223\ttotal: 4.03s\tremaining: 3.52s\n",
      "534:\tlearn: 0.3118191\ttotal: 4.03s\tremaining: 3.51s\n",
      "535:\tlearn: 0.3118160\ttotal: 4.04s\tremaining: 3.5s\n",
      "536:\tlearn: 0.3118149\ttotal: 4.04s\tremaining: 3.49s\n",
      "537:\tlearn: 0.3117972\ttotal: 4.05s\tremaining: 3.48s\n",
      "538:\tlearn: 0.3117293\ttotal: 4.06s\tremaining: 3.47s\n",
      "539:\tlearn: 0.3117073\ttotal: 4.07s\tremaining: 3.46s\n",
      "540:\tlearn: 0.3116928\ttotal: 4.07s\tremaining: 3.45s\n",
      "541:\tlearn: 0.3116734\ttotal: 4.08s\tremaining: 3.44s\n",
      "542:\tlearn: 0.3116719\ttotal: 4.08s\tremaining: 3.44s\n",
      "543:\tlearn: 0.3116711\ttotal: 4.09s\tremaining: 3.43s\n",
      "544:\tlearn: 0.3116621\ttotal: 4.09s\tremaining: 3.42s\n",
      "545:\tlearn: 0.3116562\ttotal: 4.1s\tremaining: 3.41s\n",
      "546:\tlearn: 0.3116543\ttotal: 4.11s\tremaining: 3.4s\n",
      "547:\tlearn: 0.3116284\ttotal: 4.11s\tremaining: 3.39s\n",
      "548:\tlearn: 0.3115752\ttotal: 4.12s\tremaining: 3.38s\n",
      "549:\tlearn: 0.3115585\ttotal: 4.12s\tremaining: 3.37s\n",
      "550:\tlearn: 0.3115401\ttotal: 4.13s\tremaining: 3.36s\n",
      "551:\tlearn: 0.3115401\ttotal: 4.13s\tremaining: 3.35s\n",
      "552:\tlearn: 0.3115263\ttotal: 4.14s\tremaining: 3.35s\n",
      "553:\tlearn: 0.3115252\ttotal: 4.14s\tremaining: 3.34s\n",
      "554:\tlearn: 0.3115234\ttotal: 4.16s\tremaining: 3.33s\n",
      "555:\tlearn: 0.3115206\ttotal: 4.16s\tremaining: 3.32s\n",
      "556:\tlearn: 0.3114872\ttotal: 4.17s\tremaining: 3.32s\n",
      "557:\tlearn: 0.3114820\ttotal: 4.17s\tremaining: 3.31s\n",
      "558:\tlearn: 0.3114454\ttotal: 4.18s\tremaining: 3.3s\n",
      "559:\tlearn: 0.3114223\ttotal: 4.19s\tremaining: 3.29s\n",
      "560:\tlearn: 0.3114210\ttotal: 4.19s\tremaining: 3.28s\n",
      "561:\tlearn: 0.3113902\ttotal: 4.2s\tremaining: 3.27s\n",
      "562:\tlearn: 0.3113419\ttotal: 4.21s\tremaining: 3.26s\n",
      "563:\tlearn: 0.3113334\ttotal: 4.21s\tremaining: 3.25s\n",
      "564:\tlearn: 0.3113117\ttotal: 4.22s\tremaining: 3.25s\n",
      "565:\tlearn: 0.3112718\ttotal: 4.22s\tremaining: 3.24s\n",
      "566:\tlearn: 0.3112588\ttotal: 4.23s\tremaining: 3.23s\n",
      "567:\tlearn: 0.3112546\ttotal: 4.23s\tremaining: 3.22s\n",
      "568:\tlearn: 0.3112295\ttotal: 4.24s\tremaining: 3.21s\n",
      "569:\tlearn: 0.3112276\ttotal: 4.25s\tremaining: 3.2s\n",
      "570:\tlearn: 0.3111922\ttotal: 4.25s\tremaining: 3.19s\n",
      "571:\tlearn: 0.3111878\ttotal: 4.26s\tremaining: 3.19s\n",
      "572:\tlearn: 0.3111860\ttotal: 4.26s\tremaining: 3.18s\n",
      "573:\tlearn: 0.3111375\ttotal: 4.27s\tremaining: 3.17s\n",
      "574:\tlearn: 0.3111366\ttotal: 4.28s\tremaining: 3.16s\n",
      "575:\tlearn: 0.3111365\ttotal: 4.29s\tremaining: 3.15s\n",
      "576:\tlearn: 0.3111193\ttotal: 4.29s\tremaining: 3.15s\n",
      "577:\tlearn: 0.3111049\ttotal: 4.3s\tremaining: 3.14s\n",
      "578:\tlearn: 0.3110859\ttotal: 4.3s\tremaining: 3.13s\n",
      "579:\tlearn: 0.3110849\ttotal: 4.31s\tremaining: 3.12s\n",
      "580:\tlearn: 0.3110630\ttotal: 4.32s\tremaining: 3.11s\n",
      "581:\tlearn: 0.3110257\ttotal: 4.32s\tremaining: 3.1s\n",
      "582:\tlearn: 0.3109904\ttotal: 4.33s\tremaining: 3.09s\n",
      "583:\tlearn: 0.3109904\ttotal: 4.33s\tremaining: 3.08s\n",
      "584:\tlearn: 0.3109888\ttotal: 4.34s\tremaining: 3.08s\n",
      "585:\tlearn: 0.3109860\ttotal: 4.34s\tremaining: 3.07s\n",
      "586:\tlearn: 0.3109542\ttotal: 4.35s\tremaining: 3.06s\n",
      "587:\tlearn: 0.3109034\ttotal: 4.36s\tremaining: 3.05s\n",
      "588:\tlearn: 0.3108999\ttotal: 4.36s\tremaining: 3.04s\n",
      "589:\tlearn: 0.3108776\ttotal: 4.37s\tremaining: 3.04s\n",
      "590:\tlearn: 0.3108770\ttotal: 4.37s\tremaining: 3.03s\n",
      "591:\tlearn: 0.3108716\ttotal: 4.38s\tremaining: 3.02s\n",
      "592:\tlearn: 0.3108486\ttotal: 4.38s\tremaining: 3.01s\n",
      "593:\tlearn: 0.3108244\ttotal: 4.39s\tremaining: 3s\n",
      "594:\tlearn: 0.3107982\ttotal: 4.4s\tremaining: 2.99s\n",
      "595:\tlearn: 0.3107301\ttotal: 4.4s\tremaining: 2.98s\n",
      "596:\tlearn: 0.3107096\ttotal: 4.41s\tremaining: 2.98s\n",
      "597:\tlearn: 0.3107081\ttotal: 4.41s\tremaining: 2.97s\n",
      "598:\tlearn: 0.3107043\ttotal: 4.42s\tremaining: 2.96s\n",
      "599:\tlearn: 0.3106682\ttotal: 4.42s\tremaining: 2.95s\n",
      "600:\tlearn: 0.3106645\ttotal: 4.43s\tremaining: 2.94s\n",
      "601:\tlearn: 0.3106249\ttotal: 4.44s\tremaining: 2.94s\n",
      "602:\tlearn: 0.3105659\ttotal: 4.45s\tremaining: 2.93s\n",
      "603:\tlearn: 0.3105608\ttotal: 4.46s\tremaining: 2.92s\n",
      "604:\tlearn: 0.3105508\ttotal: 4.46s\tremaining: 2.91s\n",
      "605:\tlearn: 0.3105463\ttotal: 4.46s\tremaining: 2.9s\n",
      "606:\tlearn: 0.3105265\ttotal: 4.47s\tremaining: 2.9s\n",
      "607:\tlearn: 0.3104923\ttotal: 4.48s\tremaining: 2.89s\n",
      "608:\tlearn: 0.3104886\ttotal: 4.49s\tremaining: 2.88s\n",
      "609:\tlearn: 0.3104751\ttotal: 4.49s\tremaining: 2.87s\n",
      "610:\tlearn: 0.3104275\ttotal: 4.5s\tremaining: 2.86s\n",
      "611:\tlearn: 0.3104237\ttotal: 4.5s\tremaining: 2.85s\n",
      "612:\tlearn: 0.3104194\ttotal: 4.51s\tremaining: 2.85s\n",
      "613:\tlearn: 0.3103984\ttotal: 4.51s\tremaining: 2.84s\n",
      "614:\tlearn: 0.3103461\ttotal: 4.52s\tremaining: 2.83s\n",
      "615:\tlearn: 0.3103027\ttotal: 4.53s\tremaining: 2.82s\n",
      "616:\tlearn: 0.3102783\ttotal: 4.53s\tremaining: 2.81s\n",
      "617:\tlearn: 0.3102736\ttotal: 4.54s\tremaining: 2.81s\n",
      "618:\tlearn: 0.3102349\ttotal: 4.54s\tremaining: 2.8s\n",
      "619:\tlearn: 0.3102296\ttotal: 4.55s\tremaining: 2.79s\n",
      "620:\tlearn: 0.3102272\ttotal: 4.56s\tremaining: 2.78s\n",
      "621:\tlearn: 0.3101829\ttotal: 4.56s\tremaining: 2.77s\n",
      "622:\tlearn: 0.3101799\ttotal: 4.57s\tremaining: 2.76s\n",
      "623:\tlearn: 0.3101552\ttotal: 4.57s\tremaining: 2.76s\n",
      "624:\tlearn: 0.3101516\ttotal: 4.58s\tremaining: 2.75s\n",
      "625:\tlearn: 0.3101362\ttotal: 4.59s\tremaining: 2.74s\n",
      "626:\tlearn: 0.3101167\ttotal: 4.59s\tremaining: 2.73s\n",
      "627:\tlearn: 0.3101154\ttotal: 4.6s\tremaining: 2.72s\n",
      "628:\tlearn: 0.3101153\ttotal: 4.6s\tremaining: 2.71s\n",
      "629:\tlearn: 0.3101119\ttotal: 4.61s\tremaining: 2.71s\n",
      "630:\tlearn: 0.3101115\ttotal: 4.61s\tremaining: 2.7s\n",
      "631:\tlearn: 0.3101102\ttotal: 4.62s\tremaining: 2.69s\n",
      "632:\tlearn: 0.3101087\ttotal: 4.62s\tremaining: 2.68s\n",
      "633:\tlearn: 0.3100682\ttotal: 4.63s\tremaining: 2.67s\n",
      "634:\tlearn: 0.3100425\ttotal: 4.64s\tremaining: 2.67s\n",
      "635:\tlearn: 0.3100408\ttotal: 4.64s\tremaining: 2.66s\n",
      "636:\tlearn: 0.3100233\ttotal: 4.65s\tremaining: 2.65s\n",
      "637:\tlearn: 0.3099695\ttotal: 4.66s\tremaining: 2.64s\n",
      "638:\tlearn: 0.3099623\ttotal: 4.67s\tremaining: 2.64s\n",
      "639:\tlearn: 0.3099598\ttotal: 4.67s\tremaining: 2.63s\n",
      "640:\tlearn: 0.3099586\ttotal: 4.68s\tremaining: 2.62s\n",
      "641:\tlearn: 0.3099390\ttotal: 4.68s\tremaining: 2.61s\n",
      "642:\tlearn: 0.3098851\ttotal: 4.69s\tremaining: 2.6s\n",
      "643:\tlearn: 0.3098631\ttotal: 4.7s\tremaining: 2.6s\n",
      "644:\tlearn: 0.3098038\ttotal: 4.7s\tremaining: 2.59s\n",
      "645:\tlearn: 0.3097866\ttotal: 4.71s\tremaining: 2.58s\n",
      "646:\tlearn: 0.3097566\ttotal: 4.71s\tremaining: 2.57s\n",
      "647:\tlearn: 0.3097543\ttotal: 4.72s\tremaining: 2.56s\n",
      "648:\tlearn: 0.3097505\ttotal: 4.72s\tremaining: 2.56s\n",
      "649:\tlearn: 0.3097479\ttotal: 4.74s\tremaining: 2.55s\n",
      "650:\tlearn: 0.3097462\ttotal: 4.74s\tremaining: 2.54s\n",
      "651:\tlearn: 0.3097045\ttotal: 4.75s\tremaining: 2.54s\n",
      "652:\tlearn: 0.3097011\ttotal: 4.76s\tremaining: 2.53s\n",
      "653:\tlearn: 0.3096600\ttotal: 4.76s\tremaining: 2.52s\n",
      "654:\tlearn: 0.3096425\ttotal: 4.77s\tremaining: 2.51s\n",
      "655:\tlearn: 0.3095864\ttotal: 4.77s\tremaining: 2.5s\n",
      "656:\tlearn: 0.3095798\ttotal: 4.78s\tremaining: 2.5s\n",
      "657:\tlearn: 0.3095798\ttotal: 4.78s\tremaining: 2.48s\n",
      "658:\tlearn: 0.3095363\ttotal: 4.79s\tremaining: 2.48s\n",
      "659:\tlearn: 0.3095357\ttotal: 4.79s\tremaining: 2.47s\n",
      "660:\tlearn: 0.3095272\ttotal: 4.79s\tremaining: 2.46s\n",
      "661:\tlearn: 0.3094572\ttotal: 4.8s\tremaining: 2.45s\n",
      "662:\tlearn: 0.3094471\ttotal: 4.8s\tremaining: 2.44s\n",
      "663:\tlearn: 0.3094431\ttotal: 4.8s\tremaining: 2.43s\n",
      "664:\tlearn: 0.3094403\ttotal: 4.81s\tremaining: 2.42s\n",
      "665:\tlearn: 0.3094038\ttotal: 4.81s\tremaining: 2.41s\n",
      "666:\tlearn: 0.3093576\ttotal: 4.82s\tremaining: 2.4s\n",
      "667:\tlearn: 0.3093424\ttotal: 4.82s\tremaining: 2.4s\n",
      "668:\tlearn: 0.3093411\ttotal: 4.82s\tremaining: 2.39s\n",
      "669:\tlearn: 0.3093410\ttotal: 4.83s\tremaining: 2.38s\n",
      "670:\tlearn: 0.3093398\ttotal: 4.83s\tremaining: 2.37s\n",
      "671:\tlearn: 0.3093116\ttotal: 4.84s\tremaining: 2.36s\n",
      "672:\tlearn: 0.3093098\ttotal: 4.84s\tremaining: 2.35s\n",
      "673:\tlearn: 0.3092984\ttotal: 4.84s\tremaining: 2.34s\n",
      "674:\tlearn: 0.3092638\ttotal: 4.85s\tremaining: 2.33s\n",
      "675:\tlearn: 0.3092553\ttotal: 4.86s\tremaining: 2.33s\n",
      "676:\tlearn: 0.3092511\ttotal: 4.86s\tremaining: 2.32s\n",
      "677:\tlearn: 0.3092476\ttotal: 4.87s\tremaining: 2.31s\n",
      "678:\tlearn: 0.3092475\ttotal: 4.87s\tremaining: 2.3s\n",
      "679:\tlearn: 0.3092324\ttotal: 4.87s\tremaining: 2.29s\n",
      "680:\tlearn: 0.3092045\ttotal: 4.88s\tremaining: 2.29s\n",
      "681:\tlearn: 0.3092011\ttotal: 4.88s\tremaining: 2.28s\n",
      "682:\tlearn: 0.3091573\ttotal: 4.89s\tremaining: 2.27s\n",
      "683:\tlearn: 0.3091535\ttotal: 4.89s\tremaining: 2.26s\n",
      "684:\tlearn: 0.3091488\ttotal: 4.89s\tremaining: 2.25s\n",
      "685:\tlearn: 0.3091353\ttotal: 4.9s\tremaining: 2.24s\n",
      "686:\tlearn: 0.3091223\ttotal: 4.9s\tremaining: 2.23s\n",
      "687:\tlearn: 0.3091201\ttotal: 4.91s\tremaining: 2.23s\n",
      "688:\tlearn: 0.3090871\ttotal: 4.91s\tremaining: 2.22s\n",
      "689:\tlearn: 0.3090437\ttotal: 4.92s\tremaining: 2.21s\n",
      "690:\tlearn: 0.3090401\ttotal: 4.92s\tremaining: 2.2s\n",
      "691:\tlearn: 0.3090103\ttotal: 4.93s\tremaining: 2.19s\n",
      "692:\tlearn: 0.3090064\ttotal: 4.93s\tremaining: 2.19s\n",
      "693:\tlearn: 0.3090054\ttotal: 4.94s\tremaining: 2.18s\n",
      "694:\tlearn: 0.3089739\ttotal: 4.94s\tremaining: 2.17s\n",
      "695:\tlearn: 0.3089727\ttotal: 4.95s\tremaining: 2.16s\n",
      "696:\tlearn: 0.3089458\ttotal: 4.95s\tremaining: 2.15s\n",
      "697:\tlearn: 0.3089023\ttotal: 4.95s\tremaining: 2.14s\n",
      "698:\tlearn: 0.3088999\ttotal: 4.96s\tremaining: 2.13s\n",
      "699:\tlearn: 0.3088981\ttotal: 4.96s\tremaining: 2.13s\n",
      "700:\tlearn: 0.3088973\ttotal: 4.97s\tremaining: 2.12s\n",
      "701:\tlearn: 0.3088845\ttotal: 4.97s\tremaining: 2.11s\n",
      "702:\tlearn: 0.3088833\ttotal: 4.97s\tremaining: 2.1s\n",
      "703:\tlearn: 0.3088822\ttotal: 4.98s\tremaining: 2.09s\n",
      "704:\tlearn: 0.3088346\ttotal: 4.98s\tremaining: 2.08s\n",
      "705:\tlearn: 0.3088001\ttotal: 4.99s\tremaining: 2.08s\n",
      "706:\tlearn: 0.3087940\ttotal: 4.99s\tremaining: 2.07s\n",
      "707:\tlearn: 0.3087917\ttotal: 5s\tremaining: 2.06s\n",
      "708:\tlearn: 0.3087672\ttotal: 5s\tremaining: 2.05s\n",
      "709:\tlearn: 0.3087579\ttotal: 5.01s\tremaining: 2.04s\n",
      "710:\tlearn: 0.3087561\ttotal: 5.01s\tremaining: 2.04s\n",
      "711:\tlearn: 0.3087537\ttotal: 5.01s\tremaining: 2.03s\n",
      "712:\tlearn: 0.3087528\ttotal: 5.02s\tremaining: 2.02s\n",
      "713:\tlearn: 0.3087478\ttotal: 5.03s\tremaining: 2.01s\n",
      "714:\tlearn: 0.3087461\ttotal: 5.03s\tremaining: 2s\n",
      "715:\tlearn: 0.3087342\ttotal: 5.03s\tremaining: 2s\n",
      "716:\tlearn: 0.3087308\ttotal: 5.04s\tremaining: 1.99s\n",
      "717:\tlearn: 0.3087290\ttotal: 5.05s\tremaining: 1.98s\n",
      "718:\tlearn: 0.3087286\ttotal: 5.05s\tremaining: 1.97s\n",
      "719:\tlearn: 0.3087238\ttotal: 5.06s\tremaining: 1.97s\n",
      "720:\tlearn: 0.3087129\ttotal: 5.06s\tremaining: 1.96s\n",
      "721:\tlearn: 0.3087104\ttotal: 5.07s\tremaining: 1.95s\n",
      "722:\tlearn: 0.3087058\ttotal: 5.08s\tremaining: 1.94s\n",
      "723:\tlearn: 0.3086884\ttotal: 5.08s\tremaining: 1.94s\n",
      "724:\tlearn: 0.3086486\ttotal: 5.08s\tremaining: 1.93s\n",
      "725:\tlearn: 0.3086445\ttotal: 5.09s\tremaining: 1.92s\n",
      "726:\tlearn: 0.3086370\ttotal: 5.1s\tremaining: 1.91s\n",
      "727:\tlearn: 0.3086347\ttotal: 5.1s\tremaining: 1.91s\n",
      "728:\tlearn: 0.3085978\ttotal: 5.11s\tremaining: 1.9s\n",
      "729:\tlearn: 0.3085955\ttotal: 5.11s\tremaining: 1.89s\n",
      "730:\tlearn: 0.3085523\ttotal: 5.12s\tremaining: 1.88s\n",
      "731:\tlearn: 0.3085476\ttotal: 5.13s\tremaining: 1.88s\n",
      "732:\tlearn: 0.3085464\ttotal: 5.13s\tremaining: 1.87s\n",
      "733:\tlearn: 0.3085283\ttotal: 5.13s\tremaining: 1.86s\n",
      "734:\tlearn: 0.3085244\ttotal: 5.14s\tremaining: 1.85s\n",
      "735:\tlearn: 0.3085114\ttotal: 5.15s\tremaining: 1.85s\n",
      "736:\tlearn: 0.3085083\ttotal: 5.15s\tremaining: 1.84s\n",
      "737:\tlearn: 0.3085001\ttotal: 5.16s\tremaining: 1.83s\n",
      "738:\tlearn: 0.3084978\ttotal: 5.16s\tremaining: 1.82s\n",
      "739:\tlearn: 0.3084516\ttotal: 5.17s\tremaining: 1.82s\n",
      "740:\tlearn: 0.3084471\ttotal: 5.18s\tremaining: 1.81s\n",
      "741:\tlearn: 0.3084467\ttotal: 5.18s\tremaining: 1.8s\n",
      "742:\tlearn: 0.3084322\ttotal: 5.19s\tremaining: 1.79s\n",
      "743:\tlearn: 0.3084245\ttotal: 5.19s\tremaining: 1.79s\n",
      "744:\tlearn: 0.3083973\ttotal: 5.2s\tremaining: 1.78s\n",
      "745:\tlearn: 0.3083762\ttotal: 5.2s\tremaining: 1.77s\n",
      "746:\tlearn: 0.3083726\ttotal: 5.21s\tremaining: 1.76s\n",
      "747:\tlearn: 0.3083695\ttotal: 5.21s\tremaining: 1.75s\n",
      "748:\tlearn: 0.3083562\ttotal: 5.22s\tremaining: 1.75s\n",
      "749:\tlearn: 0.3083385\ttotal: 5.22s\tremaining: 1.74s\n",
      "750:\tlearn: 0.3083325\ttotal: 5.22s\tremaining: 1.73s\n",
      "751:\tlearn: 0.3083325\ttotal: 5.23s\tremaining: 1.72s\n",
      "752:\tlearn: 0.3083288\ttotal: 5.23s\tremaining: 1.72s\n",
      "753:\tlearn: 0.3083230\ttotal: 5.24s\tremaining: 1.71s\n",
      "754:\tlearn: 0.3083097\ttotal: 5.24s\tremaining: 1.7s\n",
      "755:\tlearn: 0.3082765\ttotal: 5.25s\tremaining: 1.69s\n",
      "756:\tlearn: 0.3082738\ttotal: 5.25s\tremaining: 1.69s\n",
      "757:\tlearn: 0.3082714\ttotal: 5.26s\tremaining: 1.68s\n",
      "758:\tlearn: 0.3082641\ttotal: 5.26s\tremaining: 1.67s\n",
      "759:\tlearn: 0.3082641\ttotal: 5.26s\tremaining: 1.66s\n",
      "760:\tlearn: 0.3082336\ttotal: 5.27s\tremaining: 1.65s\n",
      "761:\tlearn: 0.3081956\ttotal: 5.27s\tremaining: 1.65s\n",
      "762:\tlearn: 0.3081945\ttotal: 5.28s\tremaining: 1.64s\n",
      "763:\tlearn: 0.3081934\ttotal: 5.28s\tremaining: 1.63s\n",
      "764:\tlearn: 0.3081900\ttotal: 5.28s\tremaining: 1.62s\n",
      "765:\tlearn: 0.3081794\ttotal: 5.29s\tremaining: 1.61s\n",
      "766:\tlearn: 0.3081724\ttotal: 5.29s\tremaining: 1.61s\n",
      "767:\tlearn: 0.3081646\ttotal: 5.3s\tremaining: 1.6s\n",
      "768:\tlearn: 0.3081491\ttotal: 5.3s\tremaining: 1.59s\n",
      "769:\tlearn: 0.3081135\ttotal: 5.3s\tremaining: 1.58s\n",
      "770:\tlearn: 0.3081113\ttotal: 5.31s\tremaining: 1.58s\n",
      "771:\tlearn: 0.3080768\ttotal: 5.31s\tremaining: 1.57s\n",
      "772:\tlearn: 0.3080637\ttotal: 5.32s\tremaining: 1.56s\n",
      "773:\tlearn: 0.3080494\ttotal: 5.32s\tremaining: 1.55s\n",
      "774:\tlearn: 0.3080444\ttotal: 5.33s\tremaining: 1.55s\n",
      "775:\tlearn: 0.3080424\ttotal: 5.33s\tremaining: 1.54s\n",
      "776:\tlearn: 0.3080358\ttotal: 5.33s\tremaining: 1.53s\n",
      "777:\tlearn: 0.3080358\ttotal: 5.34s\tremaining: 1.52s\n",
      "778:\tlearn: 0.3080239\ttotal: 5.34s\tremaining: 1.51s\n",
      "779:\tlearn: 0.3080071\ttotal: 5.36s\tremaining: 1.51s\n",
      "780:\tlearn: 0.3079603\ttotal: 5.39s\tremaining: 1.51s\n",
      "781:\tlearn: 0.3079568\ttotal: 5.4s\tremaining: 1.5s\n",
      "782:\tlearn: 0.3079531\ttotal: 5.4s\tremaining: 1.5s\n",
      "783:\tlearn: 0.3079522\ttotal: 5.42s\tremaining: 1.49s\n",
      "784:\tlearn: 0.3079458\ttotal: 5.43s\tremaining: 1.49s\n",
      "785:\tlearn: 0.3079456\ttotal: 5.45s\tremaining: 1.48s\n",
      "786:\tlearn: 0.3079446\ttotal: 5.45s\tremaining: 1.48s\n",
      "787:\tlearn: 0.3079418\ttotal: 5.46s\tremaining: 1.47s\n",
      "788:\tlearn: 0.3079285\ttotal: 5.48s\tremaining: 1.46s\n",
      "789:\tlearn: 0.3079276\ttotal: 5.49s\tremaining: 1.46s\n",
      "790:\tlearn: 0.3079268\ttotal: 5.5s\tremaining: 1.45s\n",
      "791:\tlearn: 0.3079109\ttotal: 5.51s\tremaining: 1.45s\n",
      "792:\tlearn: 0.3079010\ttotal: 5.53s\tremaining: 1.44s\n",
      "793:\tlearn: 0.3078906\ttotal: 5.54s\tremaining: 1.44s\n",
      "794:\tlearn: 0.3078845\ttotal: 5.54s\tremaining: 1.43s\n",
      "795:\tlearn: 0.3078775\ttotal: 5.55s\tremaining: 1.42s\n",
      "796:\tlearn: 0.3078645\ttotal: 5.55s\tremaining: 1.41s\n",
      "797:\tlearn: 0.3078629\ttotal: 5.56s\tremaining: 1.41s\n",
      "798:\tlearn: 0.3078593\ttotal: 5.56s\tremaining: 1.4s\n",
      "799:\tlearn: 0.3078254\ttotal: 5.57s\tremaining: 1.39s\n",
      "800:\tlearn: 0.3078159\ttotal: 5.57s\tremaining: 1.38s\n",
      "801:\tlearn: 0.3078153\ttotal: 5.58s\tremaining: 1.38s\n",
      "802:\tlearn: 0.3078066\ttotal: 5.58s\tremaining: 1.37s\n",
      "803:\tlearn: 0.3077803\ttotal: 5.59s\tremaining: 1.36s\n",
      "804:\tlearn: 0.3077534\ttotal: 5.59s\tremaining: 1.35s\n",
      "805:\tlearn: 0.3077508\ttotal: 5.6s\tremaining: 1.35s\n",
      "806:\tlearn: 0.3077475\ttotal: 5.6s\tremaining: 1.34s\n",
      "807:\tlearn: 0.3077283\ttotal: 5.6s\tremaining: 1.33s\n",
      "808:\tlearn: 0.3077226\ttotal: 5.61s\tremaining: 1.32s\n",
      "809:\tlearn: 0.3077205\ttotal: 5.61s\tremaining: 1.32s\n",
      "810:\tlearn: 0.3076833\ttotal: 5.62s\tremaining: 1.31s\n",
      "811:\tlearn: 0.3076198\ttotal: 5.64s\tremaining: 1.3s\n",
      "812:\tlearn: 0.3076181\ttotal: 5.64s\tremaining: 1.3s\n",
      "813:\tlearn: 0.3075687\ttotal: 5.65s\tremaining: 1.29s\n",
      "814:\tlearn: 0.3075675\ttotal: 5.65s\tremaining: 1.28s\n",
      "815:\tlearn: 0.3075546\ttotal: 5.66s\tremaining: 1.28s\n",
      "816:\tlearn: 0.3075527\ttotal: 5.66s\tremaining: 1.27s\n",
      "817:\tlearn: 0.3075473\ttotal: 5.67s\tremaining: 1.26s\n",
      "818:\tlearn: 0.3075424\ttotal: 5.68s\tremaining: 1.25s\n",
      "819:\tlearn: 0.3075245\ttotal: 5.69s\tremaining: 1.25s\n",
      "820:\tlearn: 0.3075025\ttotal: 5.69s\tremaining: 1.24s\n",
      "821:\tlearn: 0.3074865\ttotal: 5.7s\tremaining: 1.23s\n",
      "822:\tlearn: 0.3074864\ttotal: 5.7s\tremaining: 1.23s\n",
      "823:\tlearn: 0.3074833\ttotal: 5.71s\tremaining: 1.22s\n",
      "824:\tlearn: 0.3074568\ttotal: 5.71s\tremaining: 1.21s\n",
      "825:\tlearn: 0.3074539\ttotal: 5.72s\tremaining: 1.21s\n",
      "826:\tlearn: 0.3074330\ttotal: 5.73s\tremaining: 1.2s\n",
      "827:\tlearn: 0.3074120\ttotal: 5.74s\tremaining: 1.19s\n",
      "828:\tlearn: 0.3073898\ttotal: 5.75s\tremaining: 1.19s\n",
      "829:\tlearn: 0.3073873\ttotal: 5.76s\tremaining: 1.18s\n",
      "830:\tlearn: 0.3073208\ttotal: 5.76s\tremaining: 1.17s\n",
      "831:\tlearn: 0.3073019\ttotal: 5.77s\tremaining: 1.17s\n",
      "832:\tlearn: 0.3073011\ttotal: 5.78s\tremaining: 1.16s\n",
      "833:\tlearn: 0.3072621\ttotal: 5.78s\tremaining: 1.15s\n",
      "834:\tlearn: 0.3072586\ttotal: 5.79s\tremaining: 1.14s\n",
      "835:\tlearn: 0.3072482\ttotal: 5.79s\tremaining: 1.14s\n",
      "836:\tlearn: 0.3071960\ttotal: 5.8s\tremaining: 1.13s\n",
      "837:\tlearn: 0.3071746\ttotal: 5.8s\tremaining: 1.12s\n",
      "838:\tlearn: 0.3071728\ttotal: 5.81s\tremaining: 1.11s\n",
      "839:\tlearn: 0.3071707\ttotal: 5.82s\tremaining: 1.11s\n",
      "840:\tlearn: 0.3071706\ttotal: 5.83s\tremaining: 1.1s\n",
      "841:\tlearn: 0.3071658\ttotal: 5.83s\tremaining: 1.09s\n",
      "842:\tlearn: 0.3071627\ttotal: 5.84s\tremaining: 1.09s\n",
      "843:\tlearn: 0.3071583\ttotal: 5.84s\tremaining: 1.08s\n",
      "844:\tlearn: 0.3071532\ttotal: 5.85s\tremaining: 1.07s\n",
      "845:\tlearn: 0.3071526\ttotal: 5.86s\tremaining: 1.06s\n",
      "846:\tlearn: 0.3071516\ttotal: 5.86s\tremaining: 1.06s\n",
      "847:\tlearn: 0.3071439\ttotal: 5.87s\tremaining: 1.05s\n",
      "848:\tlearn: 0.3071402\ttotal: 5.87s\tremaining: 1.04s\n",
      "849:\tlearn: 0.3071035\ttotal: 5.88s\tremaining: 1.04s\n",
      "850:\tlearn: 0.3070986\ttotal: 5.89s\tremaining: 1.03s\n",
      "851:\tlearn: 0.3070931\ttotal: 5.89s\tremaining: 1.02s\n",
      "852:\tlearn: 0.3070866\ttotal: 5.9s\tremaining: 1.02s\n",
      "853:\tlearn: 0.3070850\ttotal: 5.9s\tremaining: 1.01s\n",
      "854:\tlearn: 0.3070839\ttotal: 5.91s\tremaining: 1s\n",
      "855:\tlearn: 0.3070824\ttotal: 5.92s\tremaining: 995ms\n",
      "856:\tlearn: 0.3070612\ttotal: 5.92s\tremaining: 988ms\n",
      "857:\tlearn: 0.3070535\ttotal: 5.93s\tremaining: 981ms\n",
      "858:\tlearn: 0.3070525\ttotal: 5.93s\tremaining: 974ms\n",
      "859:\tlearn: 0.3070458\ttotal: 5.94s\tremaining: 967ms\n",
      "860:\tlearn: 0.3070416\ttotal: 5.95s\tremaining: 960ms\n",
      "861:\tlearn: 0.3070225\ttotal: 5.96s\tremaining: 954ms\n",
      "862:\tlearn: 0.3070217\ttotal: 5.96s\tremaining: 947ms\n",
      "863:\tlearn: 0.3070207\ttotal: 5.97s\tremaining: 940ms\n",
      "864:\tlearn: 0.3070201\ttotal: 5.98s\tremaining: 933ms\n",
      "865:\tlearn: 0.3070052\ttotal: 5.98s\tremaining: 926ms\n",
      "866:\tlearn: 0.3069434\ttotal: 5.99s\tremaining: 918ms\n",
      "867:\tlearn: 0.3069378\ttotal: 5.99s\tremaining: 911ms\n",
      "868:\tlearn: 0.3069165\ttotal: 6s\tremaining: 904ms\n",
      "869:\tlearn: 0.3069106\ttotal: 6s\tremaining: 897ms\n",
      "870:\tlearn: 0.3069031\ttotal: 6.01s\tremaining: 890ms\n",
      "871:\tlearn: 0.3068892\ttotal: 6.02s\tremaining: 883ms\n",
      "872:\tlearn: 0.3068800\ttotal: 6.02s\tremaining: 876ms\n",
      "873:\tlearn: 0.3068741\ttotal: 6.03s\tremaining: 869ms\n",
      "874:\tlearn: 0.3068563\ttotal: 6.03s\tremaining: 861ms\n",
      "875:\tlearn: 0.3068561\ttotal: 6.04s\tremaining: 854ms\n",
      "876:\tlearn: 0.3068341\ttotal: 6.04s\tremaining: 847ms\n",
      "877:\tlearn: 0.3068270\ttotal: 6.05s\tremaining: 841ms\n",
      "878:\tlearn: 0.3068247\ttotal: 6.06s\tremaining: 834ms\n",
      "879:\tlearn: 0.3068077\ttotal: 6.06s\tremaining: 827ms\n",
      "880:\tlearn: 0.3068059\ttotal: 6.07s\tremaining: 820ms\n",
      "881:\tlearn: 0.3068059\ttotal: 6.08s\tremaining: 813ms\n",
      "882:\tlearn: 0.3067694\ttotal: 6.08s\tremaining: 806ms\n",
      "883:\tlearn: 0.3067575\ttotal: 6.09s\tremaining: 799ms\n",
      "884:\tlearn: 0.3067548\ttotal: 6.1s\tremaining: 793ms\n",
      "885:\tlearn: 0.3067280\ttotal: 6.11s\tremaining: 786ms\n",
      "886:\tlearn: 0.3067231\ttotal: 6.11s\tremaining: 778ms\n",
      "887:\tlearn: 0.3067209\ttotal: 6.11s\tremaining: 771ms\n",
      "888:\tlearn: 0.3067135\ttotal: 6.12s\tremaining: 764ms\n",
      "889:\tlearn: 0.3067105\ttotal: 6.12s\tremaining: 757ms\n",
      "890:\tlearn: 0.3067050\ttotal: 6.13s\tremaining: 749ms\n",
      "891:\tlearn: 0.3067013\ttotal: 6.13s\tremaining: 742ms\n",
      "892:\tlearn: 0.3067013\ttotal: 6.13s\tremaining: 735ms\n",
      "893:\tlearn: 0.3066869\ttotal: 6.14s\tremaining: 728ms\n",
      "894:\tlearn: 0.3066775\ttotal: 6.15s\tremaining: 721ms\n",
      "895:\tlearn: 0.3066767\ttotal: 6.15s\tremaining: 714ms\n",
      "896:\tlearn: 0.3066738\ttotal: 6.15s\tremaining: 707ms\n",
      "897:\tlearn: 0.3066722\ttotal: 6.16s\tremaining: 699ms\n",
      "898:\tlearn: 0.3066702\ttotal: 6.16s\tremaining: 692ms\n",
      "899:\tlearn: 0.3066614\ttotal: 6.17s\tremaining: 685ms\n",
      "900:\tlearn: 0.3066410\ttotal: 6.17s\tremaining: 678ms\n",
      "901:\tlearn: 0.3066160\ttotal: 6.18s\tremaining: 671ms\n",
      "902:\tlearn: 0.3065826\ttotal: 6.18s\tremaining: 664ms\n",
      "903:\tlearn: 0.3065780\ttotal: 6.19s\tremaining: 657ms\n",
      "904:\tlearn: 0.3065760\ttotal: 6.19s\tremaining: 650ms\n",
      "905:\tlearn: 0.3065752\ttotal: 6.2s\tremaining: 643ms\n",
      "906:\tlearn: 0.3065323\ttotal: 6.2s\tremaining: 636ms\n",
      "907:\tlearn: 0.3065151\ttotal: 6.21s\tremaining: 629ms\n",
      "908:\tlearn: 0.3064852\ttotal: 6.21s\tremaining: 622ms\n",
      "909:\tlearn: 0.3064641\ttotal: 6.22s\tremaining: 615ms\n",
      "910:\tlearn: 0.3064479\ttotal: 6.22s\tremaining: 608ms\n",
      "911:\tlearn: 0.3064223\ttotal: 6.23s\tremaining: 601ms\n",
      "912:\tlearn: 0.3063944\ttotal: 6.24s\tremaining: 594ms\n",
      "913:\tlearn: 0.3063927\ttotal: 6.24s\tremaining: 587ms\n",
      "914:\tlearn: 0.3063912\ttotal: 6.25s\tremaining: 580ms\n",
      "915:\tlearn: 0.3063895\ttotal: 6.25s\tremaining: 573ms\n",
      "916:\tlearn: 0.3063872\ttotal: 6.26s\tremaining: 567ms\n",
      "917:\tlearn: 0.3063616\ttotal: 6.27s\tremaining: 560ms\n",
      "918:\tlearn: 0.3063616\ttotal: 6.27s\tremaining: 553ms\n",
      "919:\tlearn: 0.3063607\ttotal: 6.29s\tremaining: 547ms\n",
      "920:\tlearn: 0.3063556\ttotal: 6.29s\tremaining: 540ms\n",
      "921:\tlearn: 0.3063510\ttotal: 6.3s\tremaining: 533ms\n",
      "922:\tlearn: 0.3063503\ttotal: 6.3s\tremaining: 526ms\n",
      "923:\tlearn: 0.3063457\ttotal: 6.31s\tremaining: 519ms\n",
      "924:\tlearn: 0.3063376\ttotal: 6.31s\tremaining: 512ms\n",
      "925:\tlearn: 0.3063353\ttotal: 6.32s\tremaining: 505ms\n",
      "926:\tlearn: 0.3063204\ttotal: 6.32s\tremaining: 498ms\n",
      "927:\tlearn: 0.3063015\ttotal: 6.32s\tremaining: 491ms\n",
      "928:\tlearn: 0.3062752\ttotal: 6.33s\tremaining: 484ms\n",
      "929:\tlearn: 0.3062524\ttotal: 6.34s\tremaining: 477ms\n",
      "930:\tlearn: 0.3062493\ttotal: 6.34s\tremaining: 470ms\n",
      "931:\tlearn: 0.3062489\ttotal: 6.34s\tremaining: 463ms\n",
      "932:\tlearn: 0.3062289\ttotal: 6.35s\tremaining: 456ms\n",
      "933:\tlearn: 0.3062282\ttotal: 6.35s\tremaining: 449ms\n",
      "934:\tlearn: 0.3062269\ttotal: 6.36s\tremaining: 442ms\n",
      "935:\tlearn: 0.3062223\ttotal: 6.36s\tremaining: 435ms\n",
      "936:\tlearn: 0.3062008\ttotal: 6.37s\tremaining: 428ms\n",
      "937:\tlearn: 0.3061972\ttotal: 6.37s\tremaining: 421ms\n",
      "938:\tlearn: 0.3061947\ttotal: 6.37s\tremaining: 414ms\n",
      "939:\tlearn: 0.3061905\ttotal: 6.38s\tremaining: 407ms\n",
      "940:\tlearn: 0.3061872\ttotal: 6.38s\tremaining: 400ms\n",
      "941:\tlearn: 0.3061700\ttotal: 6.39s\tremaining: 393ms\n",
      "942:\tlearn: 0.3061655\ttotal: 6.39s\tremaining: 386ms\n",
      "943:\tlearn: 0.3061574\ttotal: 6.39s\tremaining: 379ms\n",
      "944:\tlearn: 0.3061450\ttotal: 6.4s\tremaining: 372ms\n",
      "945:\tlearn: 0.3061219\ttotal: 6.41s\tremaining: 366ms\n",
      "946:\tlearn: 0.3061074\ttotal: 6.41s\tremaining: 359ms\n",
      "947:\tlearn: 0.3061034\ttotal: 6.41s\tremaining: 352ms\n",
      "948:\tlearn: 0.3061019\ttotal: 6.42s\tremaining: 345ms\n",
      "949:\tlearn: 0.3061012\ttotal: 6.43s\tremaining: 338ms\n",
      "950:\tlearn: 0.3061001\ttotal: 6.43s\tremaining: 331ms\n",
      "951:\tlearn: 0.3060711\ttotal: 6.43s\tremaining: 324ms\n",
      "952:\tlearn: 0.3060695\ttotal: 6.44s\tremaining: 318ms\n",
      "953:\tlearn: 0.3060693\ttotal: 6.44s\tremaining: 311ms\n",
      "954:\tlearn: 0.3060682\ttotal: 6.45s\tremaining: 304ms\n",
      "955:\tlearn: 0.3060634\ttotal: 6.45s\tremaining: 297ms\n",
      "956:\tlearn: 0.3060627\ttotal: 6.45s\tremaining: 290ms\n",
      "957:\tlearn: 0.3060613\ttotal: 6.46s\tremaining: 283ms\n",
      "958:\tlearn: 0.3060474\ttotal: 6.46s\tremaining: 276ms\n",
      "959:\tlearn: 0.3060474\ttotal: 6.46s\tremaining: 269ms\n",
      "960:\tlearn: 0.3060465\ttotal: 6.47s\tremaining: 263ms\n",
      "961:\tlearn: 0.3060271\ttotal: 6.47s\tremaining: 256ms\n",
      "962:\tlearn: 0.3060245\ttotal: 6.48s\tremaining: 249ms\n",
      "963:\tlearn: 0.3060239\ttotal: 6.48s\tremaining: 242ms\n",
      "964:\tlearn: 0.3060233\ttotal: 6.49s\tremaining: 235ms\n",
      "965:\tlearn: 0.3060191\ttotal: 6.49s\tremaining: 229ms\n",
      "966:\tlearn: 0.3060151\ttotal: 6.5s\tremaining: 222ms\n",
      "967:\tlearn: 0.3060107\ttotal: 6.5s\tremaining: 215ms\n",
      "968:\tlearn: 0.3060095\ttotal: 6.5s\tremaining: 208ms\n",
      "969:\tlearn: 0.3060080\ttotal: 6.51s\tremaining: 201ms\n",
      "970:\tlearn: 0.3060067\ttotal: 6.52s\tremaining: 195ms\n",
      "971:\tlearn: 0.3060059\ttotal: 6.52s\tremaining: 188ms\n",
      "972:\tlearn: 0.3060052\ttotal: 6.53s\tremaining: 181ms\n",
      "973:\tlearn: 0.3060041\ttotal: 6.53s\tremaining: 174ms\n",
      "974:\tlearn: 0.3059945\ttotal: 6.53s\tremaining: 168ms\n",
      "975:\tlearn: 0.3059933\ttotal: 6.54s\tremaining: 161ms\n",
      "976:\tlearn: 0.3059927\ttotal: 6.54s\tremaining: 154ms\n",
      "977:\tlearn: 0.3059902\ttotal: 6.55s\tremaining: 147ms\n",
      "978:\tlearn: 0.3059694\ttotal: 6.55s\tremaining: 141ms\n",
      "979:\tlearn: 0.3059678\ttotal: 6.56s\tremaining: 134ms\n",
      "980:\tlearn: 0.3059654\ttotal: 6.56s\tremaining: 127ms\n",
      "981:\tlearn: 0.3059648\ttotal: 6.57s\tremaining: 120ms\n",
      "982:\tlearn: 0.3059591\ttotal: 6.57s\tremaining: 114ms\n",
      "983:\tlearn: 0.3059220\ttotal: 6.58s\tremaining: 107ms\n",
      "984:\tlearn: 0.3059216\ttotal: 6.58s\tremaining: 100ms\n",
      "985:\tlearn: 0.3058805\ttotal: 6.59s\tremaining: 93.5ms\n",
      "986:\tlearn: 0.3058795\ttotal: 6.59s\tremaining: 86.8ms\n",
      "987:\tlearn: 0.3058454\ttotal: 6.6s\tremaining: 80.1ms\n",
      "988:\tlearn: 0.3058426\ttotal: 6.6s\tremaining: 73.4ms\n",
      "989:\tlearn: 0.3058418\ttotal: 6.61s\tremaining: 66.7ms\n",
      "990:\tlearn: 0.3058418\ttotal: 6.61s\tremaining: 60ms\n",
      "991:\tlearn: 0.3058407\ttotal: 6.61s\tremaining: 53.3ms\n",
      "992:\tlearn: 0.3058379\ttotal: 6.62s\tremaining: 46.7ms\n",
      "993:\tlearn: 0.3058324\ttotal: 6.62s\tremaining: 40ms\n",
      "994:\tlearn: 0.3058203\ttotal: 6.63s\tremaining: 33.3ms\n",
      "995:\tlearn: 0.3058196\ttotal: 6.63s\tremaining: 26.6ms\n",
      "996:\tlearn: 0.3058167\ttotal: 6.63s\tremaining: 20ms\n",
      "997:\tlearn: 0.3057937\ttotal: 6.64s\tremaining: 13.3ms\n",
      "998:\tlearn: 0.3057897\ttotal: 6.64s\tremaining: 6.65ms\n",
      "999:\tlearn: 0.3057859\ttotal: 6.65s\tremaining: 0us\n",
      "Training:  46.15%\n",
      "Test set:  39.69%\n"
     ]
    },
    {
     "ename": "ValueError",
     "evalue": "Classification metrics can't handle a mix of binary and continuous targets",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-217-cb999dbf1f2d>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Training: {:6.2f}%\"\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mmodel_cat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscore\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Test set: {:6.2f}%\"\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mmodel_cat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscore\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_test\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 9\u001b[0;31m \u001b[0maccuracy\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0maccuracy_score\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0myhat\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     10\u001b[0m \u001b[0mroc_auc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mroc_auc_score\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0myhat\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'Accuracy_: %.2f'\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0maccuracy\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/anaconda3/envs/metis/lib/python3.8/site-packages/sklearn/utils/validation.py\u001b[0m in \u001b[0;36minner_f\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     70\u001b[0m                           FutureWarning)\n\u001b[1;32m     71\u001b[0m         \u001b[0mkwargs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mupdate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m{\u001b[0m\u001b[0mk\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0marg\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mk\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0marg\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msig\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mparameters\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 72\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     73\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0minner_f\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     74\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/anaconda3/envs/metis/lib/python3.8/site-packages/sklearn/metrics/_classification.py\u001b[0m in \u001b[0;36maccuracy_score\u001b[0;34m(y_true, y_pred, normalize, sample_weight)\u001b[0m\n\u001b[1;32m    185\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    186\u001b[0m     \u001b[0;31m# Compute accuracy for each possible representation\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 187\u001b[0;31m     \u001b[0my_type\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_true\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_pred\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_check_targets\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_true\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_pred\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    188\u001b[0m     \u001b[0mcheck_consistent_length\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_true\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_pred\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    189\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0my_type\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstartswith\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'multilabel'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/anaconda3/envs/metis/lib/python3.8/site-packages/sklearn/metrics/_classification.py\u001b[0m in \u001b[0;36m_check_targets\u001b[0;34m(y_true, y_pred)\u001b[0m\n\u001b[1;32m     88\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0my_type\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 90\u001b[0;31m         raise ValueError(\"Classification metrics can't handle a mix of {0} \"\n\u001b[0m\u001b[1;32m     91\u001b[0m                          \"and {1} targets\".format(type_true, type_pred))\n\u001b[1;32m     92\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: Classification metrics can't handle a mix of binary and continuous targets"
     ]
    }
   ],
   "source": [
    "model_cat = CatBoostRegressor()\n",
    "# fit on the training set\n",
    "model_cat.fit(X_train, y_train)\n",
    "# predict on test set\n",
    "yhat = model_cat.predict(X_test)\n",
    "# evaluate predictions\n",
    "print(\"Training: {:6.2f}%\".format(100*model_cat.score(X_train, y_train)))\n",
    "print(\"Test set: {:6.2f}%\".format(100*model_cat.score(X_test, y_test)))\n",
    "accuracy = accuracy_score(y_test, yhat)\n",
    "roc_auc = roc_auc_score(y_test, yhat)\n",
    "print('Accuracy_: %.2f' % (accuracy*100))\n",
    "print('roc_auc_score: %.2f' % (roc_auc*100))\n",
    "print(classification_report(y_test, yhat))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "roc_auc_score: 90.40\n"
     ]
    }
   ],
   "source": [
    "roc_auc = roc_auc_score(y_test, yhat)\n",
    "print('roc_auc_score: %.2f' % (roc_auc*100))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "fpr, tpr, thresholds = roc_curve(y_test, yhat)\n",
    "plt.plot(fpr, tpr,lw=2,c='navy')\n",
    "plt.plot([0,1],[0,1],c='purple',ls='--')\n",
    "plt.xlim([-0.05,1.05])\n",
    "plt.ylim([-0.05,1.05])\n",
    "\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve for fraud problem');\n",
    "plt.savefig(\"Roc_CatBoost.svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "importance = model_cat.get_feature_importance()\n",
    "plt.bar([x for x in range(len(importance))], importance);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = np.argsort(importance)[-30:]\n",
    "\n",
    "feature_raw=[]\n",
    "for i in indices:\n",
    "    feature_raw.append(names[i])\n",
    "\n",
    "feature_clean=[]\n",
    "for i in feature_raw:\n",
    "    if \".\" in (i.split(\"_\")[-1]):\n",
    "        feature_clean.append(\"_\".join(i.split(\"_\")[:-1]))b\n",
    "    else:\n",
    "        feature_clean.append(i)\n",
    "        \n",
    "zip_list=list(zip(feature_clean, importance[-30:]))\n",
    "\n",
    "imp=defaultdict(float)\n",
    "for i in zip_list:\n",
    "    imp[i[0]]+=i[1]\n",
    "    \n",
    "imp_df=pd.Series(imp).sort_values(ascending = False)\n",
    "imp_df = pd.DataFrame(imp_df)\n",
    "imp_df.reset_index(inplace=True)\n",
    "imp_df.rename(columns = {'index':'features', 0:'importance'}, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "ax = sns.barplot(x='importance', y='features', data=imp_df, palette=sns.color_palette(\"YlOrRd\", 10))\n",
    "# plt.xticks(rotation='vertical')\n",
    "ax.grid(b=True, which='major', color='#d3d3d3', linewidth=1.0)\n",
    "ax.grid(b=True, which='minor', color='#d3d3d3', linewidth=0.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'age_group_55 - 64 Years',\n",
       " 'behavioral_touch_face',\n",
       " 'census_msa_MSA, Not Principle  City',\n",
       " 'census_msa_MSA, Principle City',\n",
       " 'doctor_recc_h1n1',\n",
       " 'doctor_recc_seasonal',\n",
       " 'education_College Graduate',\n",
       " 'employment_occupation_cmhcxjea',\n",
       " 'h1n1_concern',\n",
       " 'h1n1_knowledge',\n",
       " 'health_worker',\n",
       " 'income_poverty_> $75,000',\n",
       " 'opinion_h1n1_risk',\n",
       " 'opinion_h1n1_sick_from_vacc',\n",
       " 'opinion_h1n1_vacc_effective',\n",
       " 'opinion_seas_risk',\n",
       " 'opinion_seas_sick_from_vacc',\n",
       " 'opinion_seas_vacc_effective',\n",
       " 'seasonal_vaccine_0',\n",
       " 'seasonal_vaccine_1'}"
      ]
     },
     "execution_count": 235,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_keep = set(feature_clean)\n",
    "feature_keep"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training:  88.00%\n",
      "Test set:  85.19%\n",
      "Accuracy_: 85.19\n",
      "roc_auc_score: 75.25\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.88      0.93      0.91      5719\n",
      "           1       0.71      0.57      0.63      1646\n",
      "\n",
      "    accuracy                           0.85      7365\n",
      "   macro avg       0.80      0.75      0.77      7365\n",
      "weighted avg       0.84      0.85      0.85      7365\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model = XGBClassifier()\n",
    "# fit on the training set\n",
    "model.fit(X_train, y_train)\n",
    "# predict on test set\n",
    "yhat = model.predict(X_test)\n",
    "# evaluate predictions\n",
    "print(\"Training: {:6.2f}%\".format(100*model.score(X_train, y_train)))\n",
    "print(\"Test set: {:6.2f}%\".format(100*model.score(X_test, y_test)))\n",
    "accuracy = accuracy_score(y_test, yhat)\n",
    "roc_auc = roc_auc_score(y_test, yhat)\n",
    "print('Accuracy_: %.2f' % (accuracy*100))\n",
    "print('roc_auc_score: %.2f' % (roc_auc*100))\n",
    "print(classification_report(y_test, yhat))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "importance = model_cat.get_feature_importance()\n",
    "plt.bar([x for x in range(len(importance))], importance);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "metadata": {},
   "outputs": [],
   "source": [
    "indices = np.argsort(importance)[-30:]\n",
    "\n",
    "feature_raw=[]\n",
    "for i in indices:\n",
    "    feature_raw.append(names[i])\n",
    "\n",
    "feature_clean=[]\n",
    "for i in feature_raw:\n",
    "    if \".\" in (i.split(\"_\")[-1]):\n",
    "        feature_clean.append(\"_\".join(i.split(\"_\")[:-1]))\n",
    "    else:\n",
    "        feature_clean.append(i)\n",
    "        \n",
    "zip_list=list(zip(feature_clean, importance[-30:]))\n",
    "\n",
    "imp=defaultdict(float)\n",
    "for i in zip_list:\n",
    "    imp[i[0]]+=i[1]\n",
    "    \n",
    "imp_df=pd.Series(imp).sort_values(ascending = False)\n",
    "imp_df = pd.DataFrame(imp_df)\n",
    "imp_df.reset_index(inplace=True)\n",
    "imp_df.rename(columns = {'index':'features', 0:'importance'}, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 303,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "ax = sns.barplot(x='importance', y='features', data=imp_df, palette=sns.color_palette(\"GnBu\", 10))\n",
    "# plt.xticks(rotation='vertical')\n",
    "ax.grid(b=True, which='major', color='#d3d3d3', linewidth=1.0)\n",
    "ax.grid(b=True, which='minor', color='#d3d3d3', linewidth=0.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Models Pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load pre-trained/tuned models\n",
    "\n",
    "model_names = [\"lr_model\", \"nb_model\", \"knn_model\", \"svc_model\", \"rf_model\", \"et_model\", \"ada_model\",'cat_model']\n",
    "\n",
    "for model_name in model_names:\n",
    "    with open(f\"models/{model_name}.pickle\", \"rb\") as pfile:\n",
    "        exec(f\"{model_name} = pickle.load(pfile)\")\n",
    "\n",
    "model_vars = [eval(n) for n in model_names]\n",
    "model_list = list(zip(model_names, model_vars))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lr_model score: 0.8537693006357856\n",
      "nb_model score: 0.584014532243415\n",
      "knn_model score: 0.8133514986376021\n",
      "svc_model score: 0.8542234332425068\n",
      "rf_model score: 0.8510445049954587\n",
      "et_model score: 0.849227974568574\n",
      "ada_model score: 0.8583106267029973\n",
      "cat_model score: 0.3815981260891006\n"
     ]
    }
   ],
   "source": [
    "# Quick peek at each model performance\n",
    "\n",
    "for model_name in model_names:\n",
    "    curr_model = eval(model_name)\n",
    "    print(f'{model_name} score: {curr_model.score(X_test, y_test)}')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lr_model roc_auc_score: 0.8104329298091846\n",
      "nb_model roc_auc_score: 0.673957951271286\n",
      "knn_model roc_auc_score: 0.7616977196704781\n",
      "svc_model roc_auc_score: 0.8015587631833269\n",
      "rf_model roc_auc_score: 0.7919610314201165\n",
      "et_model roc_auc_score: 0.7906798014393349\n",
      "ada_model roc_auc_score: 0.8136360047611385\n",
      "cat_model roc_auc_score: 0.8664817444715727\n"
     ]
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "    curr_model = eval(model_name)\n",
    "    print(f'{model_name} roc_auc_score: {roc_auc_score(y_test, curr_model.predict(X_test))}')\n",
    "#     print(f'{model_name} accuracy: {accuracy_score(X_test.reshape(-1,1), y_test)}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lr_model roc_auc_score: 0.9035419712752635\n",
      "nb_model roc_auc_score: 0.8065337732023052\n",
      "knn_model roc_auc_score: 0.844565416184873\n",
      "svc_model roc_auc_score: 0.9032031748918898\n",
      "rf_model roc_auc_score: 0.9009984999015591\n",
      "et_model roc_auc_score: 0.9011868866811047\n",
      "ada_model roc_auc_score: 0.9017300602337995\n"
     ]
    },
    {
     "ename": "AttributeError",
     "evalue": "'CatBoostRegressor' object has no attribute 'predict_proba'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-106-763d27b73c2b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mmodel_name\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mmodel_names\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m     \u001b[0mcurr_model\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0meval\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m     \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34mf'{model_name} roc_auc_score: {roc_auc_score(y_test, curr_model.predict_proba(X_test)[:,1])}'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0;31m#     print(f'{model_name} accuracy: {accuracy_score(X_test.reshape(-1,1), y_test)}')\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAttributeError\u001b[0m: 'CatBoostRegressor' object has no attribute 'predict_proba'"
     ]
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "    curr_model = eval(model_name)\n",
    "    print(f'{model_name} roc_auc_score: {roc_auc_score(y_test, curr_model.predict_proba(X_test)[:,1])}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "pal = dict(enumerate(sns.color_palette(\"GnBu\", 2)))\n",
    "X_train2 = pca.fit_transform(X_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Logistic Regression on Selected Features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [],
   "source": [
    "select = ['h1n1_vaccine', 'seasonal_vaccine',\n",
    "          'doctor_recc_h1n1',\n",
    " 'doctor_recc_seasonal',\n",
    " 'h1n1_concern',\n",
    " 'h1n1_knowledge',\n",
    " 'health_worker',\n",
    " 'opinion_h1n1_risk',\n",
    " 'opinion_h1n1_sick_from_vacc',\n",
    " 'opinion_h1n1_vacc_effective',\n",
    " 'opinion_seas_risk',\n",
    " 'opinion_seas_sick_from_vacc',\n",
    " 'opinion_seas_vacc_effective']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_sel=df[select]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(26707, 13)"
      ]
     },
     "execution_count": 129,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sel.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "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>h1n1_vaccine</th>\n",
       "      <th>seasonal_vaccine</th>\n",
       "      <th>doctor_recc_h1n1</th>\n",
       "      <th>doctor_recc_seasonal</th>\n",
       "      <th>h1n1_concern</th>\n",
       "      <th>h1n1_knowledge</th>\n",
       "      <th>health_worker</th>\n",
       "      <th>opinion_h1n1_risk</th>\n",
       "      <th>opinion_h1n1_sick_from_vacc</th>\n",
       "      <th>opinion_h1n1_vacc_effective</th>\n",
       "      <th>opinion_seas_risk</th>\n",
       "      <th>opinion_seas_sick_from_vacc</th>\n",
       "      <th>opinion_seas_vacc_effective</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>26707.000000</td>\n",
       "      <td>26707.000000</td>\n",
       "      <td>24547.000000</td>\n",
       "      <td>24547.000000</td>\n",
       "      <td>26615.000000</td>\n",
       "      <td>26591.000000</td>\n",
       "      <td>25903.000000</td>\n",
       "      <td>26319.000000</td>\n",
       "      <td>26312.000000</td>\n",
       "      <td>26316.000000</td>\n",
       "      <td>26193.000000</td>\n",
       "      <td>26170.000000</td>\n",
       "      <td>26245.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.212454</td>\n",
       "      <td>0.465608</td>\n",
       "      <td>0.220312</td>\n",
       "      <td>0.329735</td>\n",
       "      <td>1.618486</td>\n",
       "      <td>1.262532</td>\n",
       "      <td>0.111918</td>\n",
       "      <td>2.342566</td>\n",
       "      <td>2.357670</td>\n",
       "      <td>3.850623</td>\n",
       "      <td>2.719162</td>\n",
       "      <td>2.118112</td>\n",
       "      <td>4.025986</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.409052</td>\n",
       "      <td>0.498825</td>\n",
       "      <td>0.414466</td>\n",
       "      <td>0.470126</td>\n",
       "      <td>0.910311</td>\n",
       "      <td>0.618149</td>\n",
       "      <td>0.315271</td>\n",
       "      <td>1.285539</td>\n",
       "      <td>1.362766</td>\n",
       "      <td>1.007436</td>\n",
       "      <td>1.385055</td>\n",
       "      <td>1.332950</td>\n",
       "      <td>1.086565</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       h1n1_vaccine  seasonal_vaccine  doctor_recc_h1n1  doctor_recc_seasonal  \\\n",
       "count  26707.000000      26707.000000      24547.000000          24547.000000   \n",
       "mean       0.212454          0.465608          0.220312              0.329735   \n",
       "std        0.409052          0.498825          0.414466              0.470126   \n",
       "min        0.000000          0.000000          0.000000              0.000000   \n",
       "25%        0.000000          0.000000          0.000000              0.000000   \n",
       "50%        0.000000          0.000000          0.000000              0.000000   \n",
       "75%        0.000000          1.000000          0.000000              1.000000   \n",
       "max        1.000000          1.000000          1.000000              1.000000   \n",
       "\n",
       "       h1n1_concern  h1n1_knowledge  health_worker  opinion_h1n1_risk  \\\n",
       "count  26615.000000    26591.000000   25903.000000       26319.000000   \n",
       "mean       1.618486        1.262532       0.111918           2.342566   \n",
       "std        0.910311        0.618149       0.315271           1.285539   \n",
       "min        0.000000        0.000000       0.000000           1.000000   \n",
       "25%        1.000000        1.000000       0.000000           1.000000   \n",
       "50%        2.000000        1.000000       0.000000           2.000000   \n",
       "75%        2.000000        2.000000       0.000000           4.000000   \n",
       "max        3.000000        2.000000       1.000000           5.000000   \n",
       "\n",
       "       opinion_h1n1_sick_from_vacc  opinion_h1n1_vacc_effective  \\\n",
       "count                 26312.000000                 26316.000000   \n",
       "mean                      2.357670                     3.850623   \n",
       "std                       1.362766                     1.007436   \n",
       "min                       1.000000                     1.000000   \n",
       "25%                       1.000000                     3.000000   \n",
       "50%                       2.000000                     4.000000   \n",
       "75%                       4.000000                     5.000000   \n",
       "max                       5.000000                     5.000000   \n",
       "\n",
       "       opinion_seas_risk  opinion_seas_sick_from_vacc  \\\n",
       "count       26193.000000                 26170.000000   \n",
       "mean            2.719162                     2.118112   \n",
       "std             1.385055                     1.332950   \n",
       "min             1.000000                     1.000000   \n",
       "25%             2.000000                     1.000000   \n",
       "50%             2.000000                     2.000000   \n",
       "75%             4.000000                     4.000000   \n",
       "max             5.000000                     5.000000   \n",
       "\n",
       "       opinion_seas_vacc_effective  \n",
       "count                 26245.000000  \n",
       "mean                      4.025986  \n",
       "std                       1.086565  \n",
       "min                       1.000000  \n",
       "25%                       4.000000  \n",
       "50%                       4.000000  \n",
       "75%                       5.000000  \n",
       "max                       5.000000  "
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sel.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "h1n1_vaccine                      0\n",
       "seasonal_vaccine                  0\n",
       "doctor_recc_h1n1               2160\n",
       "doctor_recc_seasonal           2160\n",
       "h1n1_concern                     92\n",
       "h1n1_knowledge                  116\n",
       "health_worker                   804\n",
       "opinion_h1n1_risk               388\n",
       "opinion_h1n1_sick_from_vacc     395\n",
       "opinion_h1n1_vacc_effective     391\n",
       "opinion_seas_risk               514\n",
       "opinion_seas_sick_from_vacc     537\n",
       "opinion_seas_vacc_effective     462\n",
       "dtype: int64"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sel.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_lr=df_sel.dropna(subset=[\"doctor_recc_h1n1\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "h1n1_vaccine                     0\n",
       "seasonal_vaccine                 0\n",
       "doctor_recc_h1n1                 0\n",
       "doctor_recc_seasonal             0\n",
       "h1n1_concern                    73\n",
       "h1n1_knowledge                 104\n",
       "health_worker                  619\n",
       "opinion_h1n1_risk              224\n",
       "opinion_h1n1_sick_from_vacc    236\n",
       "opinion_h1n1_vacc_effective    230\n",
       "opinion_seas_risk              345\n",
       "opinion_seas_sick_from_vacc    370\n",
       "opinion_seas_vacc_effective    297\n",
       "dtype: int64"
      ]
     },
     "execution_count": 133,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_lr.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "h1n1_vaccine                     int64\n",
       "seasonal_vaccine                 int64\n",
       "doctor_recc_h1n1               float64\n",
       "doctor_recc_seasonal           float64\n",
       "h1n1_concern                   float64\n",
       "h1n1_knowledge                 float64\n",
       "health_worker                  float64\n",
       "opinion_h1n1_risk              float64\n",
       "opinion_h1n1_sick_from_vacc    float64\n",
       "opinion_h1n1_vacc_effective    float64\n",
       "opinion_seas_risk              float64\n",
       "opinion_seas_sick_from_vacc    float64\n",
       "opinion_seas_vacc_effective    float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 134,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_lr.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [],
   "source": [
    "y=df_lr.iloc[:,0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    19058\n",
       "1     5489\n",
       "Name: h1n1_vaccine, dtype: int64"
      ]
     },
     "execution_count": 140,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MICE Imputation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fancyimpute import IterativeImputer "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [],
   "source": [
    "mice_imputer = IterativeImputer() \n",
    "X = mice_imputer.fit_transform(df_lr.iloc[:,1:]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.30, random_state=4444)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classification Report and Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {},
   "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>h1n1_vaccine</th>\n",
       "      <th>seasonal_vaccine</th>\n",
       "      <th>doctor_recc_h1n1</th>\n",
       "      <th>doctor_recc_seasonal</th>\n",
       "      <th>h1n1_concern</th>\n",
       "      <th>h1n1_knowledge</th>\n",
       "      <th>health_worker</th>\n",
       "      <th>opinion_h1n1_risk</th>\n",
       "      <th>opinion_h1n1_sick_from_vacc</th>\n",
       "      <th>opinion_h1n1_vacc_effective</th>\n",
       "      <th>opinion_seas_risk</th>\n",
       "      <th>opinion_seas_sick_from_vacc</th>\n",
       "      <th>opinion_seas_vacc_effective</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26702</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26703</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26704</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26705</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26706</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>24547 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       h1n1_vaccine  seasonal_vaccine  doctor_recc_h1n1  doctor_recc_seasonal  \\\n",
       "0                 0                 0               0.0                   0.0   \n",
       "1                 0                 1               0.0                   0.0   \n",
       "3                 0                 1               0.0                   1.0   \n",
       "4                 0                 0               0.0                   0.0   \n",
       "5                 0                 0               0.0                   1.0   \n",
       "...             ...               ...               ...                   ...   \n",
       "26702             0                 0               0.0                   0.0   \n",
       "26703             0                 0               1.0                   1.0   \n",
       "26704             0                 1               0.0                   0.0   \n",
       "26705             0                 0               0.0                   0.0   \n",
       "26706             0                 0               0.0                   0.0   \n",
       "\n",
       "       h1n1_concern  h1n1_knowledge  health_worker  opinion_h1n1_risk  \\\n",
       "0               1.0             0.0            0.0                1.0   \n",
       "1               3.0             2.0            0.0                4.0   \n",
       "3               1.0             1.0            0.0                3.0   \n",
       "4               2.0             1.0            0.0                3.0   \n",
       "5               3.0             1.0            0.0                2.0   \n",
       "...             ...             ...            ...                ...   \n",
       "26702           2.0             0.0            0.0                1.0   \n",
       "26703           1.0             2.0            1.0                2.0   \n",
       "26704           2.0             2.0            0.0                4.0   \n",
       "26705           1.0             1.0            0.0                1.0   \n",
       "26706           0.0             0.0            0.0                1.0   \n",
       "\n",
       "       opinion_h1n1_sick_from_vacc  opinion_h1n1_vacc_effective  \\\n",
       "0                              2.0                          3.0   \n",
       "1                              4.0                          5.0   \n",
       "3                              5.0                          3.0   \n",
       "4                              2.0                          3.0   \n",
       "5                              1.0                          5.0   \n",
       "...                            ...                          ...   \n",
       "26702                          1.0                          3.0   \n",
       "26703                          2.0                          4.0   \n",
       "26704                          2.0                          4.0   \n",
       "26705                          2.0                          3.0   \n",
       "26706                          1.0                          5.0   \n",
       "\n",
       "       opinion_seas_risk  opinion_seas_sick_from_vacc  \\\n",
       "0                    1.0                          2.0   \n",
       "1                    2.0                          4.0   \n",
       "3                    4.0                          1.0   \n",
       "4                    1.0                          4.0   \n",
       "5                    4.0                          4.0   \n",
       "...                  ...                          ...   \n",
       "26702                2.0                          2.0   \n",
       "26703                1.0                          1.0   \n",
       "26704                4.0                          2.0   \n",
       "26705                1.0                          2.0   \n",
       "26706                1.0                          1.0   \n",
       "\n",
       "       opinion_seas_vacc_effective  \n",
       "0                              2.0  \n",
       "1                              4.0  \n",
       "3                              5.0  \n",
       "4                              3.0  \n",
       "5                              5.0  \n",
       "...                            ...  \n",
       "26702                          5.0  \n",
       "26703                          5.0  \n",
       "26704                          5.0  \n",
       "26705                          2.0  \n",
       "26706                          5.0  \n",
       "\n",
       "[24547 rows x 13 columns]"
      ]
     },
     "execution_count": 227,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_lr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training:  84.43%\n",
      "Test set:  85.25%\n",
      "Accuracy_: 85.25\n",
      "roc_auc_score: 88.11\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.88      0.94      0.91      5719\n",
      "           1       0.73      0.55      0.62      1646\n",
      "\n",
      "    accuracy                           0.85      7365\n",
      "   macro avg       0.80      0.74      0.77      7365\n",
      "weighted avg       0.84      0.85      0.84      7365\n",
      "\n"
     ]
    }
   ],
   "source": [
    "lr_model = LogisticRegression(solver=\"lbfgs\")\n",
    "lr_model.fit(X_train, y_train)\n",
    "# predict on test set\n",
    "yhat = lr_model.predict(X_test)\n",
    "# evaluate predictions\n",
    "print(\"Training: {:6.2f}%\".format(100*lr_model.score(X_train, y_train)))\n",
    "print(\"Test set: {:6.2f}%\".format(100*lr_model.score(X_test, y_test)))\n",
    "accuracy = accuracy_score(y_test, yhat)\n",
    "roc_auc = roc_auc_score(y_test, lr_model.predict_proba(X_test)[:,1])\n",
    "print('Accuracy_: %.2f' % (accuracy*100))\n",
    "print('roc_auc_score: %.2f' % (roc_auc*100))\n",
    "print(classification_report(y_test, yhat))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 354,
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_confusion_matrix(model, threshold=0.6):\n",
    "    # Predict class 1 if probability of being in class 1 is greater than threshold\n",
    "    # (model.predict(X_test) does this automatically with a threshold of 0.5)\n",
    "   \n",
    "    y_predict = (model.predict_proba(X_test)[:, 1] >= threshold)\n",
    "    confusion = confusion_matrix(y_test, y_predict)\n",
    "    plt.figure(dpi=80)\n",
    "    sns.heatmap(confusion, cmap=plt.cm.Reds, annot=True, square=True, fmt='d',\n",
    "           xticklabels=['Take Vaccine', 'Not Take Vaccine'],\n",
    "           yticklabels=['Take Vaccine', 'Not Take Vaccine']);\n",
    "    plt.xlabel('prediction')\n",
    "    plt.ylabel('actual');\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 480x320 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "make_confusion_matrix(lr_model);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.13029188, 0.26684702, 0.09858978, ..., 0.49397668, 0.22401546,\n",
       "       0.0533654 ])"
      ]
     },
     "execution_count": 192,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_model.predict_proba(X_test)[:,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 1.9715433 ,  2.16502768, -0.96553423, -0.14485422,  0.06348789,\n",
       "         0.54119514,  0.38042418, -0.0089533 ,  0.65047906, -0.0300498 ,\n",
       "        -0.03827215, -0.08272063]])"
      ]
     },
     "execution_count": 305,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_model.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['seasonal_vaccine',\n",
       " 'doctor_recc_h1n1',\n",
       " 'doctor_recc_seasonal',\n",
       " 'h1n1_concern',\n",
       " 'h1n1_knowledge',\n",
       " 'health_worker',\n",
       " 'opinion_h1n1_risk',\n",
       " 'opinion_h1n1_sick_from_vacc',\n",
       " 'opinion_h1n1_vacc_effective',\n",
       " 'opinion_seas_risk',\n",
       " 'opinion_seas_sick_from_vacc',\n",
       " 'opinion_seas_vacc_effective']"
      ]
     },
     "execution_count": 228,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "select[1:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('seasonal_vaccine', 1.994603648426964),\n",
       " ('doctor_recc_h1n1', 2.221215880625276),\n",
       " ('doctor_recc_seasonal', -1.0529970588122486),\n",
       " ('h1n1_concern', -0.08436955386268696),\n",
       " ('h1n1_knowledge', 0.039013202286953534),\n",
       " ('health_worker', 0.49809820062037524),\n",
       " ('opinion_h1n1_risk', 0.36705834085549743),\n",
       " ('opinion_h1n1_sick_from_vacc', -0.005149009745327946),\n",
       " ('opinion_h1n1_vacc_effective', 0.6255080949515864),\n",
       " ('opinion_seas_risk', -0.008981373646964863),\n",
       " ('opinion_seas_sick_from_vacc', -0.05182779461435615),\n",
       " ('opinion_seas_vacc_effective', -0.09867030268637103)]"
      ]
     },
     "execution_count": 229,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(zip(select[1:], lr_model.coef_[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_coef=pd.DataFrame(zip(select[1:], lr_model.coef_[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_coef.columns = [\"Feature\",\"Coefficient\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 341,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_coef.to_csv(\"feature_coef.csv\", index=False)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_coef.to_csv(\"Data/lr_features.csv\", index= False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.94      0.79      0.86      5719\n",
      "           1       0.53      0.83      0.65      1646\n",
      "\n",
      "    accuracy                           0.80      7365\n",
      "   macro avg       0.74      0.81      0.75      7365\n",
      "weighted avg       0.85      0.80      0.81      7365\n",
      "\n"
     ]
    }
   ],
   "source": [
    "final_model = LogisticRegression(max_iter=1000, solver=\"saga\")\n",
    "\n",
    "X_oversampled_train, y_oversampled_train = RandomOverSampler(random_state=0).fit_sample(X_train, y_train)\n",
    "\n",
    "final_model.fit(X_oversampled_train, np.array(y_oversampled_train).ravel())\n",
    "y_pred = final_model.predict(X_test)\n",
    "\n",
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 1.99453922,  2.22135133, -1.05302193, -0.08449893,  0.03924903,\n",
       "         0.49816263,  0.36720353, -0.00513484,  0.62556286, -0.00902012,\n",
       "        -0.05172694, -0.0984812 ]])"
      ]
     },
     "execution_count": 203,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_model.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training:  79.16%\n",
      "Test set: 100.00%\n"
     ]
    }
   ],
   "source": [
    "print(\"Training: {:6.2f}%\".format(100*final_model.score(X_oversampled_train, y_oversampled_train)))\n",
    "print(\"Test set: {:6.2f}%\".format(100*final_model.score(X_test, y_pred)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predict H1N1 Vaccine with 12 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_try = pd.DataFrame({'seasonal_vaccine': 1,\n",
    "                      'doctor_recc_h1n1': 1,\n",
    " 'doctor_recc_seasonal': 1,\n",
    " 'h1n1_concern': 3,\n",
    " 'h1n1_knowledge': 2,\n",
    " 'health_worker': 0,\n",
    " 'opinion_h1n1_risk': 5,\n",
    " 'opinion_h1n1_sick_from_vacc': 4,\n",
    " 'opinion_h1n1_vacc_effective': 5,\n",
    " 'opinion_seas_risk': 1,\n",
    " 'opinion_seas_vacc_effective': 3,\n",
    " 'opinion_seas_sick_from_vacc': 1}, index = range(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['seasonal_vaccine', 'doctor_recc_h1n1', 'doctor_recc_seasonal',\n",
       "       'h1n1_concern', 'h1n1_knowledge', 'health_worker', 'opinion_h1n1_risk',\n",
       "       'opinion_h1n1_sick_from_vacc', 'opinion_h1n1_vacc_effective',\n",
       "       'opinion_seas_risk', 'opinion_seas_vacc_effective',\n",
       "       'opinion_seas_sick_from_vacc'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 240,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_try.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9623989763698817"
      ]
     },
     "execution_count": 246,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_model.predict_proba(X_try)[0][1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "metadata": {},
   "outputs": [],
   "source": [
    "example_data = df_drop[select[1:]].iloc[:100]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {},
   "outputs": [],
   "source": [
    "example_data.to_csv('Data/example_data.csv', index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {},
   "outputs": [],
   "source": [
    " with open(\"Data/lr_model.pickle\", \"wb\") as pfile:\n",
    "        pickle.dump(lr_model, pfile)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "metadata": {},
   "outputs": [],
   "source": [
    "    with open(\"Data/lr_model.pickle\", \"rb\") as pfile:\n",
    "        exec(\"lr_model_2 = pickle.load(pfile)\")\n",
    "#https://github.com/vanessa920/H1N1_seasonal_flu_shot_prediction/blob/eda/Data/lr_model.pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.03760102, 0.96239898]])"
      ]
     },
     "execution_count": 259,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_model_2.predict_proba(X_try)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 7338 entries, 0 to 7337\n",
      "Data columns (total 13 columns):\n",
      " #   Column                       Non-Null Count  Dtype  \n",
      "---  ------                       --------------  -----  \n",
      " 0   h1n1_vaccine                 7338 non-null   int64  \n",
      " 1   seasonal_vaccine             7338 non-null   int64  \n",
      " 2   doctor_recc_h1n1             7338 non-null   float64\n",
      " 3   doctor_recc_seasonal         7338 non-null   float64\n",
      " 4   h1n1_concern                 7327 non-null   float64\n",
      " 5   h1n1_knowledge               7313 non-null   float64\n",
      " 6   health_worker                7332 non-null   float64\n",
      " 7   opinion_h1n1_risk            7330 non-null   float64\n",
      " 8   opinion_h1n1_sick_from_vacc  7336 non-null   float64\n",
      " 9   opinion_h1n1_vacc_effective  7318 non-null   float64\n",
      " 10  opinion_seas_risk            7332 non-null   float64\n",
      " 11  opinion_seas_sick_from_vacc  7336 non-null   float64\n",
      " 12  opinion_seas_vacc_effective  7325 non-null   float64\n",
      "dtypes: float64(11), int64(2)\n",
      "memory usage: 745.4 KB\n"
     ]
    }
   ],
   "source": [
    " df_drop[select].info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 350,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA5QAAAJcCAYAAACG+uslAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAACjI0lEQVR4nOzdfXjP9f///9tOGnZStpmzZCWM2fnYxmgnqGFofdWbkJMtZznLyczJJBLWTBgbOY1OUEhRiZSQ80h5F6Ilb+enO2N28vtjv70+rQ3by2zD9XIcOw6v58njeX8+Xq8dx24ej9fjaZKTk5MjAAAAAACKybSsCwAAAAAA3J8IlAAAAAAAoxAoAQAAAABGIVACAAAAAIxCoAQAAAAAGIVACQAAAAAwCoESAAAAAGAU87IuAMDduXw5VdnZ5fNxsvb21rp4MaWsy3io0Oeljz4vffR56aPPSx99Xvro88KZmprI1tbqlvsJlMB9Ljs7p9wGSknlurYHFX1e+ujz0keflz76vPTR56WPPi8+prwCAAAAAIzCCCUAAAAKsLWykLllBcNrBwebMqym6DLTbuhyakZZlwE8NAiUAAAAKMDcsoIOO/Yp6zKKzTlpvkSgBEoNU14BAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqB8SOXk5JR1CQAAAADucwTKcm716tVycnLSmTNnSqS9jIwMTZ06VZ9//nmJtPegcHJy0ty5c4t8/NmzZ9W4cWPt3bv3rq89ePBgjR079q7bAQAAAEobgfIhc+nSJS1evFiZmZllXcp969y5c4qIiFBycvJdtZOTk6OYmBh9/fXXJVQZAAAAULoIlEAR5eTkaMOGDXrhhRd07ty5u2rr+PHjioiI0PLly1WxYsUSqhAAAAAoXQTKciQ7O1tz585VYGCg3N3dNWDAAF29ejXfMd9//706d+4sT09PNW3aVOPGjdPly5fzHXPgwAH16tVLXl5eatq0qSIjI3Xx4kX9/fffCggIkCSNHj1awcHBRW539uzZCgkJ0axZs+Tr66vnnntOqampRbqvqKgo9e7dW9HR0fL29lbnzp2Vk5Oj7OxsJSYmqlWrVnJxcVFISIhWrVpV4Py1a9fq+eefl7u7u4KDgzVr1ixlZWXd8X6LKzk5WaNHj1aTJk3k5eWloUOH6tKlS4b9p06dUmRkpEJCQjRt2rRb3mt4eLhWrVqlZ599Vi4uLurYsaN++OGHfMe98cYbSk5O1ooVK2Rvb1/sWgEAAIDygEBZjrzzzjuaM2eOOnXqpPj4eNna2mr69OmG/Z9++qn69Omj2rVra+bMmXr99de1ZcsWvfLKK0pPT5ckHT58WN26dVNWVpZiYmIUHR2tvXv3qm/fvqpataoSEhIkSf3791d8fHyR25WkkydPasuWLYqLi9PQoUNlZWVV5HvbtWuX/ve//yk+Pl59+/aViYmJJkyYoPj4eIWFhSkxMVFBQUGKjo7WsmXLDOd98MEHGjVqlNzc3DRnzhz17NlT7733nqFfbne/xbVkyRJlZWVp1qxZGjZsmDZt2qRJkyYZ9tva2uqrr77SuHHjVKlSpVu2c/DgQS1evFhDhgzRnDlzZGZmpsGDB+ebIjtu3DitXLlSDRs2LHadAAAAQHlhXtYFINe1a9e0bNky9e7dWwMHDpQktWjRQmfPntUPP/yg7OxsxcXFKTAwUDExMYbzGjRooBdffFGrV69W165dlZiYKHt7ey1YsEAWFhaSpMqVK2v8+PE6ffq0nJ2dJUm1a9eWs7NzkduVpMzMTEVFRcnX17fY95eZmak333xTtWrVkiSdOHFCK1euVGRkpHr37i1Jat68ubKysjRz5kx16tRJFSpU0Jw5cxQSEqKJEycajrl27Zq2b9+unJyc295vUlKSHB0di1yjh4eHoQ+aNm2qgwcPatu2bYb9VlZWRQrRycnJWrNmjZ544glJkqWlpbp166Zdu3apVatWknIXAQIAAADud4xQlhMHDhzQzZs31bJly3zb27RpI0n63//+pwsXLqhdu3b59ru5ucnR0VG7du2SJO3bt08BAQGGcCVJzZo106ZNmwoNVydOnChSu3nq169v1P1ZWloawqQk7dy5Uzk5OQoKClJmZqbhJzg4WMnJyfr555914sQJXbx4Ua1bt87X1sCBA/XRRx/JxMSk2Pd7O40bN873ulatWkYtvOPg4GAIk5JUvXp1Sco32gsAAAA8CBihLCfyvitpZ2eXb7uDg4Ok3O9X/vP1P9nb2yslJUWSdOXKlQJt3M6VK1eK1K4kmZmZydbWtsht/1OVKlUKvW5ISEihx587d07m5uaGOm6luPd7O/+exmpqamro97tpx8TERJKMagsAAAAozwiU5UReULtw4YJq165t2J4XvPJCyfnz5wuce/78ebm7u0uSrK2t8y0kI+UGma1bt8rV1bXAuY899liR2i1pNjY2knTLVU5r1aplqOnf93PhwgUdPXpUXl5ed7xfFrwBAAAA7h2mvJYTnp6eqlixor766qt827ds2SJJevzxx1WlShWtX78+3/6ff/5ZJ0+elJeXlyTJ29tb27Zt082bNw3H7N+/X3379tWJEydkapr/La9Tp06R2i1pedNLr169KldXV8PP6dOnNWvWLKWnp6tOnTqqXLmyvv3223znrlixQgMGDJB05/sFAAAAcO8wQllOWFlZacCAAXr33XdVsWJF+fj46LvvvjMESlNTUw0dOlTjxo1TZGSkQkNDdfbsWc2cOVNPPfWUwsLCJEkDBgxQ586d1a9fP3Xr1k1paWmKi4uTj4+PvLy8dP36dZmYmOjHH3/U008/LXd39yK1W9IaNGig0NBQjRkzRidPnlTDhg117NgxxcXFqVGjRqpZs6ak3O9LTp48Wba2tgoODtaRI0c0f/58hYeHq0KFCne8XwAAAAD3DoGyHOnbt68sLS21dOlSLV68WJ6enho1apQmTJggSXrxxRdlaWmp9957TwMGDNBjjz2mli1b6vXXX5elpaUkycXFRUuXLtWMGTM0ZMgQPfroowoODtbw4cNlamoqS0tL9e/fX0uWLNHWrVu1ffv2IrV7L0ydOlWJiYlavny5zp49qypVqqhTp04aPHiw4Zju3burUqVKWrRokT7++GPVrFlTgwcPVq9evYp0vwAAAADuHZOcnJycsi4CgPEuXkxRdnb5/DV2cLDR+fPFXykXxqPPSx99Xvro89Lh4GCjw459yrqMYnNOmv9AfD74nJc++rxwpqYmsre3vuV+RihhlJycHGVlZd3xODMzM8OCQqUtOzu7SCur5q0mCwAAAKB4+EsaRtm9e7deeeWVOx43ZcoUvfDCC6VQUUFjxozRmjVr7njc77//XgrVAAAAAA8eAiWM0qhRI33yySd3PK5WrVqlUE3hBg4cqK5du5bZ9QEAAIAHHYESRrG2ti70uZblSa1atco00AIAAAAPOpbBBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKPw2BAAAAAUkJl2Q85J88u6jGLLTLtR1iUADxUCJQAAAAq4nJohpWZIkhwcbHT+fHIZVwSgPGLKKwAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABiFQAkAAAAAMAqBEgAAAABgFPOyLgAAAAB3x9bKQuaWFe7pNRwcbEq8zcy0G7qcmlHi7QIoPQRKAACA+5y5ZQUdduxT1mUUm3PSfIlACdzXmPIKAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABiFQAkAAAAAMAqBEgAAAABgFAIlytzq1avl5OSkM2fOlHUpxXY/1w4AAADcLQIlcBcCAwO1YsUK2dnZlXUpAAAAQKkzL+sCgPuZnZ0dYRIAAAAPLUYoy5lffvlFPXr0kLe3tzw9PdWzZ08dOHDAsH/Pnj3q2rWr3N3d5evrq3HjxunatWv52ti1a5d69+6tJk2ayMXFRS1btlR8fLyys7MNx3zxxRfq0KGD3Nzc1LRpU40YMUJnz5417M/MzNSSJUvUrl07ubm5qWXLlkpISFBWVpbhmO7du2v8+PGaN2+eAgIC5Orqqs6dO+vQoUP56vn666/VpUsXeXp6ysXFRW3atNGHH35odB+dOnVKDRo00Mcff5xv++nTp9WgQQN9+umnkqSTJ09q5MiRat68uRo1aqRmzZopKipKV69eNZyTk5OjJUuWKCQkRG5ubnruuee0bNmyfO1+//336ty5szw8PNSiRQu99dZbSk1NlVRwymtUVJTCw8O1atUqPfvss3JxcVHHjh31ww8/FLiHoUOHqkmTJvLw8FB4eLiOHTtmdJ8AAAAAZYFAWY6kpKQoIiJCtra2mj17tmbMmKH09HRFREQoJSVFe/bsUa9evWRlZaWZM2cqMjJS3333ncLDw5WZmSlJ+vXXX9W7d2/Z29vr3XffVUJCgry9vTV79mx99dVXkqR9+/YpMjJSzz77rBYsWKCoqCjt3LlTI0aMMNQyduxYxcbGqk2bNkpISNDzzz+vOXPmKDo6Ol/NGzZs0JYtWxQdHa24uDhduHBBQ4YMMYTXzZs3a/DgwXJzc9PcuXM1e/Zs1apVS2+++aZ+/vlno/rp8ccfV+PGjbV+/fp829evX68KFSroueeeU3p6urp166Y///xTEyZM0MKFC9W9e3d9/vnnmjFjhuGcmJgYxcTE6Nlnn1ViYqLat2+vyZMn64MPPpAkbdmyRX379lXVqlU1c+ZMDR06VOvWrdOYMWNuWd/Bgwe1ePFiDRkyRHPmzJGZmZkGDx6s5ORkSdKlS5fUpUsX/fbbb5owYYJiY2OVmpqql19+WadOnTKqTwAAAICywJTXcuTYsWO6fPmyXnnlFXl5eUmS6tSpoxUrVig1NVXTp0/X008/rcTERJma5v5fgLOzs8LCwrRhwwZ16NBBR44cUfPmzRUTEyMTExNJkr+/v7799lvt2bNHbdu21b59+1SxYkX16dNHFhYWkqTKlSvr0KFDysnJ0bFjx7R27VpFRkYqPDzc0EbFihUVGxurnj17qn79+pKkrKwsLViwQNbW1pKk1NRUjRo1SkeOHFGDBg30xx9/6IUXXtDo0aMN9+np6SlfX1/t3r1bbm5uRvVVx44dNX78eJ07d05Vq1aVlBtug4ODZW1trV9//VWPP/64YmJiVKtWLUmSn5+fDh48qD179kiSrl27pvfff189e/bUsGHDJEnNmjXTmTNnDCPBs2fPlouLi2bNmmW4dk5OjhYtWqSUlJRCa0tOTtaaNWv0xBNPSJIsLS3VrVs37dq1S61atdLSpUt19epVrVy5UtWrV5ckNW/eXK1bt1ZCQoLeeusto/oEAAAAKG0EynKkXr16srOzU79+/RQSEqIWLVrI399fI0eOVHp6ug4ePKg+ffooOzvbMAJYr1491axZUzt27FCHDh0UFhamsLAw3bhxQydOnNBff/2lw4cPKysrSzdv3pQkNWnSRDNmzFD79u317LPPKiAgQM2bN1dAQIAkGQJXaGhovvo6dOig2NhY7dmzxxAonZycDGFSkqpVqyZJSktLkyT16dNHUm7QzKsnb0psXj3GCAkJ0aRJk/Tll1+qR48e+vPPP/Xrr79q0KBBkqRGjRrpww8/VHZ2tv78808lJSXp2LFjOn78uKGNAwcOKDMzU61bt87Xdl6gu379ug4fPqzXX3893/5OnTqpU6dOt6zNwcHBECYlGUJjenq6JOnHH39Uo0aNVKVKFcPIsrm5ufz9/bVjxw5juwQAAAAodQTKcsTKykoffPCBEhIS9OWXX2rFihWqWLGiOnbsqL59+yo7O1uJiYlKTEwscO65c+ck5YagSZMm6bPPPlNmZqZq1aolT09PmZubKycnR1LuCOH8+fO1ZMkSLV68WPPnz1eVKlXUr18/de/e3fAdQ3t7+3zXyHudN3VTkipWrJjvmLyR07zAe+nSJb3xxhvatGmTTExM5OjoKG9vb0ky1GMMGxsbBQUFacOGDerRo4e++OIL2dnZqUWLFoZjFi9erMTERF25ckVVqlSRi4uLKlWqZAi7V65cKfQ+81y9elU5OTnFXnSnUqVK+V7njRTn9cmVK1eUlJSkRo0aFTj3kUceKda1AAAAgLJEoCxn6tSpo3feeUdZWVn6+eef9dlnn+mjjz5S1apVZWJiot69e6tNmzYFzrOyspIkTZ48WRs3btTMmTPVtGlTWVpaSpKaNm2a7/gWLVqoRYsWSk9P186dO/X+++/rrbfekqenpx599FFJ0sWLFw0jjpJ0/vx5SZKtrW2R72fEiBE6ceKElixZIk9PT1lYWCg9PV2rVq0qXscUomPHjurfv79Onz6tL7/8Um3btpW5ee5H+vPPP9fUqVMVGRmpsLAwQygcMmSIDh8+LCk3lEq5obd27dqGdk+ePKnTp0/L2dlZknT58uV8101JSdFPP/0kDw8Po+q2traWn59fvu+sAgAAAPcjFuUpR7755hv5+fnp/PnzMjMzk6enpyZMmKBHH31UV69elbOzs/7880+5uroafp566im9++67OnjwoKTcBXeaNm2qli1bGsLkL7/8okuXLhlGyN555x116tRJOTk5qlSpkoKCgjRq1ChJ0pkzZ+Tj4yMpdyXYf8p7nTfCWBT79u1TSEiIfH19Dd/X3Lp1qyTlW3XWGC1atJCtra0WLFigY8eOqWPHjvmua2trq/DwcEOYTE1N1b59+wzXdXd31yOPPKItW7bkazchIUFjxoyRtbW1GjRooG+//Tbf/k2bNikiIiLfSG1x+Pj46MSJE3r66afzvZcrV64ssNAQAAAAUJ4xQlmOeHl5KScnR6+99pr69OkjKysrffnll0pJSdGzzz6r5s2bq1+/foqKilLbtm2VkZGh9957T0eOHDEEQjc3N3311VdasWKFnnrqKf32229KSEiQiYmJ4Tt8zZo108KFCxUVFaUOHTro5s2bWrBggWxtbeXj46NHH31UHTp0MKwy6+npqZ9++kmJiYnq2LGj6tatW+R7cnNz07p169SwYUNVq1ZN+/fv1/z58/PVY6xHHnlE7dq100cffaQnn3wy3wI/bm5u+uijjxQTE6PAwECdOXNGixYt0oULFwwB087OTt26ddPChQtlbm6uxo0ba9++fVqzZo0mTZokSRo8eLBee+01jRgxQh07dtSZM2c0ffp0Pf/886pZs6ZRdffq1Utr165V79691bNnTz366KNau3atPvvsM7399tt31ScAAABAaSJQliP29vZauHChZsyYobFjxyo9PV316tXT7Nmz1aRJE0nSggULFB8fr0GDBqlChQpydXXV+++/b1gkJyoqSjdv3lRcXJwyMjJUq1Yt9e/fX8eOHdP333+v7Oxs+fv7Ky4uTgsWLNDAgQNlYmIib29vvf/++4bprlOmTJGjo6NWr16txMRE1axZU4MGDVJERESx7mnq1KmaNGmSJk6cKEl68skn9eabb2rdunXat2/fXfdZx44dtXz5crVv3z7f9rCwMP3999/69NNPtXz5clWrVk0BAQF6+eWXFR0drRMnTuipp55SZGSk7OzstHLlSs2fP1+Ojo56++23FRYWJklq2bKl5s6dq/j4eA0YMED29vZ66aWX9Nprrxldc7Vq1fTxxx8rLi5O0dHRunnzpurUqaO4uDi1a9furvoDAAAAKE0mOXezMgqAMnfxYoqys8vnr7GDg43OnzduajCMQ5+XPvq89NHnBTk42OiwY5+yLqPYnJPm817eAp/z0kefF87U1ET29ta33M8IJcqVvMdo3I6pqalhNVkAAAAAZYdAiXKlsEdp/FtYWJimTp1aCtUAAAAAuB0CJcqVTz755I7HFOexJQAAAADuHQIlyhVXV9eyLgEAAABAEfFFNAAAAACAUQiUAAAAAACjECgBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBSeQwkAAHCfy0y7Ieek+WVdRrFlpt0o6xIA3CUCJQAAwH3ucmqGlJpxz9p3cLDR+fPJ96x9APcvprwCAAAAAIxCoAQAAAAAGIVACQAAAAAwCoESAAAAAGAUAiUAAAAAwCgESgAAAACAUQiUAAAAAACj8BxKAAAeELZWFjK3rFDWZRRbZtqN3OcoAgDuOwRKAAAeEOaWFXTYsU9Zl1FszknzJQIlANyXmPIKAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABiFQAkAAAAAMAqBEgAAAABgFAIlYIScnJyyLgEAAAAoc+UqUEZFRal169b3/JySsmvXLjk5OWnv3r1FPue7775To0aN7vraqampCg4O1meffXbXbT0IUlJSNGDAALm7u6tJkyY6efKkli1bpubNm8vNzU3z588vketkZGRo6tSp+vzzzw3byvIzCAAAAJQl87Iu4J8GDBig1NTUe35OWdm/f7+GDx9+16NbeeHp1KlTJVTZ/e/zzz/X5s2bNX78eNWrV0/29vaaMmWKAgIC1Lt3bz3xxBMlcp1Lly5p8eLFmjJlimHb/fQZBAAAAEpSuQqUtWvXLpVzStv169e1aNEizZ07V1ZWVnfV1rZt2zRp0iRduXKlZIp7QOT1x8svvywTExOdPXtWWVlZatWqlZo0aXJPr30/fAYBAACAe6HEprxmZmZqyZIlateundzc3NSyZUslJCQoKytLktS9e3eNHTtW8fHx8vPzU5MmTTRs2DBdunTJ0Ma/pw4GBwcrPj5eU6dOVbNmzeTu7q7w8HAlJSXd8pw71ZFXy/jx4zVv3jwFBATI1dVVnTt31qFDh4y69z/++EO9evWSm5ub/P39FRcXl+96GzZs0NKlSzV69Gh169at0DaKcq+SFBERIRcXF7333ntG1Zqeni5PT09Nnz493/br16/L09NTs2bNkpQ7EvfGG28oKChILi4u8vHx0aBBgwqMiq5du1bPP/+83N3dFRwcrFmzZuW79wMHDqhXr17y8vJS06ZNFRkZqYsXLxar5uzsbCUmJqpVq1ZycXFRSEiIVq1aZdjfvXt3vfvuu5KkBg0aKDg4WM8884wkacyYMXJycjIc+8033+iFF16Qq6urmjdvrmnTpikjIyPf9W5V899//62AgABJ0ujRoxUcHCwp/2dw9OjRatGihbKzs/O1OWbMGAUHBxtGp/fs2aOuXbvK3d1dvr6+GjdunK5du1asfgEAAADKWokFyrFjxyo2NlZt2rRRQkKCnn/+ec2ZM0fR0dGGYzZu3Kj169dr4sSJioqK0vbt2xUREVHgj+9/WrJkiU6cOKEpU6Zo0qRJ+uWXXzR69Oi7qkPKDXlbtmxRdHS04uLidOHCBQ0ZMuS2tdzK5MmT5ePjo3nz5ikkJETz5s3TypUrDft9fX21efNmde3a9bbtFOVe161bp+nTp8vOzq7YdUpSpUqV9Oyzz2rDhg35tm/ZskVpaWnq0KGDcnJyFBERoZ07d2rEiBFauHChBg4cqO3bt2vChAmGcz744AONGjVKbm5umjNnjnr27Kn33nvPEFYPHz6sbt26KSsrSzExMYqOjtbevXvVt2/fYtU8YcIExcfHKywsTImJiQoKClJ0dLSWLVsmSXrjjTf0n//8R5K0YsUKzZgxQwkJCZKk/v37a8WKFZJyp8UOHDhQ9erV05w5c9SvXz+tWLFCw4cPN1zrdjVXrVo1X7vx8fEFau3YsaPOnTuX73u1GRkZ2rRpk9q3by8TExPt2bNHvXr1kpWVlWbOnKnIyEh99913Cg8PV2ZmZrH6BgAAAChLJTLl9ejRo1q7dq0iIyMVHh4uSfL391fFihUVGxurnj17SsodHVu0aJFq1KghSbKzs1O/fv20detWBQYGFtp25cqVNXfuXJmZmUmS/vrrL82ePVvJycmysbEpdh3169eXJGVlZWnBggWytraWlLvIzahRo3TkyBE1aNCgWPffq1cv9e/fX5Lk5+enzZs3a+fOnerSpYsk6fHHHy9SO0W517z670bHjh21du1aHTx4UO7u7pJyA7a7u7uefPJJnTlzRlZWVho3bpy8vLwk5Ybiv/76S5988omk3FHDOXPmKCQkRBMnTpQkNW/eXNeuXdP27duVk5OjxMRE2dvba8GCBbKwsDDc4/jx45WUlCRHR8c71nrixAmtXLlSkZGR6t27t+E6WVlZmjlzpjp16qS6deuqevXqkiQPDw9J0pkzZyTlTkf18PBQTk6OYmNjFRQUpGnTphnar169ul577TXt27dP3t7et6359OnTcnZ2NrSb9+9/8vX1VY0aNbR+/Xr5+PhIyp2mfPXqVXXo0EGSNH36dD399NNKTEyUqWnu/+k4OzsrLCxMGzZsMBwHAAAAlHclMkK5Z88eSVJoaGi+7Xl/GOft9/b2NoRJSQoMDJSFhcVtV0l1d3c3BCxJhuCQlpZmdB2S5OTkZAiTklStWrVbtnsnjRs3NvzbxMREjz/+uJKTk4vdTnHu9W74+fmpWrVqWr9+vaTcRX6+//57Qz9Vr15dy5Ytk6enp/7++29t375dy5Yt0/79+3Xz5k1JuUHv4sWLBVY3HThwoD766COZmJho3759CggIMAQzSWrWrJk2bdpUpDApSTt37lROTo6CgoKUmZlp+AkODlZycrJ+/vnnIrVz/PhxnTlzRsHBwfnaadGihR555BHt2LFDku66ZhMTE7Vv314bN240jDauX79ejRo10tNPP6309HQdPHhQgYGBys7ONtRRr1491axZ01AHAAAAcD8okRHKq1evSpLs7e3zbc97nReuqlatmm+/iYmJ7OzsbvvdsYoVK+Z7nTeiU9hKqUWt43btGjPltVKlSgXaMqad4tzr3TA1NVVoaKi++OILRUVFadOmTcrOzlbbtm0Nx6xbt05xcXE6ffq0KleurIYNG6pixYqGWvIWwfl3X//TlStXjJ6a+882JCkkJKTQ/efOnStWO9HR0QWmP/+znZKouWPHjpo/f75+/PFHNWnSRN9++62GDh0qSbp27ZrhO6GJiYm3rAMAAAC4H5RIoHz00UclSRcvXjSM9EnS+fPnJUm2traSVGBl0pycHF28ePGu/4Avbh3IDT0LFy7U/v37tWHDBjVv3tzwPuzdu1ejRo1Sjx491KtXL0NfxsTE6MCBA5JkmIL7z0WVJOnChQs6evSovLy8ZG1tXWB/dna2tm7dKldX19uG0Tx511m+fHmBwC1JtWrVKtL95rUzevRoeXt7F9if99m4U81FUbduXTVq1EhfffWVUlJSdOPGDbVr106SZGVlJRMTE/Xu3Vtt2rQpcO7drgIMAAAAlKYSmfKa912xL774It/2vNd5f8Dv378/32jkt99+q5s3b8rPz68kyihyHcid8tugQQOtW7dOO3bsyPe9vZ9++knZ2dkaNGiQIUxmZWVpx44dhpHXOnXqqHLlyvr222/ztbtixQoNGDBAUm5/b9u2zTBNVsr9DPTt21cnTpwoUp1504mvXr0qV1dXw8/p06c1a9YspaenF6mdp59+WnZ2djp16lS+dmxtbRUbG6s//vijSDXnjRrfSYcOHfT999/ryy+/lL+/v6pUqSIpN7A6Ozvrzz//zFfHU089pXfffVcHDx4sUvsAAABAeVAiI5T16tVThw4dNGPGDMNjKX766SclJiaqY8eOqlu3rqTchW/69Omjvn376sKFC4qNjZW/v3+JBcqi1oFcHTt2VGxsrCpVqqSWLVsatru5uUmSJk2apOeff15Xr17VBx98oN9++005OTm6fv26KlasqIEDB2ry5MmytbVVcHCwjhw5ovnz5ys8PFwVKlTQgAED1LlzZ/Xr10/dunVTWlqa4uLi5OPjY1js504aNGig0NBQjRkzRidPnlTDhg117NgxxcXFqVGjRqpZs2aR2jEzM9PQoUP15ptvytTUVM8884yuXr2qWbNmKTk52bDAzp1qvn79ukxMTPTjjz/q6aefNixq9G+hoaGKiYnRpk2b8i0CJElDhgxRv379FBUVpbZt2yojI0Pvvfeejhw5olGjRhXpfgAAAIDyoEQCpSRNmTJFjo6OWr16tRITE1WzZk0NGjRIERERhmN8fHzk6empkSNHytzcXKGhoRoxYkRJlVDkOpArNDRUsbGxevbZZ1WhQgXDdl9fX40fP16LFy/W+vXrVaVKFfn6+io+Pl6vvfaa9u7dq+bNm6t79+6qVKmSFi1apI8//lg1a9bU4MGD1atXL0mSi4uLli5dqhkzZmjIkCF69NFHFRwcrOHDhxd5pE+Spk6dqsTERC1fvlxnz55VlSpV1KlTJw0ePLhY9/uf//xH1tbWWrBggT788ENZW1sbnofq4OBQpJotLS3Vv39/LVmyRFu3btX27dsLvVaVKlXk7++vvXv3qlWrVvn2BQQEaMGCBYqPj9egQYNUoUIFubq66v333y+RVXwBAACA0mKSU9IrvtxC9+7dZWZmpiVLlpTG5YCHxsWLKcrOLpVf42JzcLDR+fPFX/EYxqPPS1956nMHBxsdduxT1mUUm3PS/GL1YXnq84cFfV766PPSR58XztTURPb21rfcX2IjlA+SrKysO66samJiku8RH+XB/VR3dnZ2kVbCNTfnIwoAAACUV/y1XojWrVvr1KlTtz3Gx8dHy5YtK6WKiuZ+qnvMmDFas2bNHY/7/fffS6EaAAAAAMYotUBZHkJMUSUkJCgjI+O2x5THxzvcT3UPHDhQXbt2LesyAAAAANwFRigL4eTkVNYlGOV+qrtWrVpFfoYkAAAAgPKpRJ5DCQAAAAB4+BAoAQAAAABGIVACAAAAAIxCoAQAAAAAGIVACQAAAAAwCoESAAAAAGAUHhsCAMADIjPthpyT5pd1GcWWmXajrEsAABiJQAkAwAPicmqGlJpR1mUAAB4iTHkFAAAAABiFQAkAAAAAMAqBEgAAAABgFAIlAAAAAMAoBEoAAAAAgFEIlAAAAAAAoxAoAQAAAABG4TmUAACUU7ZWFjK3rFDWZRRbZtqN3GdiAgAeeARKAADKKXPLCjrs2Kesyyg256T5EoESAB4KTHkFAAAAABiFQAkAAAAAMAqBEgAAAABgFAIlAAAAAMAoBEoAAAAAgFEIlAAAAAAAoxAoAQAAAABGIVACAAAAAIxCoAQAAAAAGIVACQAAAAAwCoESZSYnJ6esSwAAAABwFwiUD7Hg4GCNHTv2nl9n165dcnJy0t69eyVJZ8+eVd++fXXq1KlSr6Ukde/eXT179izrMgAAAIAyQ6BEqdu5c6e+++67si4DAAAAwF0iUAIAAAAAjEKgfMjdvHlTU6dOVbNmzeTh4aHw8HCdPHnSsH/Pnj3q2rWr3N3d5evrq3HjxunatWv52ti1a5d69+6tJk2ayMXFRS1btlR8fLyys7MLXG/16tWKjIyUJLVs2VJRUVFFruVONm3aJCcnJ/3++++GbR988IGcnJy0YcMGw7bt27fLycnJ0PaBAwfUq1cvNWnSRE2aNNGQIUP0999/56vZ1dVVH3/8sZo1a6bAwMBC69q0aZOcnZ01ZcoUw7ZTp05p6NChatKkieGejh07Ztj/999/y8nJSUuWLNFzzz0nX1/ffLUCAAAA5RmB8iH3+eef6/jx45o2bZreeOMNHTp0SMOHD5eUGyZ79eolKysrzZw5U5GRkfruu+8UHh6uzMxMSdKvv/6q3r17y97eXu+++64SEhLk7e2t2bNn66uvvipwvcDAQA0aNEiSFB8frwEDBhSplqJo1qyZLCws9OOPPxq27dq1y3AveX744QfVrVtXTzzxhHbs2KGXX35Z5ubmmjZtmsaPH6///ve/6ty5sy5cuGA45+bNm1qwYIGmTJmioUOH6oknnsh37R07duj1119X586dNXr0aEnSpUuX1KVLF/3222+aMGGCYmNjlZqaqpdffjnf90clacaMGerbt68mTJggHx+fIt8zAAAAUJbMy7oAlK0aNWpozpw5euSRRyRJSUlJSkhIUFpamqZPn66nn35aiYmJMjXN/b8HZ2dnhYWFacOGDerQoYOOHDmi5s2bKyYmRiYmJpIkf39/ffvtt9qzZ4/atm2b73p2dnaGMNawYUPVqlWrSLVYWlre8V4sLS3VpEkT7dy5Uz179lROTo52796tRo0aFQiUgYGBkmS4x3nz5hnu0dvbW88995wWLlyoUaNGScpdkXbAgAEKCAgocN2ffvpJr732msLCwhQdHW3YvnTpUl29elUrV65U9erVJUnNmzdX69atlZCQoLfeestwbJs2bfTCCy/c8R4BAACA8oQRyoech4eHIcBJMgS85ORkHTx4UIGBgcrOzlZmZqYyMzNVr1491axZUzt27JAkhYWFad68ecrIyNBvv/2mjRs3atasWcrKytLNmzdLrJaiCggI0O7du5WZmanffvtNV65cUUREhI4dO6ZLly7p9OnTOnbsmIKCgpSWlqZff/1Vbdu2NYRJSapZs6YaN26s3bt352u7fv36Ba536tQp9enTR2ZmZhozZowhVEvSjz/+qEaNGqlKlSqG/jM3N5e/v7+h/27XNgAAAFDeMUL5kKtUqVK+13nBKjMzU9nZ2UpMTFRiYmKB886dOydJun79uiZNmqTPPvtMmZmZqlWrljw9PWVubl7s50zeqpbitBMYGKi3335bhw4d0oEDB+Tk5KTAwECZmZlp7969unz5sipXrixPT09duHBBOTk5qlKlSoF27O3t9b///S/ftsKO++uvv9S8eXP9+OOPSkxM1NChQw37rly5oqSkJDVq1KjAef8MznnXAwAAAO43BEoUytTUVCYmJurdu7fatGlTYL+VlZUkafLkydq4caNmzpyppk2bGqamNm3atFTrzePo6Kgnn3xSO3bs0KFDh+Tj4yNLS0u5urpq9+7dOnPmjJo3by4zMzNZW1vLxMQk33cl85w/f162trZ3vF6DBg00f/58xcTEaMGCBWrbtq1htNHa2lp+fn4aMWJEid8nAAAAUB4w5RWFMjExkbOzs/7880+5uroafp566im9++67OnjwoCRp3759atq0qVq2bGkIk7/88osuXbpU6CqvkmRmZnZPaw8ICNCOHTv0008/ydfXV5Lk4+OjnTt3aufOnQoKCpKUG4obNWqkDRs25Kv19OnT2r9/v7y8vO54LVtbW5mZmWnQoEGys7NTdHS0oS0fHx+dOHFCTz/9dL4+XLlypdavX38P7hwAAAAoXQRK3NKQIUO0ZcsWRUVFaevWrdq0aZPCw8O1f/9+wzRONzc3bd26VStWrNDu3bv1/vvv69VXX5WJiYnS09MLbdfGxkaS9M033+iPP/4o8boDAwO1d+9eXb16VY0bN5Yk+fr66ujRo0pLS1OLFi0Mx77++us6duyY+vXrpy1btuiLL75Qz549ZW1trZ49exb5mtbW1hozZowOHDigjz76SJLUq1cvZWRkqHfv3vrqq6+0Y8cORUZGauXKlapXr16J3jMAAABQFpjyilsKCAjQggULFB8fr0GDBqlChQpydXXV+++/b5jWGRUVpZs3byouLk4ZGRmqVauW+vfvr2PHjun7778vdJTSz89PQUFBmj59unbt2lXodzTvRuPGjWVpaanatWurcuXKkiQvLy898sgj8vDw0GOPPWY4tnnz5lq4cKFmzZqlIUOGqFKlSmrWrJlGjBihqlWrFuu6ISEheuaZZzR9+nS1bNlS1atX18cff6y4uDhFR0fr5s2bqlOnjuLi4tSuXbuSvGUAAACgTJjkFHflFADlysWLKcrOLp+/xg4ONjp/vuir9OLu0eel7172uYODjQ479rknbd9Lzknz7+nnkM956aPPSx99Xvro88KZmprI3t76lvsZocR9ITMz847HmJqa5nv8BwAAAIB7i0CJ+0Jhj974t7CwME2dOrUUqgEAAAAgEShxn/jkk0/ueExRHvMBAAAAoOQQKHFfcHV1LesSAAAAAPwLXzgDAAAAABiFQAkAAAAAMAqBEgAAAABgFAIlAAAAAMAoBEoAAAAAgFEIlAAAAAAAo/DYEAAAyqnMtBtyTppf1mUUW2bajbIuAQBQSgiUAACUU5dTM6TUjLIuAwCAW2LKKwAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABiFQAkAAAAAMArPoQQAoJTYWlnI3LJCWZdRbJlpN3KfiQkAwL8QKAEAKCXmlhV02LFPWZdRbM5J8yUCJQCgEEx5BQAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABiFQAkAAAAAMAqBEgAAAABgFALlAyIqKkqtW7e+5+eUlF27dsnJyUl79+4t8jnfffedGjVqdNfXTk1NVXBwsD777LO7bitPcfuyLPseAAAAKCnmZV0ASsaAAQOUmpp6z88pK/v379fw4cOVk5NzV+2kpKRowIABOnXqVAlVlut+6ksAAACgpBAoHxC1a9culXNK2/Xr17Vo0SLNnTtXVlZWd9XWtm3bNGnSJF25cqVkivuH+6EvAQAAgJLGlNdyIDMzU0uWLFG7du3k5uamli1bKiEhQVlZWZKk7t27a+zYsYqPj5efn5+aNGmiYcOG6dKlS4Y2/j2FMjg4WPHx8Zo6daqaNWsmd3d3hYeHKykp6Zbn3KmOvFrGjx+vefPmKSAgQK6ururcubMOHTpk1L3/8ccf6tWrl9zc3OTv76+4uLh819uwYYOWLl2q0aNHq1u3boW2UZR7laSIiAi5uLjovffeM6pW6f+m6q5YsUKBgYFq3ry59u7dW6Avf/nlF/Xo0UPe3t7y9PRUz549deDAgVu2e+DAAXl6emrw4MH57h8AAAAozwiU5cDYsWMVGxurNm3aKCEhQc8//7zmzJmj6OhowzEbN27U+vXrNXHiREVFRWn79u2KiIhQdnb2LdtdsmSJTpw4oSlTpmjSpEn65ZdfNHr06LuqQ8oNeVu2bFF0dLTi4uJ04cIFDRky5La13MrkyZPl4+OjefPmKSQkRPPmzdPKlSsN+319fbV582Z17dr1tu0U5V7XrVun6dOny87Orth1/tuMGTM0ZswYDR8+XG5ubvn2paSkKCIiQra2tpo9e7ZmzJih9PR0RUREKCUlpUBbv/32m/r06aNmzZopLi5OZmZmd10fAAAAUBqY8lrGjh49qrVr1yoyMlLh4eGSJH9/f1WsWFGxsbHq2bOnJCk9PV2LFi1SjRo1JEl2dnbq16+ftm7dqsDAwELbrly5subOnWsIKH/99Zdmz56t5ORk2djYFLuO+vXrS5KysrK0YMECWVtbS8pd5GbUqFE6cuSIGjRoUKz779Wrl/r37y9J8vPz0+bNm7Vz50516dJFkvT4448XqZ2i3Gte/SWha9euevbZZwvdd+zYMV2+fFmvvPKKvLy8JEl16tTRihUrlJqaaug3Sfrzzz8VHh4uT09PzZgxQ+bm/EoCAADg/sEIZRnbs2ePJCk0NDTf9g4dOuTb7+3tbQiTkhQYGCgLC4vbrpLq7u6eb7SrevXqkqS0tDSj65AkJyenfKGoWrVqt2z3Tho3bmz4t4mJiR5//HElJycXu53i3GtJuF04rVevniHwjx8/Xt98842qVKmikSNHGvpKkpKTk9WrVy9dunRJb775piwsLO5JrQAAAMC9QqAsY1evXpUk2dvb59ue9zovXFWtWjXffhMTE9nZ2enatWu3bLtixYr5Xpua5r7dha2UWtQ6bteuMVNeK1WqVKAtY9opzr2WhH/30z9ZWVnpgw8+UEBAgL788ksNHDhQTZs21fjx45WRkWE47vLly6pZs6YsLS01ffr0e1InAAAAcC8RKMvYo48+Kkm6ePFivu3nz5+XJNna2kpSgZVJc3JydPHixRL5PmBx6kDR1KlTR++884527typjz/+WGFhYVqxYoWWL19uOMbe3l4LFizQkCFDtG7dOm3btq0MKwYAAACKj0BZxnx8fCRJX3zxRb7tea+9vb0l5T6H8Z+jkd9++61u3rwpPz+/Uq0Dd/bNN9/Iz89P58+fl5mZmTw9PTVhwgQ9+uijOn36tOE4KysrVapUSV27dlWjRo30xhtvKD09vQwrBwAAAIqHFUDKWL169dShQwfDSqCenp766aeflJiYqI4dO6pu3bqSche+6dOnj/r27asLFy4oNjZW/v7+JRYoi1oH7szLy0s5OTl67bXX1KdPH1lZWenLL79USkpKoQv5mJmZacKECfrPf/6jWbNmadSoUWVQNQAAAFB8BMpyYMqUKXJ0dNTq1auVmJiomjVratCgQYqIiDAc4+PjI09PT40cOVLm5uYKDQ3ViBEjSr0O3Jm9vb0WLlyoGTNmaOzYsUpPT1e9evU0e/ZsNWnSpNBz3Nzc9NJLL2np0qUKDQ1Vo0aNSrlqAAAAoPhMcu7VqiUoMd27d5eZmZmWLFlS1qWgHLp4MUXZ2eXz19jBwUbnzxd/1V4Yjz4vfcXpcwcHGx127HOPKyp5zknzy9Xnis956aPPSx99Xvro88KZmprI3t76lvsZoUSJysrKuuPKqiYmJvke8VEe3K91AwAAAGWJQIkS1bp1a506deq2x/j4+GjZsmWlVFHR3K91AwAAAGWJQHkfuJ9CTEJCQr5nLRbGysqqlKopuvu1bgAAAKAsEShRopycnMq6BKPcr3UDAAAAZYnnUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKPwHEoAAEpJZtoNOSfNL+syii0z7UZZlwAAKKcIlAAAlJLLqRlSakZZlwEAQIlhyisAAAAAwCgESgAAAACAUQiUAAAAAACjECgBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKz6EEADxUbK0sZG5ZoUTbdHCwKdH2biUz7UbusywBACgnCJQAgIeKuWUFHXbsU9ZlGMU5ab5EoAQAlCNMeQUAAAAAGIVACQAAAAAwCoESAAAAAGAUAiUAAAAAwCgESgAAAACAUQiUAAAAAACjECgBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQC5X0iODhYY8eOLdKxqampCg4O1meffXbX1502bZp69uxp1LnFqbk0rV69Wk5OTjpz5sxdt1Ve7xEAAAAoDQTKB0xKSor69++vU6dO3XVby5cv16JFi0qgKgAAAAAPIvOyLgAlZ9u2bZo0aZKuXLlyV+2cPXtWMTEx2rBhg2xsbEqmOAAAAAAPHEYo7yM3b97U1KlT1axZM3l4eCg8PFwnT5407I+IiJCLi4vee++9Qs+fPXu2QkJCtHnzZrVv314uLi567rnnCkyNnTFjhg4fPqzFixerYcOGJVb/8uXL5eTkpKVLlxarnjNnzigyMlItWrSQu7u7unbtqt27d0uSrly5ImdnZy1btsxw/LFjx+Tk5KQJEyYYtqWlpcnFxUWffvppobXt2bNHXbt2lbu7u3x9fTVu3Dhdu3Yt3zG//fabevXqJU9PTwUFBWndunUF2jl9+rQGDRqkxo0by8/PT9OnT9fo0aPVvXt3wzHZ2dlKTExUq1at5OLiopCQEK1atap4nQkAAACUAwTK+8jnn3+u48ePa9q0aXrjjTd06NAhDR8+3LB/3bp1mj59uuzs7G7ZxtmzZzV58mT16NFD8+fPV61atTRq1Cj9+eefhmMiIiK0fv16+fn5lVjta9eu1VtvvaXIyEj16NGjyPWcO3dOnTp10sGDBxUZGakZM2aoYsWK6tWrl3788UdVrlxZ7u7u+vHHHw1t7ty5U1JuSPzntszMTAUEBBSobc+ePerVq5esrKw0c+ZMRUZG6rvvvlN4eLgyMzMNdXbr1k3Jycl65513NGTIEMXGxurs2bOGdjIyMtSzZ0/9+uuvmjBhgiZMmKDNmzfriy++yHe9CRMmKD4+XmFhYUpMTFRQUJCio6PzhWIAAADgfsCU1/tIjRo1NGfOHD3yyCOSpKSkJCUkJCgtLU2WlpaqX7/+HdtIS0tTQkKCISw++eSTCgoK0vfff68nn3xSklS3bt0Srfubb77RmDFjNHToUIWHhxernsWLF+vatWtatWqVatSoIUkKDAxUx44dFRsbq08//VQBAQFasGCBsrKyZGZmpp07d6pRo0b69ddfdenSJdnZ2emHH36Qq6urqlSpUqC+6dOn6+mnn1ZiYqJMTXP/j8XZ2VlhYWHasGGDOnTooCVLligrK0vvvfeebG1tJUlPPfWUXnrpJUM769atU1JSkj777DM5OTlJkjw8PNSqVSvDMSdOnNDKlSsVGRmp3r17S5KaN2+urKwszZw5U506dVKlSpVKqusBAACAe4oRyvuIh4eHIUxKUq1atSRJycnJxWrHy8vL8O/q1atLktLT00ugwoJ+/vlnvf7666pdu7b69u1b7Hr27t0rb29vQ5iUJFNTU7Vt21a//vqrUlJSFBgYqOTkZP3666/Kzs42jDiam5sbRil/+OEHBQYGFrh2enq6Dh48qMDAQGVnZyszM1OZmZmqV6+eatasqR07dkiS9u3bJy8vL0OYlCR3d3fVrFnT8Hrnzp168sknDWEy7348PT3zHZOTk6OgoCDDtTIzMxUcHKzk5GT9/PPPRe5bAAAAoKwxQnkf+ffIVd5oWk5OTpHbMDMzk4WFRYE2srOzS6DCgo4cOaKgoCBt2bJFn3zyiV588cVi1XP16lXDyOk/ValSRTk5OUpNTVWDBg1UvXp1/fjjjzIzM9PVq1fVokULubi4aM+ePapfv75OnjypoKCgAu1cu3bN8J3GxMTEAvvPnTtnqMPR0bHAfgcHB8O/L1++XOh0YwcHB50/f16SDAsmhYSEFDjun9cDAAAA7gcEStxTgYGBSkxM1MCBA/XOO+8oKCio0Gmnt/Loo4/qwoULBbbnBa+8EcNnnnlGO3bskIWFhZycnFS5cmX5+vrqu+++U+3atVWtWjU5OzsXaMfKykomJibq3bu32rRpU+j+vOtcvHixwP5/rqhbtWpV7d+/v8Ax/zwvb9Xc5cuXq2LFigWOzRt1BgAAAO4HTHnFPZUXHseOHaubN2/qrbfeKtb5TZo00b59+3TmzBnDtuzsbH311VdydXU1jG4GBARo//792r59u3x8fCRJPj4+OnLkiNavX1/odFdJsra2lrOzs/7880+5uroafp566im9++67OnjwoCTJz89P+/btM4w0Srmryf5zld0mTZooKSlJx44dM2y7ePGiDhw4YHjduHFjSbkjnv+83unTpzVr1qx7NvUYAAAAuBcIlCgVNWrU0MCBA/Xll1/qu+++K/J5vXr1krW1tXr06KEvvvhCW7ZsUd++ffXHH3/o9ddfNxzXrFkzSbnflfT19ZUkeXt7y9zcXAcOHLhloJSkIUOGaMuWLYqKitLWrVu1adMmhYeHa//+/WrUqJEkqUePHrKxsVHv3r21ceNGbdiwQf3798/3ndb27dvL0dFR/fv31/r167Vp0yZFREQoIyNDJiYmkqQGDRooNDRUY8aM0eLFi7Vz504tX75cUVFRun79er7vZAIAAADlHYESpaZHjx6qX7++3nzzTaWmphbpnKpVq+qjjz5S/fr19cYbb2jYsGG6fv26Fi9eLH9/f8NxlpaW8vHxkYmJiWEUsFKlSnJ1dVWFChXUtGnTW14jb5XYpKQkDRo0SGPGjJG1tbXef/99w8q5tra2+uijj/T4449r1KhRevvtt9W1a1c1aNDA0M4jjzyiRYsW6emnn9bYsWMVHR2t4OBgubu7y9LS0nDc1KlT1b17dy1fvlwRERFasGCBOnXqpISEhGL1JwAAAFDWTHKKs6ILgFs6evSo/vzzT7Vu3dqwLTMzU4GBgWrXrp1Gjx59T6578WKKsrPL56+xg4ONzp8v3irEuDv0+Z05ONjosGOfsi7DKM5J83l/xee8LNDnpY8+L330eeFMTU1kb299y/0syoNiyc7OLtKKsObmD99HKyUlRYMGDVLPnj0VGBioGzduaOXKlbp27Vq+51UCAAAAD4qH769+3JU5c+YoPj7+jsdt3rz5oVux1NPTU9OnT9fChQu1YsUKmZqaysPDQ8uXL9fTTz9d1uUBAAAAJY5AiWJ56aWXbrvATZ6qVave+2LKoXbt2qldu3ZlXQYAAABQKgiUKJZq1aqpWrVqZV0GAAAAgHKAVV4BAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKDw2BADwUMlMuyHnpPllXYZRMtNulHUJAADkQ6AEADxULqdmSKkZJdaeg4ONzp9PLrH2AAC4nzDlFQAAAABgFAIlAAAAAMAoBEoAAAAAgFEIlAAAAAAAoxAoAQAAAABGIVACAAAAAIxCoAQAAAAAGIXnUAIAHmi2VhYyt6xwT6/h4GBzT9rNTLuR+9xMAADKKQIlAOCBZm5ZQYcd+5R1GUZxTpovESgBAOUYU14BAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABil1AJlVFSUWrdufc/PKSm7du2Sk5OT9u7dW+RzvvvuOzVq1Oiur52amqrg4GB99tlnd91WYZycnDR37tx7dvytLFu2TM2bN5ebm5vmz59/1+0BAAAAKFvmpXWhAQMGKDU19Z6fU1b279+v4cOHKycn567aSUlJ0YABA3Tq1KkSqqygFStWqEaNGves/cKkpaVpypQpCggIUO/evfXEE0+U6vUBAAAAlLxSC5S1a9culXNK2/Xr17Vo0SLNnTtXVlZWd9XWtm3bNGnSJF25cqVkirsFDw+Pe9p+YZKTk5WVlaVWrVqpSZMmpX59AAAAACWvSFNeMzMztWTJErVr105ubm5q2bKlEhISlJWVJUnq3r27xo4dq/j4ePn5+alJkyYaNmyYLl26ZGjj39NXg4ODFR8fr6lTp6pZs2Zyd3dXeHi4kpKSbnnOnerIq2X8+PGaN2+eAgIC5Orqqs6dO+vQoUNGddAff/yhXr16yc3NTf7+/oqLi8t3vQ0bNmjp0qUaPXq0unXrVmgbRblXSYqIiJCLi4vee+89o2rNs3TpUoWEhMjV1VUtWrTQhAkTlJKSYtj/7yms586dU2RkpPz8/OTl5aUePXro119/vWX706ZNU6NGjbRx48Yi1bN69Wo988wzkqQxY8bIyclJUu57NWrUKA0cOFBeXl4aPHiwJOnKlSuaNGmSgoOD5erqqhdeeKHAtZycnLRixQqNGDFCnp6e8vPzU3x8vFJSUjR69Gh5e3vL399f77zzTrFGjV955RX95z//KbC9R48ehvc3KytL8+bNU2hoqNzc3OTh4aEuXbpo165d+c45cOCAevXqJS8vLzVt2lSRkZG6ePGiYX9x+x0AAAAob4oUKMeOHavY2Fi1adNGCQkJev755zVnzhxFR0cbjtm4caPWr1+viRMnKioqStu3b1dERISys7Nv2e6SJUt04sQJTZkyRZMmTdIvv/yi0aNH31UdUm7I27Jli6KjoxUXF6cLFy5oyJAht63lViZPniwfHx/NmzdPISEhmjdvnlauXGnY7+vrq82bN6tr1663baco97pu3TpNnz5ddnZ2xa4zzxdffKF33nlHXbt21cKFC/Xaa6/ps88+0+TJkws9PjU1VV26dNHevXsVFRWlmTNnKjs7W7169dKZM2cKHD9nzhwtXbpUMTExevbZZ4tUU2BgoBISEiRJ/fv314oVK/LVW6lSJc2ZM0ddunRRenq6Xn75ZX399dfq37+/4uPjVadOHQ0aNEhr167N1+60adNka2uruXPnKigoSLNnz1anTp1UqVIlzZw5U0FBQVqwYEGRg68kdezYUQcOHNDff/9t2HbhwgXt2bNHHTp0kCTFxMQoMTFRXbp00YIFCzRp0iRdvnxZQ4YMUXp6uiTp8OHD6tatm7KyshQTE6Po6Gjt3btXffv2NarfAQAAgPLojlNejx49qrVr1yoyMlLh4eGSJH9/f1WsWFGxsbHq2bOnJCk9PV2LFi0yfDfPzs5O/fr109atWxUYGFho25UrV9bcuXNlZmYmSfrrr780e/ZsJScny8bGpth11K9fX1LuCNKCBQtkbW0tKfeP91GjRunIkSNq0KBBsTqoV69e6t+/vyTJz89Pmzdv1s6dO9WlSxdJ0uOPP16kdopyr3n1343du3erVq1a6tatm0xMTOTj4yNLS0tdvXq10OPXrFmjU6dOad26dYbre3h4KCwsTPv371fbtm0Nx77//vuKj4/XlClT1K5duyLXZGdnJ2dnZ0m505j/OeXW3NxckyZNUsWKFSVJH3zwgf744w+tWrVKbm5ukqSAgABdvXpV77zzjtq3b2/ow0aNGmns2LGSpAYNGmj16tWyt7fX+PHjJeV+Pr788ksdOHBAzz33XJFqfe655zRx4kRt2LBBffr0kSR9+eWXMjMzU0hIiKTckcVhw4bl+0+EChUqaNCgQTp69Kjc3NyUmJgoe3t7LViwQBYWFpJyPwPjx49XUlKSfvjhhyL3OwAAAFBe3XGEcs+ePZKk0NDQfNvzRmvy9nt7e+db6CUwMFAWFha3XSXV3d3dEA4kqXr16pJyF3Axtg4pdzpkXpiUpGrVqt2y3Ttp3Lix4d8mJiZ6/PHHlZycXOx2inOvd8PPz08nTpxQWFiY4uPjdejQIbVv317du3cv9Ph9+/bJ0dExX5i1sbHRpk2b8oWaTZs26e2331azZs30/PPPl1i9tWvXNoRJKfd9dHR0NITJPO3bt9eFCxd0/Phxw7Z/HmNrayszMzO5u7sbtpmYmOixxx7TtWvXilyPtbW1WrZsqQ0bNhi2rV+/XkFBQXr00UclSTNmzFD37t116dIl7d27V59++qnWrVsnSbp586ak3H4NCAgwhElJatasmTZt2iRHR8ci9zsAAABQnt0xUOaNbNnb2+fbnvc6L1xVrVo1334TExPZ2dnd9o/5fwYJSTI1zS2nsO+8FbWO27VrzJTXSpUqFWjLmHaKc693o23btpo+fbosLS01d+5cderUqUBA+qcrV64U6NPC/Prrr3rmmWe0bds27dixo8Tq/fe1r169qipVqhQ4Lm/bP9/nwhZB+vf7ZYyOHTvqv//9r44fP67//e9/OnDggOE/LiTp0KFD6tSpk5o2baqIiAh99NFHBd7PK1eu3HbqclH7HQAAACjP7hgo80Zl/rmYiCSdP39eUu7IkKQCK5Pm5OTo4sWLd/V9QGPqQO4o7ocffqhdu3bp3XffVeXKlTVy5EhduHChwLE2Njb5Fk/Ks3fvXv3555+G1127dlViYqJcXFz0xhtv6Pr16/ek9kcffbTQOs+dOyepdN5nf39/2dvb66uvvtKGDRv02GOPGRYVSklJUUREhKytrbV+/Xrt379fn3zyif7f//t/+dqwtrYu0K/Z2dn67rvvdPHixSL3OwAAAFCe3TFQ+vj4SMpdPOWf8l57e3tLyn0O4z9HI7/99lvdvHlTfn5+JVJoUet42A0fPlwDBw6UlBsW27RpowEDBigzM9MQvv/Jy8tLSUlJ+uOPPwzbUlNT1a9fP61fv96wrUqVKjI1NdWECRP0999/a/bs2fekfh8fHyUlJennn3/Ot339+vVycHCQo6PjPbnuP5mbm6tdu3basmWLvvrqK7Vp08YwdfX48eO6cuWKevbsqbp16xpGJrdu3Srp/0Yovb29tW3bNsMUWCn3d6Rv3746ceJEkfsdAAAAKM/uuChPvXr11KFDB82YMUPp6eny9PTUTz/9pMTERHXs2FF169aVlPvHcJ8+fdS3b19duHBBsbGx8vf3L7FAWdQ6HnZ+fn4aN26cpk2bpmeeeUbXrl1TfHy8nnrqKdWrV6/A8Z06ddKyZcvUv39/DRo0SI899pgWLlyoChUq6MUXXyxwvKurq7p06aIlS5YoNDRUDRs2LNH6w8LCtGzZMg0YMEBDhgxRtWrV9MUXX2jr1q166623DAHuXuvYsaOWLVsmKfdRJ3meeuopWVtba+7cuTIxMZGpqak2btyoTz75RNL/fSd2wIAB6ty5s/r166du3bopLS1NcXFx8vHxkZeXlxo0aFCsfgcAAADKozsGSkmaMmWKHB0dtXr1aiUmJqpmzZoaNGiQIiIiDMf4+PjI09NTI0eOlLm5uUJDQzVixIgSLbYodTzsXnzxRWVkZOjDDz/Uhx9+qIoVKxqegWhuXvDttra21gcffKBp06Zp4sSJysnJkZeXl95///0C34vN8/rrr2vjxo0aN26cVq5cmW+xobtlaWmp5cuXa/r06YqNjVV6errq16+v2bNnF/kxJSXBxcVFderUUUZGhry8vAzbbWxsNHfuXMXExGjw4MGysrJSw4YNtXz5cr366quGxXhcXFy0dOlSzZgxQ0OGDNGjjz6q4OBgDR8+XKampkb1OwAAAFDemOSUwKow3bt3l5mZmZYsWVICJQEojosXU5SdXbKLO5UUBwcbnT9f/FWRYTz6vCAHBxsdduxT1mUYxTlpPu9nIficlz76vPTR56WPPi+cqamJ7O2tb7m/SCOUD5KsrKw7rqxqYmJSoqNuJaG81p2ZmXnHY0xNTUttquqtZGdnF2l13sJGcQEAAAAU7qH767l169Y6derUbY/x8fExfH+uvCivdTdq1OiOx4SFhWnq1KmlUM2tjRkzRmvWrLnjcb///nspVAMAAAA8GEokUJa38HU7CQkJysjIuO0xhT3fsKyV17rzFqO5nfLwSJeBAweqa9euZV0GAAAA8EB56EYonZycyroEo5TXul1dXcu6hCKpVauWatWqVdZlAAAAAA+Usv1iGwAAAADgvkWgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRHrrHhgAAHi6ZaTfknDS/rMswSmbajbIuAQCA2yJQAgAeaJdTM6TUjHvWvoODjc6fT75n7QMAUJ4x5RUAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABiF51ACAB5otlYWMrescE+v4eBgU+JtZqbdyH2GJgAA5RiBEgDwQDO3rKDDjn3Kuoxic06aLxEoAQDlHFNeAQAAAABGIVACAAAAAIxCoAQAAAAAGIVACQAAAAAwCoESAAAAAGAUAiUAAAAAwCgESgAAAACAUQiUAAAAAACjECgBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUD5gImKilLr1q3v+TnIb/Xq1XJyctKZM2fuyfEAAABAeWRe1gWgZA0YMECpqan3/BzkFxgYqBUrVsjOzq6sSwEAAABKDYHyAVO7du1SOQf52dnZESYBAADw0GHKazmSmZmpJUuWqF27dnJzc1PLli2VkJCgrKwsSVL37t01duxYxcfHy8/PT02aNNGwYcN06dIlQxv/nr4aHBys+Ph4TZ06Vc2aNZO7u7vCw8OVlJR0y3PuVEdeLePHj9e8efMUEBAgV1dXde7cWYcOHSr2fS9dulQhISFydXVVixYtNGHCBKWkpBj2Z2dnKzExUa1atZKLi4tCQkK0atWqfG1kZWVp3rx5Cg0NlZubmzw8PNSlSxft2rXLcMz169c1YcIEPfPMM4Z2Fi5cWOx6nZycNGfOHL3wwgtq3LixFi9eXGAK66VLlzR8+HD5+/vLzc1NHTt21Nq1a2/Z5qVLl9SmTRu1adNGFy5cKHZNAAAAQFlghLIcGTt2rNavX69+/frJ09NT+/fv15w5c3Ty5Em9/fbbkqSNGzeqSpUqmjhxopKTkxUTE6OIiAh98sknMjUt/P8HlixZIm9vb02ZMkVXr17V5MmTNXr0aH344YdG1yFJGzZsUN26dRUdHa2cnBxNmzZNQ4YM0aZNm25Zy7998cUXeueddzRq1Cg5OTnp+PHjmjZtmm7cuKEpU6ZIkiZMmKDVq1erf//+cnd31/bt2xUdHa3r16+re/fukqSYmBitXLlSI0aMUL169XT27FnNmTNHQ4YM0ZYtW1SpUiW9/fbb2rZtm6KiomRvb6+tW7cqJiZGdnZ2CgsLK/L7JElz587VsGHD9NRTT8nR0VEHDx7Mt3/kyJG6ePGi3nzzTVlbW+uzzz7TqFGjVKNGDfn6+uY7Njk5WREREcrOztayZctUpUqVYtUCAAAAlBUCZTlx9OhRrV27VpGRkQoPD5ck+fv7q2LFioqNjVXPnj0lSenp6Vq0aJFq1KghKXeqZb9+/bR161YFBgYW2nblypU1d+5cmZmZSZL++usvzZ49W8nJybKxsSl2HfXr15eUOyq4YMECWVtbS5JSU1M1atQoHTlyRA0aNCjSfe/evVu1atVSt27dZGJiIh8fH1laWurq1auSpBMnTmjlypWKjIxU7969JUnNmzdXVlaWZs6cqU6dOqlSpUo6d+6chg0bpq5duxrarlChggYNGqSjR4/Kzc1Nu3fvlr+/v9q2bStJ8vX1laWlpWxtbYtU6z95eXkZ+kdSgUC5e/duvfbaa2rVqpUkycfHR5UrV9YjjzyS77j09HQNGzZM165d0/Lly1W1atVi1wIAAACUFaa8lhN79uyRJIWGhubb3qFDh3z7vb29DWFSyl0MxsLCQnv37r1l2+7u7oYwKUnVq1eXJKWlpRldh5Q79TMvTEpStWrVbtnurfj5+enEiRMKCwtTfHy8Dh06pPbt2xtGHnfu3KmcnBwFBQUpMzPT8BMcHKzk5GT9/PPPkqQZM2aoe/fuunTpkvbu3atPP/1U69atkyTdvHlTUm6AXLlypV599VUtX75cJ0+e1GuvvXbLIH47eaH6Vnx9fTV79mwNHjxYq1at0oULFzRq1Ch5eXnlO2748OHat2+fhg4danhfAAAAgPsFI5TlRN6InL29fb7tea+Tk5MlqcAIlomJiezs7HTt2rVbtl2xYsV8r/Omo+bk5Bhdx+3azc7OvmUt/9a2bVtlZ2frww8/1Ny5czV79mw9/vjjGjFihNq2basrV65IkkJCQgo9/9y5c5KkQ4cO6c0339ShQ4dUqVIl1a1bVzVr1sx3n2PHjlX16tW1bt06TZo0SZMmTZKnp6cmTJhQ5BHVPP/un3+bMWOGEhMT9eWXX+rrr7+WqampmjVrpokTJ+rxxx83HHfhwgU1bNhQs2bNUuvWrVWhQoVi1QEAAACUJQJlOfHoo49Kki5evGgY6ZOk8+fPS5JhWmZewMqTk5OjixcvltgKo0WtoySFhoYqNDRUycnJ2rZtm9577z2NHDlSPj4+him5y5cvLxBgJalWrVpKSUlRRESEGjZsqPXr16tOnToyNTXV999/r6+//tpwrIWFhfr376/+/fvrf//7n7Zs2aK5c+dq5MiR+vzzz0v0nmxsbDRy5EiNHDlSx48f1+bNmzV37lxNmjRJiYmJhuPmzJkjMzMzderUSXPmzNGwYcNKtA4AAADgXmLKaznh4+MjKXeRmn/Ke+3t7S1J2r9/f77RyG+//VY3b96Un59fqdZRUoYPH66BAwdKyg1hbdq00YABA5SZmanz58+rcePGknJHTl1dXQ0/p0+f1qxZs5Senq7jx4/rypUr6tmzp+rWrWsYKd26dauk3NCdkZGhkJAQLVq0SJJUs2ZNde3aVe3atdPp06dL9J7OnDmjgIAAffXVV5KkOnXq6NVXX1WzZs0KXMvBwUHOzs56+eWXtXDhQv3+++8lWgsAAABwLzFCWU7Uq1dPHTp00IwZM5Seni5PT0/99NNPSkxMVMeOHVW3bl1JuQvf9OnTR3379tWFCxcUGxsrf3//EguURa2jpPj5+WncuHGaNm2annnmGV27dk3x8fF66qmnVK9ePZmbmys0NFRjxozRyZMn1bBhQx07dkxxcXFq1KiRatasKRsbG1lbW2vu3LkyMTGRqampNm7cqE8++URS7nc6LSws5Orqqvj4eD3yyCNycnLSiRMntGbNGj333HMlek/Vq1fX448/rrfeekspKSmqXbu2fvnlF33//fcaMGBAoecMHTpUX3/9tcaNG6cVK1YUeZVcAAAAoCwRKMuRKVOmyNHRUatXr1ZiYqJq1qypQYMGKSIiwnCMj4+PPD09NXLkSEPYGjFiRKnXUVJefPFFZWRk6MMPP9SHH36oihUrqmnTpoqMjJS5ee7Hc+rUqUpMTNTy5ct19uxZValSRZ06ddLgwYMl5Y5szp07VzExMRo8eLCsrKzUsGFDLV++XK+++qr27dungIAAvfnmm7K1tdWiRYt0/vx52dvbq1OnTho6dGiJ39fs2bMVGxurmTNn6vLly6pRo4YGDRqkV199tdDjra2tNXr0aL3++utavny5XnnllRKvCQAAAChpJjmFrcyCcql79+4yMzPTkiVLyroUlCMXL6YoO7t8/ho7ONjo/PnkOx+IEkOfF+TgYKPDjn3Kuoxic06az3t5C3zOSx99Xvro89JHnxfO1NRE9vbWt9zPCCXuiaysrEJXkf0nExOTfI8zKUvZ2dlFWp02b9QUAAAAAIES90jr1q116tSp2x7j4+OjZcuWlVJFtzdmzBitWbPmjsexaA4AAADwfwiU95HyEr6KIiEhQRkZGbc9xsrKqpSqubOBAweqa9euZV0GAAAAcF8hUOKecHJyKusSiqVWrVqqVatWWZcBAAAA3Fd4NgEAAAAAwCgESgAAAACAUQiUAAAAAACjECgBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKz6EEADzQMtNuyDlpflmXUWyZaTfKugQAAO6IQAkAeKBdTs2QUjPuWfsODjY6fz75nrUPAEB5xpRXAAAAAIBRCJQAAAAAAKMQKAEAAAAARiFQAgAAAACMQqAEAAAAABiFQAkAAAAAMAqBEgAAAABgFJ5DCQB4oNhaWcjcskKpXtPBweau28hMu5H7zEwAAO4jBEoAwAPF3LKCDjv2Kesyis05ab5EoAQA3GeY8goAAAAAMAqBEgAAAABgFAIlAAAAAMAoBEoAAAAAgFEIlAAAAAAAoxAoAQAAAABGIVACAAAAAIxCoAQAAAAAGIVACQAAAAAwCoESAAAAAGCUUg2UUVFRat269T0/B7fWvXt39ezZ854dfyu7d+9WSEiIXFxc1Ldv37tuDwAAAEDZMy/Niw0YMECpqan3/Bzc2htvvCETE5NSv25sbKyysrI0f/58ValSpdSvDwAAAKDklWqgrF27dqmcg1urW7dumVz3ypUr8vLyUrNmzcrk+gAAAABKXpGnvGZmZmrJkiVq166d3Nzc1LJlSyUkJCgrK0tS7tTIsWPHKj4+Xn5+fmrSpImGDRumS5cuGdr49/TV4OBgxcfHa+rUqWrWrJnc3d0VHh6upKSkW55zpzryahk/frzmzZungIAAubq6qnPnzjp06FCxO2jp0qUKCQmRq6urWrRooQkTJiglJcWwPzs7W4mJiWrVqpVcXFwUEhKiVatW5WsjKytL8+bNU2hoqNzc3OTh4aEuXbpo165dhmOuX7+uCRMm6JlnnjG0s3DhwmLXu337dr300kvy9PRUkyZNNGDAAP3xxx/5+uafU1gzMjL07rvvKjg4WO7u7mrfvr02bNhwy/Y3bdokZ2dnTZkypUj1/P3333JyclJSUpLWrl0rJycn7dq1S7Nnz1ZISIhmzZolX19fPffcc0pNTS3y+zthwgTNnj1b/v7+8vT01Ouvv66UlBTNnz9fLVq0kLe3twYNGqTLly8Xue/i4+Pl6uqq5OTkfNvnzJkjDw8Pw0j5119/rS5dusjT01MuLi5q06aNPvzww3znnDt3TpGRkfLz85OXl5d69OihX3/91bC/uP0OAAAAlEdFHqEcO3as1q9fr379+snT01P79+/XnDlzdPLkSb399tuSpI0bN6pKlSqaOHGikpOTFRMTo4iICH3yyScyNS08uy5ZskTe3t6aMmWKrl69qsmTJ2v06NEF/kAvTh2StGHDBtWtW1fR0dHKycnRtGnTNGTIEG3atOmWtfzbF198oXfeeUejRo2Sk5OTjh8/rmnTpunGjRuGQDVhwgStXr1a/fv3l7u7u7Zv367o6Ghdv35d3bt3lyTFxMRo5cqVGjFihOrVq6ezZ89qzpw5GjJkiLZs2aJKlSrp7bff1rZt2xQVFSV7e3tt3bpVMTExsrOzU1hYWJHqPXnypAYMGKD/9//+n4YNG6arV69qxowZ6tu3r7755ptCp7qOGDFCW7du1YABA+Tq6qpvvvlGw4YNU6VKlRQUFJTv2B07duj1119X586dNXr06CLVVLVqVa1YsUJDhgyRk5OTBgwYoLp162r37t06efKktmzZori4OF27dk1WVlYaNWpUkd7fdevWycPDQ9OmTdPRo0c1depU/fe//1XVqlX11ltv6cSJE4qJiVHVqlUVHR1dpFo7duyo2bNna9OmTfn6/Msvv1TLli1lZWWlzZs3a/DgwerZs6cGDx6s69ev68MPP9Sbb74pFxcXubm5KTU1VV26dFFOTo7h/Zw/f7569eqldevWqXr16sXqdwAAAKC8KlKgPHr0qNauXavIyEiFh4dLkvz9/VWxYkXFxsYaRrzS09O1aNEi1ahRQ5JkZ2enfv36aevWrQoMDCy07cqVK2vu3LkyMzOTJP3111+aPXu2kpOTZWNjU+w66tevLyl3VHDBggWytraWJKWmpmrUqFE6cuSIGjRoUKTO2b17t2rVqqVu3brJxMREPj4+srS01NWrVyVJJ06c0MqVKxUZGanevXtLkpo3b66srCzNnDlTnTp1UqVKlXTu3DkNGzZMXbt2NbRdoUIFDRo0SEePHpWbm5t2794tf39/tW3bVpLk6+srS0tL2draFqlWSfr55591/fp19e3bV9WqVZMk1ahRQ5s3b1ZqaqqhL/IcOXJEX3/9tcaPH2+orWnTpvrrr7+0a9eufMHmp59+0muvvaawsLAiBzRJsrCwkIeHhywsLGRnZycPDw/DvszMTEVFRcnX11dS8d7fnJwczZo1S9bW1mrevLlWr16tU6dOadWqVbKxsVFAQIB27typn376qci1PvHEE/Ly8tL69esNgfL333/X0aNHFRkZKUn6448/9MILL+QL1J6envL19dXu3bvl5uamNWvW6NSpU1q3bp2hXg8PD4WFhWn//v2qW7dukfsdAAAAKM+KFCj37NkjSQoNDc23vUOHDoqNjTXs9/b2NoRJSQoMDJSFhYX27t17y0Dp7u5uCJOSVL16dUlSWlpagUBZlDry/oB3cnLKF6DyAlZaWlpRblmS5OfnpxUrVigsLEytWrVSQECA2rdvbxjp27lzp3JychQUFKTMzEzDecHBwVq6dKl+/vln+fr6asaMGZKkS5cu6fjx40pKStKWLVskSTdv3pSUGyA//vhjnTlzRgEBAQoICNBrr71W5Fql3L6sUKGCOnXqpJCQED3zzDPy9fWVm5tbocfv27dPkgqsortgwYJ8r0+dOqU+ffrIzMxMY8aMKdFFffLeL6l472/dunXzvb/29vaqUKFCvs9M5cqV8033LYqOHTtq0qRJunz5smxtbbVhwwZVqVJF/v7+kqQ+ffpIyv0PihMnTuivv/4yTKXOey/37dsnR0fHfPdmY2OjTZs2SZI++ugjSXfudwAAAKC8K9Lcz7wROXt7+3zb817nfeesatWq+fabmJjIzs5O165du2XbFStWzF/Q/z8dNScnx+g6btdudnb2LWv5t7Zt22r69OmytLTU3Llz1alTJ7Vs2dLwXbcrV65IkkJCQtSoUSPDT48ePSTlfo9Okg4dOqROnTqpadOmioiI0EcffVTgPseOHauhQ4fq77//1qRJk9SqVSt17txZv/32W5HrrVWrlpYvXy53d3d98sknioiIkL+/v2bMmFFof+bV/+/+/Le//vpLbm5uSktLU2JiYpHruRMzM7N8I7DFeX+trKwKtFepUqW7rqlNmzYyNTXV119/LSl36nS7du0M/+lx6dIlDRo0SI0bN9ZLL71kGE2X/u+9vHLlym37tKj9DgAAAJR3RRqhfPTRRyVJFy9eNIz0SdL58+clyRAK8v5QzpOTk6OLFy/Kzs6uJGotch0lKTQ0VKGhoUpOTta2bdv03nvvaeTIkfLx8TGMhi1fvrxAgJVyA15KSooiIiLUsGFDrV+/XnXq1JGpqam+//57Q2iRcqeG9u/fX/3799f//vc/bdmyRXPnztXIkSP1+eefF7leNzc3xcfHKyMjQ/v27dOKFSuUmJgoZ2dnPffcc/mOzav/0qVLcnBwMGw/cuSI0tPT5e7uLklq0KCB5s+fr5iYGC1YsEBt27bNN/pWUsri/f23xx57TIGBgfrqq6/k7Oysv/76Sx06dDDsHzFihE6cOKElS5bI09NTFhYWSk9Pz7cQk42NjU6fPl2g7b1796pKlSpF7ncAAACgvCvSCKWPj4+k3EVq/invtbe3tyRp//79+UYjv/32W928eVN+fn4lUmxR6ygpw4cP18CBAyXlhoQ2bdpowIAByszM1Pnz59W4cWNJuSNrrq6uhp/Tp09r1qxZSk9P1/Hjx3XlyhX17NlTdevWNYxMbt26VVJu6M7IyFBISIgWLVokSapZs6a6du2qdu3aFRpMbmXZsmUKDg5WRkaGLCws1LRpU02aNEmSCm0nr7/ypt/mmTx5suLi4gyvbW1tZWZmpkGDBsnOzk7R0dHFGuktqtJ+f2+lY8eO2rt3rz755BM9/fTTcnFxMezbt2+fQkJC5OvrKwsLC0n/917m9YmXl5eSkpLyTbdNTU1Vv379tH79+iL3OwAAAFDeFWmEsl69eurQoYNmzJih9PR0eXp66qefflJiYqI6duxoeLZhamqq+vTpo759++rChQuKjY2Vv79/iQXKotZRUvz8/DRu3DhNmzZNzzzzjK5du6b4+Hg99dRTqlevnszNzRUaGqoxY8bo5MmTatiwoY4dO6a4uDg1atRINWvWlI2NjaytrTV37lyZmJjI1NRUGzdu1CeffCIp9zudFhYWcnV1VXx8vB555BE5OTnpxIkTWrNmTYFRxTvVGxMTo9dee03dunWTmZmZPv74Y1WoUKHQhV4aNmyoZ599VlOmTFFaWpqcnJy0adMm7d69u9BHllhbW2vMmDEaMmSIPvroo3yLDJWE0n5/b+WZZ56RlZWVPvnkEw0ePDjfPjc3N61bt04NGzZUtWrVtH//fs2fP18mJiZKT0+XJHXq1EnLli1T//79NWjQID322GNauHChKlSooBdffFFVq1YtVr8DAAAA5VWRHxsyZcoUOTo6avXq1UpMTFTNmjU1aNAgRUREGI7x8fGRp6enRo4caQhbI0aMKNGCi1JHSXnxxReVkZGhDz/8UB9++KEqVqyopk2bKjIyUubmuV03depUJSYmavny5Tp79qyqVKmiTp06GYKIjY2N5s6dq5iYGA0ePFhWVlZq2LChli9frldffVX79u1TQECA3nzzTdna2mrRokU6f/687O3t1alTJw0dOrTI9darV0/z5s3T7NmzNWzYMGVlZcnFxUWLFi2So6NjoedMnz5dM2fO1KJFi3T16lU9/fTTSkhIULNmzQo9Pm+xn+nTp6tly5aGRZRKSmm+v7diYWGhNm3a6OOPP1b79u3z7Zs6daomTZqkiRMnSpKefPJJvfnmm1q3bp1hkSNra2t98MEHmjZtmiZOnKicnBx5eXnp/fffN3zPuLj9DgAAAJRHJjmFrdZihO7du8vMzExLliwpieYAFNHFiynKzi6RX+MS5+Bgo/Pnk+98IEoMfZ7bB4cd+5R1GcXmnDT/oX/viorPeemjz0sffV766PPCmZqayN7e+pb7izxC+SDJysoqdNXTfzIxMcn3OJOylJ2dXaTvLOaNmpaW+6Ufc3JylJWVdcfjzMzMSvSRKAAAAMCD7qEMlK1bt9apU6due4yPj4+WLVtWShXd3pgxY7RmzZo7Hvf777+XQjX/p2fPntq9e/dtj3n88cf17bffllJFhVuzZo1Gjx59x+Pef/99+fr6lkJFAAAAwIOhxAJleQlfRZGQkKCMjIzbHlPYcw7LysCBA0t8AZyS8Oabbyo1NfW2x+SthFqWgoKCDIsg3c5TTz1VCtUAAAAAD46HcoTSycmprEsollq1aqlWrVplXUYBderUKesSisTW1rZUnmEJAAAAPGyK9BxKAAAAAAD+jUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKARKAAAAAIBRCJQAAAAAAKMQKAEAAAAARnkon0MJAHhwZabdkHPS/LIuo9gy026UdQkAABQbgRIA8EC5nJohpWaU2vUcHGx0/nxyqV0PAIDyhCmvAAAAAACjECgBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKDyHEgBw37O1spC5ZYUyu76Dg41R52Wm3ch9biYAAPcpAiUA4L5nbllBhx37lHUZxeacNF8iUAIA7mNMeQUAAAAAGIVACQAAAAAwCoESAAAAAGAUAiUAAAAAwCgESgAAAACAUQiUAAAAAACjECgBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQC5UMqODhYY8eOLdKxqampCg4O1meffXbX1502bZp69ux51+0AAAAAKHsEStxWSkqK+vfvr1OnTt11W8uXL9eiRYtKoCoAAAAA5YF5WReA8mvbtm2aNGmSrly5clftnD17VjExMdqwYYNsbGxKpjgAAAAAZY4RyofYzZs3NXXqVDVr1kweHh4KDw/XyZMnDfsjIiLk4uKi9957r9DzZ8+erZCQEG3evFnt27eXi4uLnnvuuQJTY2fMmKHDhw9r8eLFatiwodH15uTkaMmSJQoJCZGbm5uee+45LVu2LN8x33//vTp37ixPT081bdpU48aN0+XLl4td87lz5xQZGSk/Pz95eXmpR48e+vXXXw37s7OzlZiYqFatWsnFxUUhISFatWpVvja6d++uUaNGaeDAgfLy8tLgwYP1999/y8nJSRs3btTAgQPl6ekpHx8fRUdHKz093ei+AQAAAMoCgfIh9vnnn+v48eOaNm2a3njjDR06dEjDhw837F+3bp2mT58uOzu7W7Zx9uxZTZ48WT169ND8+fNVq1YtjRo1Sn/++afhmIiICK1fv15+fn53VW9MTIxiYmL07LPPKjExUe3bt9fkyZP1wQcfSJI+/fRT9enTR7Vr19bMmTP1+uuva8uWLXrllVfyhbU71ZyamqouXbpo7969ioqK0syZM5Wdna1evXrpzJkzkqQJEyYoPj5eYWFhSkxMVFBQkKKjowsE3C+++EKVKlXSnDlz1KVLF8P2cePG6YknntDcuXMVHh6uVatWad68eXfVPwAAAEBpY8rrQ6xGjRqaM2eOHnnkEUlSUlKSEhISlJaWJktLS9WvX/+ObaSlpSkhIcEQFp988kkFBQXp+++/15NPPilJqlu37l3Xeu3aNb3//vvq2bOnhg0bJklq1qyZzpw5oz179qhLly6Ki4tTYGCgYmJiDOc1aNBAL774olavXq2uXbsWqeY1a9bo1KlTWrdunaEPPDw8FBYWpv3796thw4ZauXKlIiMj1bt3b0lS8+bNlZWVpZkzZ6pTp06qVKmSJMnc3FyTJk1SxYoVJUl///23JCkoKEijRo2SJDVt2lTbt2/Xd999p6FDh951XwEAAAClhRHKh5iHh4chTEpSrVq1JEnJycnFasfLy8vw7+rVq0tSiU/fPHDggDIzM9W6det829966y29++67OnHihC5cuKB27drl2+/m5iZHR0ft2rWryDXv27dPjo6O+QK1jY2NNm3apLZt22rnzp3KyclRUFCQMjMzDT/BwcFKTk7Wzz//bDivdu3ahjB5q+vn1cCUVwAAANxvGKF8iOWNouUxNc39/4WcnJwit2FmZiYLC4sCbWRnZ5dAhf8nb2Ege3v72+53cHAosM/e3l4pKSmG13eq+cqVK7e8zj+vFRISUuj+c+fO5bt2Yf4dMk1NTUu8zwAAAIB7jUCJ+0Le6rCXLl1S7dq1DdtPnjyp06dPG77nef78+QLnnj9/Xu7u7sW61unTpwts37t3r6pUqWKoZfny5YWOPuaN9AIAAAAPOqa84r7g7u6uRx55RFu2bMm3PSEhQWPGjFGdOnVUpUoVrV+/Pt/+n3/+WSdPniwwxfR2vLy8lJSUpD/++MOwLTU1Vf369dP69evVuHFjSdLVq1fl6upq+Dl9+rRmzZrF1FUAAAA8NBihxH3Bzs5O3bp108KFC2Vubq7GjRtr3759WrNmjSZNmiRTU1MNHTpU48aNU2RkpEJDQ3X27FnNnDlTTz31lMLCwop8rU6dOmnZsmXq37+/Bg0apMcee0wLFy5UhQoV9OKLL6pq1aoKDQ3VmDFjdPLkSTVs2FDHjh1TXFycGjVqpJo1a97DngAAAADKDwIl7huRkZGys7PTypUrNX/+fDk6Ourtt982hMUXX3xRlpaWeu+99zRgwAA99thjatmypV5//XVZWloW+TrW1tb64IMPNG3aNE2cOFE5OTny8vLS+++/r6pVq0qSpk6dqsTERC1fvlxnz55VlSpV1KlTJw0ePPie3DsAAABQHpnkFGcFFgDlzsWLKcrOLp+/xg4ONjp/vnirBuPuPKx97uBgo8OOfcq6jGJzTpr/UL5fd+th/ZyXJfq89NHnpY8+L5ypqYns7a1vuZ8RSpSp7OzsIq1uam7ORxUAAAAob/grHWVqzpw5io+Pv+NxmzdvZvVUAAAAoJwhUKJMvfTSSwoMDLzjcXnfXQQAAABQfhAoUaaqVaumatWqlXUZAAAAAIzAcygBAAAAAEYhUAIAAAAAjEKgBAAAAAAYhUAJAAAAADAKgRIAAAAAYBQCJQAAAADAKDw2BABw38tMuyHnpPllXUaxZabdKOsSAAC4KwRKAMB973JqhpSaUSbXdnCw0fnzyWVybQAAyhpTXgEAAAAARiFQAgAAAACMQqAEAAAAABiFQAkAAAAAMAqBEgAAAABgFAIlAAAAAMAoBEoAAAAAgFF4DiUA4L5la2Uhc8sKZV2GHBxsin1OZtqN3OdnAgBwHyNQAgDuW+aWFXTYsU9Zl2EU56T5EoESAHCfY8orAAAAAMAoBEoAAAAAgFEIlAAAAAAAoxAoAQAAAABGIVACAAAAAIxCoAQAAAAAGIVACQAAAAAwCoESAAAAAGAUAiUAAAAAwCgESgAAAACAUQiUgBFycnLKugQAAACgzJXLQBkVFaXWrVvf83PwYImLi5Ovr688PDz0+eefa/fu3QoJCZGLi4v69u1bYtf59NNPNW3aNMPr1atXy8nJSWfOnCmxawAAAAD3A/OyLqAwAwYMUGpq6j0/Bw+OP/74Q/PmzdNLL72kjh07qk6dOurXr5+ysrI0f/58ValSpcSulZiYKG9vb8PrwMBArVixQnZ2diV2DQAAAOB+UC4DZe3atUvlHDw4rly5Iklq166dGjdubNjm5eWlZs2a3dNr29nZESYBAADwUCrxKa+ZmZlasmSJ2rVrJzc3N7Vs2VIJCQnKysqSJHXv3l1jx45VfHy8/Pz81KRJEw0bNkyXLl0ytPHv6avBwcGKj4/X1KlT1axZM7m7uys8PFxJSUm3POdOdeTVMn78eM2bN08BAQFydXVV586ddejQoWLf99KlSxUSEiJXV1e1aNFCEyZMUEpKimF/dna2EhMT1apVK7m4uCgkJESrVq3K10ZWVpbmzZun0NBQubm5ycPDQ126dNGuXbsMx1y/fl0TJkzQM888Y2hn4cKFxar1lVde0X/+858C23v06KFu3boVuRZJOnDggHr16iUvLy81bdpUkZGRunjxomH/uXPnFBkZKT8/P3l5ealHjx769ddfi1WvJH3zzTd64YUX5OrqqubNm2vatGnKyMiQJM2ePVsvv/yy4R6Cg4Pl5OSkpKQkrV27Vk5OToa6f//9d7366qvy9PSUt7e3hgwZUmCq6u1qDg4O1l9//aU1a9bIyclJf//9d74pr3nXO378eL4287afPHlSknTq1CkNHTpUTZo0kYeHh8LDw3Xs2LFi9wsAAABQlko8UI4dO1axsbFq06aNEhIS9Pzzz2vOnDmKjo42HLNx40atX79eEydOVFRUlLZv366IiAhlZ2ffst0lS5boxIkTmjJliiZNmqRffvlFo0ePvqs6JGnDhg3asmWLoqOjFRcXpwsXLmjIkCG3reXfvvjiC73zzjvq2rWrFi5cqNdee02fffaZJk+ebDhmwoQJio+PV1hYmBITExUUFKTo6GgtW7bMcExMTIwSExPVpUsXLViwQJMmTdLly5c1ZMgQpaenS5Lefvttbd26VVFRUVq4cKFatmypmJgYrVmzpsj1duzYUQcOHNDff/9t2HbhwgXt2bNHHTp0KHIthw8fVrdu3ZSVlaWYmBhFR0dr7969hu8rpqamqkuXLtq7d6+ioqI0c+ZMZWdnq1evXsX6vuHnn3+ugQMHql69epozZ4769eunFStWaPjw4ZKkF198URMnTpQkjR8/XjNmzNCKFStUvXp1BQQEaMWKFWrUqJFOnDihLl266OrVq3rnnXc0adIkHTlyRF27dlVycnKRao6Pj8/XbtWqVfPV2rp1a1laWmrDhg35tm/YsEGenp564okndOnSJXXp0kW//fabJkyYoNjYWKWmpurll1/WqVOnitwvAAAAQFkr0SmvR48e1dq1axUZGanw8HBJkr+/vypWrKjY2Fj17NlTkpSenq5FixapRo0aknKnDPbr109bt25VYGBgoW1XrlxZc+fOlZmZmSTpr7/+0uzZs5WcnCwbG5ti11G/fn1JuSNxCxYskLW1taTcQDFq1CgdOXJEDRo0KNJ97969W7Vq1VK3bt1kYmIiHx8fWVpa6urVq5KkEydOaOXKlYqMjFTv3r0lSc2bN1dWVpZmzpypTp06qVKlSjp37pyGDRumrl27GtquUKGCBg0apKNHj8rNzU27d++Wv7+/2rZtK0ny9fWVpaWlbG1ti1SrJD333HOaOHGiNmzYoD59+kiSvvzyS5mZmSkkJESSilRLYmKi7O3ttWDBAllYWEjKfZ/Gjx+vpKQk/fDDDzp16pTWrVtn6G8PDw+FhYVp//79hnu4nZycHMXGxiooKCjfQjjVq1fXa6+9pn379snb21tPP/20JKlu3bpyd3eXJFlYWMjOzk4eHh6SpDfeeEOWlpZavHixrKysJElNmjRRq1attHz5cvXv319r1qy5Y83/bvefrKys1LJlS61fv14DBw6UlDv1dseOHRo7dqyk3NHsq1evauXKlapevbqk3M9D69atlZCQoLfeeuuO/QIAAACUByU6Qrlnzx5JUmhoaL7teaNeefu9vb0NYVLKXdTEwsJCe/fuvWXb7u7uhjApyfCHeFpamtF1SJKTk5MhTEpStWrVbtnurfj5+enEiRMKCwtTfHy8Dh06pPbt26t79+6SpJ07dyonJ0dBQUHKzMw0/AQHBys5OVk///yzJGnGjBnq3r27Ll26pL179+rTTz/VunXrJEk3b96UlBsgV65c+f+1d+dhVVX7H8ffzCSgoZJd5SfOiAKKJKDmgFKCol6N1ELNLAdwCOf0aklmDnlRERXNzAExy1Iru1aWN1Mrhwbtp9YvIUQFM8wBHBA4vz94OHUC9HBEUPu8nsfnibXXWue7117a+bL2XpuhQ4eSlJREeno6I0eOLDURL4mzszNdunQxWUXbtm0bwcHBVK1a1exYDh48SMeOHY3JJEDbtm3ZsWMHHh4eHDx4EA8PD2NiBuDi4sKOHTvMSiYBUlJSyMzMpHPnziZj1759e+zs7Ni7d6/Z5/3VV18RFBSEg4ODsR9XV1d8fX2N/ZRHzL169SIlJYVjx44BhSvyAGFhYQB8+eWXNG/enJo1axrjsLW1pV27dmU6HxERERGRylauK5RFK3I1atQwKS/6uei2wr/eJmhlZUX16tW5ePFiqX07Ojqa/GxtXZgLl/Q+QHPjuFG/ZbnltVu3bhQUFJCcnMzSpUtZvHgxderUYcKECXTr1s24YUzR6t9f/frrrwAcPnyY2NhYDh8+zH333UejRo2oXbu2yXn+61//4sEHH+S9995j5syZzJw5Ez8/P2bMmGH2iioUJj3Dhg0jJSUFR0dHvvvuOxISEozHzYnl/PnzN9yM5vz588WuQVkVjd306dOL3a4Mf4yduX29//77vP/++8WO1atXz1jnVmNu27Ytbm5ubNu2jaZNm/LBBx/QsWNH7r//fuNnpKWl0bx582Jt7ezsbumzRUREREQqUrkmlEWrW1lZWcaVPoCzZ88CGG/LLEoSihgMBrKyssptp0xz4yhP4eHhhIeHc+nSJXbv3s1rr73GxIkTCQgIMN6Sm5SUVCyBBXB3dyc7O5tnn30WLy8vtm3bRoMGDbC2tubzzz/no48+Mta1t7cnKiqKqKgoTp8+zc6dO1m6dCkTJ04sMVEqTbt27ahRowbbt2/H3t6eatWq0aFDBwCzY3F2djbZTAkKE/Fdu3bh4+ODi4sLGRkZxT77wIED1KxZ05jE3UjR2E2ZMsXkVR1FynItnZ2d6dChA4MGDSp2rGiVtTxitrGxITw8nO3btzNo0CD279/PwoULTeIICgpiwoQJZscuIiIiInInKtdbXgMCAoDCTWr+rOjnooTgm2++MVmN/Oyzz7h+/TpBQUEVGkd5GT9+vPF5ORcXF8LCwoiOjiYvL4+zZ88aX2Nx4cIFfHx8jH8yMjKIj4/nypUrpKSkcP78eQYPHkyjRo2MK6W7du0CCpPu3NxcQkNDWbVqFQC1a9cmMjKS7t27l5gE3YitrS3du3dn586dbN++nbCwMGNSZU4sUDiOu3fvNt4CC4XXdvjw4aSmptKqVSvS0tI4fvy48XhOTg4jRoxg27ZtZsXZsGFDqlevzqlTp0zGztXVlfnz55v0fTMBAQEcP36c5s2bG/tp1qwZK1asMJ6bOTEXjceN9OrVixMnTpCYmIizszPBwcEmcaSmptKwYUOTc3rrrbfMHhcRERERkTtBua5QNm7cmJ49e7JgwQKuXLmCn58f3377LYmJifTq1YtGjRoBhV/Qhw0bxvDhw/ntt9+YP38+7dq1K7eE0tw4yktQUBDTpk1j7ty5dOjQgYsXL5KQkED9+vVp3Lgxtra2hIeHM3XqVNLT0/Hy8uLnn38mLi6O5s2bU7t2bVxcXHB2dmbp0qVYWVlhbW3Nxx9/zKZNm4DCZzrt7e3x8fEhISEBOzs7PD09SU1NZfPmzXTt2rXMcffq1cu4y+zUqVON5fXr179pLADR0dH079+fESNGMGDAAC5fvkxcXBwBAQG0atWKpk2bsm7dOqKiohg9ejTVqlXj9ddfx8HBgccff9ysGG1sbIiJiSE2NhZra2s6dOjAhQsXiI+P59KlSzRr1szs8x05ciR9+/YlKiqKvn37YmtrS1JSEnv37uWJJ54AICIi4qYxV61alSNHjrBv3z58fX1L/CwvLy+aNGnChg0beOyxx0yeM3366afZsmULQ4YMYfDgwVStWpUtW7awdetWXnnlFbPPR0RERESkspVrQgkwe/ZsPDw8ePfdd0lMTKR27dqMHj2aZ5991lgnICAAPz8/Jk6caEy2yvv2P3PiKC+PP/44ubm5JCcnk5ycjKOjo/GdjLa2hUM8Z84cEhMTSUpK4syZM9SsWZOIiAjGjBkDFK5sLl26lHnz5jFmzBicnJzw8vIiKSmJoUOHGjfAiY2NxdXVlVWrVnH27Flq1KhBREQEMTExZY7b29ubBg0akJubS6tWrYzl5sbi7e3NmjVrWLBgAc899xxVq1alc+fOjB8/Hmtra5ydnVm/fj1z587lpZdewmAw0KpVK9auXVvsOdob6devH87OzqxcuZLk5GScnZ2N7y91c3Mzu5+mTZuyfv16Fi5cyIQJE7CysqJp06asWLGCtm3bApgV84gRI5g+fTrPPPMMa9asKfXzevbsyfz5842bQRWpVasWb775JnFxcUyfPp3r16/ToEED4uLi6N69u9nnIyIiIiJS2awMJe1qcxsNHDgQGxsbVq9eXZEfK3LPysrKpqCgQv8am83NzYWzZy/dvKKUm7/bmLu5uXDEY1hlh2GRZmkr/lbXqjz93eb5nUBjXvE05hVPY14ya2sratRwLvV4ua9Q3kvy8/NL3EX2z6ysrExeZ1KZCgoKzNqdtmjVtDIZDAby8/NvWs/GxgYrK6sKiEhERERERMqq8jOLO9gjjzzCqVOnblgnICDA+BxiZZs6dSqbN2++ab0ff/yxAqK5sc2bNzNlypSb1lu7di2BgYEVEJGIiIiIiJRVhSeUd0ryZY5ly5aRm5t7wzpOTk4VFM3NjRo1isjIyMoOwyzBwcHGTX5upH79+hUQjYiIiIiIWEIrlDfg6elZ2SGUibu7O+7u7pUdhllcXV1vy/tARURERESk4pTreyhFRERERETk70MJpYiIiIiIiFhECaWIiIiIiIhYRAmliIiIiIiIWEQJpYiIiIiIiFhECaWIiIiIiIhYRK8NERGRu1be5Ws0S1tR2WFYJO/ytcoOQURE5JYpoRQRkbvW7zm5kJNbqTG4ublw9uylSo1BRESksuiWVxEREREREbGIEkoRERERERGxiBJKERERERERsYgSShEREREREbGIEkoRERERERGxiBJKERERERERsYgSShEREREREbGIEkoRERERERGxiG1lByAici9zdbLHtopDhX6mm5vLLfeRd/kav+fklkM0IiIici9TQikichvZVnHgiMewyg6jzJqlrQAllCIiInITuuVVRERERERELKKEUkRERERERCyihFJEREREREQsooRSRERERERELKKEUkRERERERCyihFJEREREREQsooRSRERERERELKKEUkRERERERCyihFJEREREREQsooTyDmAwGCo7BBERERERkTJTQlkG7777Lp6enmRmZpZLf7m5ucyZM4f333+/XPqTu9vixYtp1qxZZYchIiIiImI2JZSV6Ny5c7zxxhvk5eVVdigiIiIiIiJlpoRSRERERERELKKEshQFBQUsXbqUTp060aJFC6Kjo7lw4YJJnc8//5z+/fvj5+dHmzZtmDZtGr///rtJne+++46nn36aVq1a0aZNGyZNmkRWVhYnT56kY8eOAEyZMoXOnTub3e/ixYsJDQ0lPj6ewMBAunbtSk5Ojlnn9fzzzzNkyBCmT5+Ov78//fv3x2AwUFBQQGJiIiEhIXh7exMaGsrbb79drP2WLVv45z//SYsWLejcuTPx8fHk5+ff9HzLYs+ePfTt2xc/Pz9at25NdHQ0x48fN6nzySef0KdPH3x8fHj44YeZO3cuubm5JnU++ugjnnjiCfz8/PD29iYsLIzk5GSTOmvWrCE0NBQfHx/at2/PjBkzyM7ONh6/evUqixYtomvXrvj4+NCtWzc2btxo0kfnzp1JSEhgzpw5tG3blhYtWvDMM8+QlpZmUm/jxo306dOHli1b4uvrS+/evfnoo4/KNDYiIiIiIncSJZSlePXVV1myZAkREREkJCTg6urKv//9b+Pxd955h2HDhlG3bl0WLVrE2LFj2blzJ4MGDeLKlSsAHDlyhAEDBpCfn8+8efOYPn06Bw4cYPjw4TzwwAMsW7YMgKioKBISEszuFyA9PZ2dO3cSFxdHTEwMTk5OZp/b119/zenTp0lISGD48OFYWVkxY8YMEhIS6N27N4mJiQQHBzN9+nTWrVtnbLd+/XomT56Mr68vS5YsYfDgwbz22mvGcbnR+ZorPT2d6OhovL29WbZsGS+//DIpKSkMHz7cuHnR+++/z6hRo2jcuDFLlixhxIgRbNy4kfHjxxv7+fTTTxkzZgy+vr4sXbqUxYsX4+7uTmxsLIcOHQLggw8+4NVXXyUyMpLXX3+dkSNHsnXrVmbNmgUUbpY0dOhQ1qxZwxNPPMGyZcto27YtL774IkuWLDGJe/Xq1aSmpjJ79mxmzpzJDz/8wJQpU4zH165dS2xsLI8++ijLly9n/vz52NraMn78eM6cOWP2+IiIiIiI3ElsKzuAO9HFixdZt24dQ4YMYdSoUQC0b9+eM2fO8MUXX1BQUEBcXBydOnVi3rx5xnZNmzbl8ccf59133yUyMpLExERq1KjBypUrsbe3B+D+++/nhRdeICMjw7gBS926dWnWrJnZ/QLk5eXx/PPPExgYWObzy8vLIzY2Fnd3dwBSU1N56623mDRpEkOGDAHg4YcfJj8/n0WLFhEREYGDgwNLliwhNDSUl156yVjn4sWL7NmzB4PBcMPzTUtLw8PD46axHTp0iKtXrzJ8+HBq1aoFwD/+8Q8+/fRTcnJycHJyYv78+QQHBzN37lxjuwcffJCRI0dy8OBB/P39OX78OH369DFJ6vz8/AgMDGTfvn34+vqyb98+3N3dGTBgAFZWVgQEBFClShXjSvTnn3/Ovn37WLRoEaGhocZzzsvLIzExkSeffBJXV1fjeS5duhQbGxsATpw4weLFi7l06RIuLi6cPHmSZ599lhEjRhjjqVOnDn369OGbb74hLCyszNdRRERERKSyKaEswXfffcf169fp0qWLSXlYWBhffPEFp0+f5rfffqN79+4mx319ffHw8ODrr78mMjKSgwcP0qVLF2NyBdC2bVt27NgBUGy32NTUVLP6LdKkSROLzq9KlSrGZBLgq6++wmAwEBwcbLJBUOfOnVmzZg2HDh2iZs2aZGVl8cgjj5j0NWrUKGPSfbPzNUeLFi1wcHAgIiKC0NBQOnToQGBgIL6+vgAcP36czMxMRo4caRJr+/btsbOzY+/evfj7+zNs2DAAcnJySE1N5cSJExw+fBiA69evAxAUFMTGjRvp3bs3ISEhdOzYkR49emBlZQXA/v37sbOz49FHHzWJsUePHmzYsIHvv/+eTp06GeMuSiahMMEFuHz5Mi4uLkydOhUo/GVFSkoKaWlpfP311ybxiIiIiIjcbZRQlqBohap69eom5W5ubkDh85V//vnPatSoYXwG7/z588X6uJHz58+b1S+AjY2NcXWsrGrWrFni5xatwv3Vr7/+iq2trTGO0pT1fEvi7u5OUlISK1asYNOmTaxdu5aqVavy5JNPEhMTY4x1+vTpTJ8+vcRYoXAH3RdffJEdO3ZgZWWFh4cH/v7+wB/v/ezWrRsFBQUkJycbb4utU6cOEyZMoFu3bly4cIEaNWpgbW16Z3jR+F26dMlY5ujoaFKnqE3RZ504cYIXXniBL7/8Ejs7Oxo0aEDTpk1N6oiIiIiI3G2UUJagKFH77bffqFu3rrG8KJkpWsE6e/ZssbZnz56lRYsWADg7O3Pu3DmT4wUFBezatQsfH59ibatVq2ZWv+XNxcUFgKSkpGKJERQmeUUx/fV8fvvtN/7v//6PVq1a3fR8b5SM/pmvry8JCQnk5uZy8OBBNm7cSGJiIs2aNaN+/fpA4UZGRQninxVduwkTJpCamsrq1avx8/PD3t6eK1euFNtoKDw8nPDwcC5dusTu3bt57bXXmDhxIgEBAVStWpWsrCwKCgpMksqisTA3oS8oKGDYsGE4ODiwadMmvLy8sLW15eeff2br1q1m9SEiIiIicifSpjwl8PPzw9HRke3bt5uU79y5Eyh89q1mzZps27bN5PihQ4dIT0+nVatWAPj7+7N7926TWxq/+eYbhg8fTmpqarGVrwYNGpjVb3l76KGHgMKVWR8fH+OfjIwM4uPjuXLlCg0aNOD+++/ns88+M2m7ceNGoqOjgZufrznWrVtH586dyc3Nxd7enjZt2jBz5kwAMjIyaNiwIdWrV+fUqVMmsbq6ujJ//nzjbrAHDx4kNDSUwMBA4y24u3btAv5YYR4/frzxdl0XFxfCwsKIjo4mLy+Ps2fPEhAQwPXr1/n4449NYvzggw+ws7Mz3oZ7M7///jupqan07dsXHx8f42pvUTxaoRQRERGRu5VWKEvg5OREdHQ0CxcuxNHRkYCAAP773/8aE0pra2tiYmKYNm0akyZNIjw8nDNnzrBo0SLq169P7969AYiOjqZ///6MGDGCAQMGcPnyZeLi4ggICKBVq1ZcvXoVKysrvvzySxo2bEiLFi3M6re8NW3alPDwcKZOnUp6ejpeXl78/PPPxMXF0bx5c2rXrg0UPi85a9YsXF1d6dy5Mz/99BMrVqzgmWeewcHB4abna46goCDmzZvHyJEjGTBgADY2Nrz55ps4ODgQHByMjY0NMTExxMbGYm1tTYcOHbhw4QLx8fFcunTJuNGRr68v7733Hl5eXtSqVYtvvvmGFStWYGVlZdwtNygoiGnTpjF37lw6dOjAxYsXSUhIoH79+jRu3BhPT09at27Nv/71LzIzM2ncuDGff/45b775JlFRUVStWtWsc6pRowZ16tRh7dq1PPDAAzg7O/PFF1+wdu1aoPA5SxERERGRu5ESylIMHz6cKlWqsGbNGt544w38/PyYPHkyM2bMAODxxx+nSpUqvPbaa0RHR1OtWjW6dOnC2LFjqVKlCgDe3t6sWbOGBQsW8Nxzz1G1alU6d+7M+PHjsba2pkqVKkRFRbF69Wp27drFnj17zOr3dpgzZw6JiYkkJSVx5swZatasSUREBGPGjDHWGThwIPfddx+rVq3izTffpHbt2owZM4ann37arPM1R+PGjVm+fDmLFy9m3Lhx5Ofn4+3tzapVq4y7xPbr1w9nZ2dWrlxJcnIyzs7OtG7dmnHjxhmfP50zZw4zZ8407khbr149YmNjee+99zh48CBQeA1zc3NJTk4mOTkZR0dH47szi1YRly9fzsKFC1m5ciUXLlygXr16zJgxg/79+5dpfJcuXcqsWbOYNGkS9vb2NGrUiGXLlvHKK69w8OBBnnzyyTL1JyIiIiJyJ7Ay6H47kbtaVlY2BQV35l9jNzcXzp69dPOK9zA3NxeOeAyr7DDKrFnair/9tTOX5nnF05hXPI15xdOYVzyNecmsra2oUcO51ONaobwHGAwG8vPzb1rPxsbGuKFQRSsoKDA+u3gjRSuDIiIiIiJy59O393vAvn37GDRo0E3rzZ49mz59+lRARMVNnTqVzZs337Tejz/+WAHRiIiIiIhIeVBCeQ9o3rw5mzZtumk9d3f3CoimZKNGjSIyMrLSPl9ERERERMqfEsp7gLOzc4nvtbyTuLu7V2pCKyIiIiIi5U/voRQRERERERGLKKEUERERERERiyihFBEREREREYsooRQRERERERGLKKEUERERERERiyihFBEREREREYsooRQRERERERGL6D2UIiK3Ud7lazRLW1HZYZRZ3uVrlR2CiIiI3AWUUIqI3Ea/5+RCTm6FfZ6bmwtnz16qsM8TERGRvzfd8ioiIiIiIiIW0QqlyF3O2tqqskO4oTs9vnuRxrziacwrnsa84mnMK57GvOJpzIu72ZhYGQwGQwXFIiIiIiIiIvcQ3fIqIiIiIiIiFlFCKSIiIiIiIhZRQikiIiIiIiIWUUIpIiIiIiIiFlFCKSIiIiIiIhZRQikiIiIiIiIWUUIpIiIiIiIiFlFCKSIiIiIiIhZRQikiIiIiIiIWUUIpIuVu7ty5DB482Ky6hw8fZuDAgfj5+fHwww8TFxfH9evXb2+A94icnBxiY2Np164dfn5+DB06lF9++eWm7QYPHoynp2exP4cPH779Qd9lPvjgA7p3746vry9hYWFs2bLlhvUtvSbyh7KO+datW0uczy+99FLFBHwPOXr0KM2bNyczM/OG9TTPy4+5Y655fmsKCgrYsGEDPXr0wM/Pj5CQEGbPnk12dnapbTTPzWdb2QGIyL0lKSmJVatW0aZNm5vWTUtLY/Dgwfj5+bFw4UKOHz/OggULyM7O5oUXXqiAaO9uY8eO5fDhw0yaNAknJycSEhIYNGgQ27Ztw8XFpdR2x44dY9CgQXTv3t2kvGHDhrc75LvKf/7zHyZMmMCgQYNo3749O3bsYPLkyTg6OhIaGlpiG0uviRSyZMyPHTuGh4cH8+bNMymvWbNmRYR8z0hJSWH48OHk5eXdtK7mefkoy5hrnt+alStXsnDhQp555hnatGlDamoq8fHx/Pzzz7z++uslttE8LwODiEg5yMzMNIwbN87QtGlTg7+/v+Gpp566aZupU6caOnbsaLh27ZqxbP369QYvLy9DZmbmbYz27rd//35DkyZNDJ9//rmxLCsry9CyZUvD8uXLS22XmZlZrJ2ULCQkxBATE2NS9txzzxlCQ0NLrG/pNZE/lHXMDQaD4emnny7WRsx3/fp1Q1JSksHPz88QEBBgaNKkiSEjI6PU+prnt66sY24waJ7fioKCAkPr1q0NM2bMMCnftm2boUmTJoYjR44Ua6N5Xja65VVEysWCBQs4cuQIb7zxBl5eXma12bNnD8HBwdjb2xvLQkNDyc/PZ/fu3bcr1HvCnj17cHJyol27dsay6tWr07p1a3bt2lVqu2PHjgHg6el522O8m6Wnp3PixAkeffRRk/KuXbuSkpJCenp6sTaWXhMpZMmYQ+Gc1ny23MGDB5k/fz5DhgxhwoQJN62veX7ryjrmoHl+K3JycujZsyfh4eEm5Q0aNADgxIkTxdponpeNEkoRKRfPPvss27ZtIygoyKz6V65cISMjg/r165uUV69eHWdnZ1JTU29HmPeMlJQUPDw8sLGxMSmvW7fuDcfu2LFj2NvbEx8fT2BgID4+PgwdOlTj/RcpKSkAxeanh4cHQInjZek1kUKWjPmvv/5KVlYWR44cITQ0lObNm9O1a9ebPncpf2jYsCE7duxg1KhRxeZuSTTPb11Zx1zz/NY4Ozszbdo0/P39Tcp37NgBQKNGjYq10TwvGz1DKSI3lJeXx7Zt20o9XrNmTdq1a1fiP8g3cunSJaDwH/q/cnJyuuGD8vc6c8Y8OzvborE7duwYubm5ODo6kpCQQEZGBkuWLCEyMpKtW7fi5uZWLudwtyttfjo5OQGUOMaWXhMpZMmYF624nzx5kokTJ+Lg4MCWLVuYPHky+fn5PPbYY7c56rtfWZ/B0zy/dWUdc83z8vf999+zYsUKQkJCStw/QPO8bJRQisgNXbt2jUmTJpV6PCAgwOSWEHMZDAYArKysSjxmbf33vYHCnDG3s7Mr9fiNxi4qKop+/fqZrCT7+fkRFhZGUlISY8eOtSzoe0xp87OovKQxLjpWkr/zfDaXJWPu7e1NYmIirVu3Nn75e/jhh8nKymLRokX6on0baJ5XPM3z8nXw4EFGjBiBu7s7L7/8col1NM/LRgmliNyQk5MTP/74Y7n3W/Q/xZJ+03f58uW/9Q5q5oz5mDFjOHnyZLHynJycEn+rWqRJkybFyv7nf/6Hhg0bGn8LLhjn31/nZ05OjsnxP3N2drbomkghS8a8evXqBAcHFyvv2LEje/fu5dy5c1SvXv02RPv3pXle8TTPy8+HH37I888/T7169Vi5ciWurq4l1tM8Lxul2CJSKZycnKhVqxZpaWkm5VlZWWRnZxd7jkpM1a9fn/T09GK/RU1LSyt17AwGA1u2bOHAgQPFjl29erXU/7H+HRWN4V83ayiaryWNsSXXRP5gyZh/++23vP3228XKr127hq2t7d/6F1O3i+Z5xdM8Lx9vvPEG48aNo2XLlqxfv54HHnig1Lqa52WjhFJEKk27du3YuXMnubm5xrKPPvoIGxsbAgICKjGyO9/DDz/MxYsX2bt3r7Hs3LlzHDhwgLZt25bYxsrKitdff51XXnmFgoICY/n//u//cuLECY35n3h4eODu7s727dtNyj/++GPq1atH7dq1i7Wx5JrIHywZ8++++45p06aZrK4XFBTw0Ucf0apVqxveGi6W0TyveJrnt+7tt99mzpw5hIWFsXLlypsm4ZrnZaNbXkWkwvz888/k5ubSrFkz4I+dYYcNG8ZTTz3FL7/8QlxcHH379i3xy6P8oXXr1gQEBDBu3DgmTJjA/fffz+LFi3FxceGJJ54w1vvrmI8ePZrRo0czYcIEHnvsMU6fPs2iRYvw8vKiV69elXU6d6SRI0cyZcoUqlWrRqdOnfjss8/4z3/+w4IFC4DCLxcnTpygUaNGODs7m31NpHRlHfM+ffqwbt06Ro0aRUxMDE5OTiQnJ/PTTz+xfv36Sj6be4PmecXTPC9fWVlZzJo1izp16hAZGcmRI0dMjtetWxdA8/xWVMbLL0Xk3jZgwADDU089VWJ5cHCwSdn+/fsNjz/+uMHb29vQvn17w7///W9Dbm5uBUV6dzt//rzh+eefNzz00EOGVq1aGYYOHWo4fvy4SZ2SxvyTTz4xPPbYY4aWLVsagoKCDNOnTzf8/vvvFRj53WPDhg2GRx55xODt7W0ICwszbN682XjsnXfeMTRp0sTw1VdfGcvMuSZyY2Ud85MnTxrGjh1raNu2rcHX19fw5JNPGvbv318Jkd/9isY3IyOjWJnm+e1h7phrnltu8+bNhiZNmpT6Z8uWLZrnt8jKYLjBNkYiIiIiIiIipdAzlCIiIiIiImIRJZQiIiIiIiJiESWUIiIiIiIiYhEllCIiIiIiImIRJZQiIiIiIiJiESWUIiIiIiIiYhEllCIiIlLusrOzWbVqFX369MHf35+WLVsSERHBxo0bKSgouG2fazAYePXVVwkMDKRly5asX7+eU6dOMXDgQHx9fQkMDOTw4cN4enqyePHiMvXt6enJ888/f5siL5Senn5b+xcRKW96D6WIiIiUq5SUFKKiojh16hQ9evTA29uba9eu8emnn3LgwAF69OjBq6++ipWVVbl/9s6dOxkxYgSdOnUiJCQEf39/Fi5cyCeffMKoUaNwc3MjPDycTz75BE9PT5o2bWp231u3bqVu3br4+fmVe9wA77zzDrGxsRw6dOi29C8icjvYVnYAIiIicu+4du0a0dHRnD9/nk2bNpkkbEOGDCE2Npbk5GR8fX0ZNGhQuX/+jz/+CMC4cePw9PQ0lnl5eTFy5EhjvV69epW5b0valMX+/fu5du3abf0MEZHyplteRUREpNwkJyeTmprKlClTSlz9mzx5MtWqVePNN9+8LZ9//fp1AJycnEzK/vyziIiUH93yKiIiIuUmIiKC48ePs2/fPuzs7Eqs88svv1C7dm3s7e0BOHDgAAkJCXz//fcA+Pj4MHr0aFq3bm3S7ttvvyU+Pp7vvvsOAD8/P2JiYvD19QWgc+fOnDp1yli/Tp06Jj8DjBo1it69e9OlSxdGjRrF6NGjjce2bt3K2rVrOX78OFWrVqVTp07ExMRQvXp1oPAZyt69ezNnzhxjm507d7J8+XKOHj2Kvb09QUFBjBs3jvr16xvreHp6Mn78eOzs7Fi/fj2ZmZnUr1+f6OhowsLCABg4cCD79u0ztvnr54iI3Km0QikiIiLlwmAwcPToUby9vUtNJgHq1atnTCY//fRTBg4cSEZGBlFRUURFRZGRkcHgwYP59NNPjW327NnDwIEDuXTpEs899xxRUVGcPn2ayMhIDhw4AMDUqVN55JFHAJgyZQoTJkxg3rx5uLq60qBBA+bNm2c8/levvfYakyZNwsHBgXHjxtG7d2/ee+89hg8fTl5eXolt3n33XaKiorjvvvuYOHEigwcP5ttvv6Vv376kpqaa1N2wYQOrV6+mb9++TJo0icuXLzN27Fh++uknAEaMGMFDDz0EwLx58+jXr585Qy4iUun0DKWIiIiUi99//528vDzc3NzMqp+Xl8dLL71ErVq1eOedd3B2dgagf//+hIeHExsbS4cOHbCxseHFF1/Ex8eHpKQkbGxsABgwYAD//Oc/efnll9myZQshISEcPXqUTz75hJCQENzd3QFYtGgRNWvWND4DefLkSZM4Lly4wOLFi2nfvj3Lly839u/u7s60adPYs2cPHTt2NGmTnZ3NrFmz6NatG3Fxccbyvn370r17d+bPn8+SJUuM5efPn+fjjz82jk2LFi3o27cv27Zto0mTJrRr147333+fAwcO3PZnNUVEypNWKEVERKRcWFsXfq3Iz883q/6RI0fIzMwkMjLSmEwCVK1alQEDBnDmzBl++OEHjhw5Qnp6OiEhIVy4cIFz585x7tw5rl69SnBwMEePHiUzM9PiuPfu3cu1a9eIjIw0JpMAPXv25N133yUgIKBYmz179pCdnU1ISIgxnnPnzmFjY0NQUBC7d+82Wdn09/c3SbS9vLwAOHv2rMVxi4jcCbRCKSIiIuWiWrVq2NnZce7cObPqF60U/vl5wyINGjQA4PTp08bXi8ybN4958+aV2FdGRgYPPvigJWEbn7P08PAwKXdwcKB58+Yltjlx4gQAY8eOLbXfc+fO8cADDwAYn8MsUnTL7+18J6eISEVQQikiIiLlwsrKCj8/P3744Qfy8vKwtS35a8aCBQtIT0+nS5cupfZVtGegnZ0dubm5ADz33HO0bNmyxPpFCaglipK6ohXWsrSZOXOm8dbav6pWrZrxv8vSt4jI3UQJpYiIiJSbRx55hH379vHhhx/Ss2fPYsevXr3Kpk2byM/P58knnwQgJSWlWL2iTW0efPBB4y20VapUoW3btib1Dh06xIULF3B0dLQ45n/84x8ApKWlUa9ePWN5bm4uEydOpEePHoSEhJi0qVOnDlC48vjXmL7++msKCgqMq5AiIvcy/bpMREREyk2/fv2oU6cOc+fONe5gWiQ/P58ZM2bw22+/MXToUFq0aIGbmxsbNmwgOzvbWC87O5vk5GTc3Nzw9vbG29sbNzc31q1bR05Ojkm9mJgYpkyZYvLsY1m1bdsWOzs73nrrLf78NrXt27ezffv2Uts4ODiwcuVK47svAc6cOUN0dDTz58833qprrqJVTN0GKyJ3E61QioiISLlxcHAgISGBIUOGEBERQY8ePfDx8eH8+fNs376do0ePEhoaytNPP421tTXTp08nJiaGxx57jIiICAA2bdrEr7/+Snx8PNbW1ib1+vTpQ0REBA4ODrz99tucPn2a+fPnl3p7rTlq1KjByJEjWbhwIUOGDCEkJITMzEySkpIIDAwkODi4WJvq1aszbtw4Zs+eTb9+/ejZsyd5eXkkJydz7do1Jk+eXOY4ip6zjI+PJzAwkDZt2lh8TiIiFUUJpYiIiJSrZs2asXXrVlavXs2uXbv48MMPMRgMeHp68sorr9CnTx/j6l3Xrl1ZtWoVS5cuZcmSJdja2tKiRQtmzZplfC/jn+stW7aMpUuXYm1tTePGjVm2bFmJCV9ZRUVF4ebmxtq1a5kzZw5ubm707duX0aNHl7r6OXjwYGrVqsUbb7zBggULcHR0pHnz5rz66qv4+/uXOYYnnniCr776ipUrV3L48GEllCJyV7Ay/PneDhEREREREREz6RlKERERERERsYgSShEREREREbGIEkoRERERERGxiBJKERERERERsYgSShEREREREbGIEkoRERERERGxiBJKERERERERsYgSShEREREREbGIEkoRERERERGxiBJKERERERERscj/AzZkDQ0Joe2/AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 864x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set(font_scale=1.5)\n",
    "plt.figure(figsize=(12,10))\n",
    "ax = sns.barplot(x='Coefficient', y='Feature', data=feature_coef, \n",
    "                 color = \"#F3005E\", order=feature_coef.sort_values('Coefficient', ascending = False).Feature)\n",
    "ax.set(ylabel=None)\n",
    "#ax.grid(False)\n",
    "def change_width(ax, new_value) :\n",
    "    for patch in ax.patches :\n",
    "        current_width = patch.get_width()\n",
    "        diff = current_width - new_value\n",
    "\n",
    "        # we change the bar width\n",
    "        patch.set_width(new_value)\n",
    "\n",
    "        # we recenter the bar\n",
    "        patch.set_x(patch.get_x() + diff * .5)\n",
    "\n",
    "change_width(ax, .25);\n",
    "plt.savefig(\"Feature_Coeffient.jpg\", bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 380,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 864x720 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,10))\n",
    "\n",
    "sns.set(font_scale=1)\n",
    "\n",
    "def make_confusion_matrix(model, threshold=0.3):\n",
    "    # Predict class 1 if probability of being in class 1 is greater than threshold\n",
    "    # (model.predict(X_test) does this automatically with a threshold of 0.5)\n",
    "    \n",
    "    y_predict = (model.predict_proba(X_test)[:, 1] >= threshold)\n",
    "    confusion = confusion_matrix(y_test, y_predict)\n",
    "    plt.figure(dpi=80)\n",
    "    sns.heatmap(confusion, cmap=plt.cm.YlOrRd, annot=True, square=True, fmt='d',\n",
    "           xticklabels=['Take Vaccine', 'Not Take Vaccine'],\n",
    "           yticklabels=['Take Vaccine', 'Not Take Vaccine']);\n",
    "    plt.xlabel('prediction')\n",
    "    plt.ylabel('actual');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 382,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 480x320 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "make_confusion_matrix(lr_model);\n",
    "plt.savefig(\"ClassBalance.jpg\", bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:metis] *",
   "language": "python",
   "name": "conda-env-metis-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  },
  "toc-showtags": false
 },
 "nbformat": 4,
 "nbformat_minor": 4
}