predicting_cluster.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split, cross_val_score, GridSearchCV\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import auc, roc_auc_score, roc_curve, accuracy_score,  recall_score, precision_score\n",
    "from regression_tools.dftransformers import (\n",
    "    ColumnSelector, Identity, FeatureUnion, MapFeature, Intercept)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from my_metr import plot_roc_curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "biz_df = pd.read_csv('biz_cluster.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 57173 entries, 0 to 57172\n",
      "Data columns (total 24 columns):\n",
      "Unnamed: 0                57173 non-null int64\n",
      "business_id               57173 non-null object\n",
      "RestaurantsPriceRange2    51111 non-null float64\n",
      "stars                     57173 non-null float64\n",
      "review_count              57173 non-null int64\n",
      "categories                57173 non-null object\n",
      "cl_0                      25785 non-null float64\n",
      "cl_1                      37611 non-null float64\n",
      "cl_2                      6306 non-null float64\n",
      "cl_3                      327 non-null float64\n",
      "cl_4                      11547 non-null float64\n",
      "cl_5                      14129 non-null float64\n",
      "cl_6                      1 non-null float64\n",
      "cl_7                      3366 non-null float64\n",
      "cl_8                      2667 non-null float64\n",
      "cl_9                      2073 non-null float64\n",
      "cl_10                     380 non-null float64\n",
      "cl_11                     2937 non-null float64\n",
      "cl_12                     470 non-null float64\n",
      "cl_13                     1706 non-null float64\n",
      "cl_14                     37 non-null float64\n",
      "cl_15                     0 non-null float64\n",
      "cl_16                     3076 non-null float64\n",
      "cl_17                     1397 non-null float64\n",
      "dtypes: float64(20), int64(2), object(2)\n",
      "memory usage: 10.5+ MB\n"
     ]
    }
   ],
   "source": [
    "biz_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>Unnamed: 0</th>\n",
       "      <th>business_id</th>\n",
       "      <th>RestaurantsPriceRange2</th>\n",
       "      <th>stars</th>\n",
       "      <th>review_count</th>\n",
       "      <th>categories</th>\n",
       "      <th>cl_0</th>\n",
       "      <th>cl_1</th>\n",
       "      <th>cl_2</th>\n",
       "      <th>cl_3</th>\n",
       "      <th>...</th>\n",
       "      <th>cl_8</th>\n",
       "      <th>cl_9</th>\n",
       "      <th>cl_10</th>\n",
       "      <th>cl_11</th>\n",
       "      <th>cl_12</th>\n",
       "      <th>cl_13</th>\n",
       "      <th>cl_14</th>\n",
       "      <th>cl_15</th>\n",
       "      <th>cl_16</th>\n",
       "      <th>cl_17</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>--9e1ONYQuAa-CB_Rrw7Tw</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1546</td>\n",
       "      <td>Steakhouses, Restaurants, Cajun/Creole</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>8.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>16.0</td>\n",
       "      <td>17.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>-VAsjhmAbKF3Pb_-8rh3xg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>10</td>\n",
       "      <td>Fast Food, Burgers, Restaurants</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>-cxD1NimFldATDUsN-oa3A</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>23</td>\n",
       "      <td>Mexican, Restaurants</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 24 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0             business_id  RestaurantsPriceRange2  stars  \\\n",
       "0           0  --9e1ONYQuAa-CB_Rrw7Tw                     4.0    4.0   \n",
       "1           1  -VAsjhmAbKF3Pb_-8rh3xg                     1.0    2.0   \n",
       "2           2  -cxD1NimFldATDUsN-oa3A                     2.0    2.0   \n",
       "\n",
       "   review_count                              categories  cl_0  cl_1  cl_2  \\\n",
       "0          1546  Steakhouses, Restaurants, Cajun/Creole   0.0   1.0   2.0   \n",
       "1            10         Fast Food, Burgers, Restaurants   NaN   1.0   NaN   \n",
       "2            23                    Mexican, Restaurants   NaN   1.0   NaN   \n",
       "\n",
       "   cl_3  ...    cl_8  cl_9  cl_10  cl_11  cl_12  cl_13  cl_14  cl_15  cl_16  \\\n",
       "0   NaN  ...     8.0   NaN    NaN    NaN    NaN   13.0    NaN    NaN   16.0   \n",
       "1   NaN  ...     NaN   NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "2   NaN  ...     NaN   NaN    NaN    NaN    NaN    NaN    NaN    NaN    NaN   \n",
       "\n",
       "   cl_17  \n",
       "0   17.0  \n",
       "1    NaN  \n",
       "2    NaN  \n",
       "\n",
       "[3 rows x 24 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "biz_cl = biz_df.copy()\n",
    "for i in range (18):\n",
    "    col_new = 'cl_t'+str(i)\n",
    "    col_old = 'cl_'+str(i)\n",
    "    biz_cl[col_new] = ~pd.isnull(biz_df[col_old])\n",
    "    biz_cl.drop(col_old, axis =1, inplace= True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(57173, 24)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_cl.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "biz_cl['how_many_cl'] = 0\n",
    "for i in range (18):\n",
    "    col_new = 'cl_t'+str(i)\n",
    "    biz_cl['how_many_cl'] += biz_cl[col_new]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5644"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(biz_cl[biz_cl['how_many_cl']>4])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2774"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(biz_cl[biz_cl['how_many_cl']>5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "57173"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(biz_cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def expand_features(X, col='features', inplace=False, sep=None):\n",
    "    X = X.copy() if not inplace else X\n",
    "    if sep==None:\n",
    "        for i in range(len(X[col].iloc[0])):\n",
    "             X['feature_' + str(i)] = X[col].apply(lambda x: x[i])\n",
    "    else:\n",
    "        lbls = np.unique(sep.join(X[col].unique()).split(sep))\n",
    "        lbls = np.unique([l.strip() for l in lbls])\n",
    "        for lbl in lbls:\n",
    "            X[lbl] = X[col].apply(lambda x: lbl in x)\n",
    "        X.drop(col, axis=1, inplace=True)\n",
    "    return X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "biz_cl1 = expand_features(biz_cl, col='categories', sep =',')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "biz_cl1['num_of_categ'] = biz_df.categories.map(lambda x: len(x.split(',')))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((57173, 732),\n",
       " Index(['Unnamed: 0', 'business_id', 'RestaurantsPriceRange2', 'stars',\n",
       "        'review_count', 'cl_t0', 'cl_t1', 'cl_t2', 'cl_t3', 'cl_t4', 'cl_t5',\n",
       "        'cl_t6', 'cl_t7', 'cl_t8', 'cl_t9', 'cl_t10', 'cl_t11', 'cl_t12',\n",
       "        'cl_t13', 'cl_t14', 'cl_t15', 'cl_t16', 'cl_t17', 'how_many_cl',\n",
       "        '& Probates', 'Acai Bowls', 'Accessories'],\n",
       "       dtype='object'))"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_cl1.shape, biz_cl1.columns[:27]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [],
   "source": [
    "colClust =['cl_t0', 'cl_t1', 'cl_t2', 'cl_t3', 'cl_t4', 'cl_t5',\n",
    "        'cl_t6', 'cl_t7', 'cl_t8', 'cl_t9', 'cl_t10', 'cl_t11', 'cl_t12',\n",
    "        'cl_t13', 'cl_t14', 'cl_t15', 'cl_t16', 'cl_t17']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "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>cl_t0</th>\n",
       "      <th>cl_t1</th>\n",
       "      <th>cl_t2</th>\n",
       "      <th>cl_t3</th>\n",
       "      <th>cl_t4</th>\n",
       "      <th>cl_t5</th>\n",
       "      <th>cl_t6</th>\n",
       "      <th>cl_t7</th>\n",
       "      <th>cl_t8</th>\n",
       "      <th>cl_t9</th>\n",
       "      <th>cl_t10</th>\n",
       "      <th>cl_t11</th>\n",
       "      <th>cl_t12</th>\n",
       "      <th>cl_t13</th>\n",
       "      <th>cl_t14</th>\n",
       "      <th>cl_t15</th>\n",
       "      <th>cl_t16</th>\n",
       "      <th>cl_t17</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>cl_t0</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.385155</td>\n",
       "      <td>0.286471</td>\n",
       "      <td>-0.011409</td>\n",
       "      <td>0.359006</td>\n",
       "      <td>0.303135</td>\n",
       "      <td>0.004614</td>\n",
       "      <td>0.064800</td>\n",
       "      <td>0.135711</td>\n",
       "      <td>0.097607</td>\n",
       "      <td>0.030550</td>\n",
       "      <td>0.142260</td>\n",
       "      <td>0.020644</td>\n",
       "      <td>0.139987</td>\n",
       "      <td>-0.013389</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.182701</td>\n",
       "      <td>0.122248</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t1</th>\n",
       "      <td>0.385155</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.207322</td>\n",
       "      <td>0.002877</td>\n",
       "      <td>0.283651</td>\n",
       "      <td>0.271736</td>\n",
       "      <td>0.003016</td>\n",
       "      <td>0.075600</td>\n",
       "      <td>0.117397</td>\n",
       "      <td>0.080720</td>\n",
       "      <td>0.040843</td>\n",
       "      <td>0.120563</td>\n",
       "      <td>0.031362</td>\n",
       "      <td>0.104162</td>\n",
       "      <td>0.005305</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.132429</td>\n",
       "      <td>0.083333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t2</th>\n",
       "      <td>0.286471</td>\n",
       "      <td>0.207322</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.005973</td>\n",
       "      <td>0.375981</td>\n",
       "      <td>0.256510</td>\n",
       "      <td>0.011878</td>\n",
       "      <td>0.036705</td>\n",
       "      <td>0.146902</td>\n",
       "      <td>0.101063</td>\n",
       "      <td>0.030295</td>\n",
       "      <td>0.150761</td>\n",
       "      <td>0.006901</td>\n",
       "      <td>0.182404</td>\n",
       "      <td>0.002018</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.237500</td>\n",
       "      <td>0.172109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t3</th>\n",
       "      <td>-0.011409</td>\n",
       "      <td>0.002877</td>\n",
       "      <td>-0.005973</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.001758</td>\n",
       "      <td>0.000640</td>\n",
       "      <td>-0.000317</td>\n",
       "      <td>-0.010102</td>\n",
       "      <td>-0.000279</td>\n",
       "      <td>0.000178</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.001889</td>\n",
       "      <td>-0.006905</td>\n",
       "      <td>0.001694</td>\n",
       "      <td>0.007190</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000418</td>\n",
       "      <td>0.010531</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t4</th>\n",
       "      <td>0.359006</td>\n",
       "      <td>0.283651</td>\n",
       "      <td>0.375981</td>\n",
       "      <td>-0.001758</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.324164</td>\n",
       "      <td>-0.002104</td>\n",
       "      <td>0.025576</td>\n",
       "      <td>0.162456</td>\n",
       "      <td>0.102158</td>\n",
       "      <td>0.051609</td>\n",
       "      <td>0.187061</td>\n",
       "      <td>0.024644</td>\n",
       "      <td>0.209060</td>\n",
       "      <td>-0.004236</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.266616</td>\n",
       "      <td>0.198332</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t5</th>\n",
       "      <td>0.303135</td>\n",
       "      <td>0.271736</td>\n",
       "      <td>0.256510</td>\n",
       "      <td>0.000640</td>\n",
       "      <td>0.324164</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.002396</td>\n",
       "      <td>-0.042693</td>\n",
       "      <td>0.066129</td>\n",
       "      <td>0.032474</td>\n",
       "      <td>0.066918</td>\n",
       "      <td>0.085634</td>\n",
       "      <td>0.044841</td>\n",
       "      <td>0.147620</td>\n",
       "      <td>-0.009796</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.252479</td>\n",
       "      <td>0.143075</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t6</th>\n",
       "      <td>0.004614</td>\n",
       "      <td>0.003016</td>\n",
       "      <td>0.011878</td>\n",
       "      <td>-0.000317</td>\n",
       "      <td>-0.002104</td>\n",
       "      <td>-0.002396</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.001046</td>\n",
       "      <td>-0.000925</td>\n",
       "      <td>-0.000811</td>\n",
       "      <td>-0.000342</td>\n",
       "      <td>-0.000973</td>\n",
       "      <td>-0.000381</td>\n",
       "      <td>-0.000733</td>\n",
       "      <td>-0.000106</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.000997</td>\n",
       "      <td>-0.000662</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t7</th>\n",
       "      <td>0.064800</td>\n",
       "      <td>0.075600</td>\n",
       "      <td>0.036705</td>\n",
       "      <td>-0.010102</td>\n",
       "      <td>0.025576</td>\n",
       "      <td>-0.042693</td>\n",
       "      <td>-0.001046</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.013043</td>\n",
       "      <td>-0.023073</td>\n",
       "      <td>-0.016801</td>\n",
       "      <td>-0.021513</td>\n",
       "      <td>-0.021948</td>\n",
       "      <td>-0.005869</td>\n",
       "      <td>-0.000521</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.007214</td>\n",
       "      <td>0.035977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t8</th>\n",
       "      <td>0.135711</td>\n",
       "      <td>0.117397</td>\n",
       "      <td>0.146902</td>\n",
       "      <td>-0.000279</td>\n",
       "      <td>0.162456</td>\n",
       "      <td>0.066129</td>\n",
       "      <td>-0.000925</td>\n",
       "      <td>-0.013043</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.043614</td>\n",
       "      <td>-0.008907</td>\n",
       "      <td>0.098812</td>\n",
       "      <td>-0.018302</td>\n",
       "      <td>0.096724</td>\n",
       "      <td>0.000894</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.081060</td>\n",
       "      <td>0.054167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t9</th>\n",
       "      <td>0.097607</td>\n",
       "      <td>0.080720</td>\n",
       "      <td>0.101063</td>\n",
       "      <td>0.000178</td>\n",
       "      <td>0.102158</td>\n",
       "      <td>0.032474</td>\n",
       "      <td>-0.000811</td>\n",
       "      <td>-0.023073</td>\n",
       "      <td>0.043614</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.004351</td>\n",
       "      <td>0.243086</td>\n",
       "      <td>0.011356</td>\n",
       "      <td>0.052872</td>\n",
       "      <td>-0.001257</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.052032</td>\n",
       "      <td>0.042026</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t10</th>\n",
       "      <td>0.030550</td>\n",
       "      <td>0.040843</td>\n",
       "      <td>0.030295</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>0.051609</td>\n",
       "      <td>0.066918</td>\n",
       "      <td>-0.000342</td>\n",
       "      <td>-0.016801</td>\n",
       "      <td>-0.008907</td>\n",
       "      <td>-0.004351</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.006318</td>\n",
       "      <td>0.023544</td>\n",
       "      <td>-0.000429</td>\n",
       "      <td>-0.002082</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.034876</td>\n",
       "      <td>0.010756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t11</th>\n",
       "      <td>0.142260</td>\n",
       "      <td>0.120563</td>\n",
       "      <td>0.150761</td>\n",
       "      <td>-0.001889</td>\n",
       "      <td>0.187061</td>\n",
       "      <td>0.085634</td>\n",
       "      <td>-0.000973</td>\n",
       "      <td>-0.021513</td>\n",
       "      <td>0.098812</td>\n",
       "      <td>0.243086</td>\n",
       "      <td>0.006318</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.008024</td>\n",
       "      <td>0.097962</td>\n",
       "      <td>0.003425</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.088138</td>\n",
       "      <td>0.065295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t12</th>\n",
       "      <td>0.020644</td>\n",
       "      <td>0.031362</td>\n",
       "      <td>0.006901</td>\n",
       "      <td>-0.006905</td>\n",
       "      <td>0.024644</td>\n",
       "      <td>0.044841</td>\n",
       "      <td>-0.000381</td>\n",
       "      <td>-0.021948</td>\n",
       "      <td>-0.018302</td>\n",
       "      <td>0.011356</td>\n",
       "      <td>0.023544</td>\n",
       "      <td>-0.008024</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.009136</td>\n",
       "      <td>-0.002317</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.007481</td>\n",
       "      <td>-0.005626</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t13</th>\n",
       "      <td>0.139987</td>\n",
       "      <td>0.104162</td>\n",
       "      <td>0.182404</td>\n",
       "      <td>0.001694</td>\n",
       "      <td>0.209060</td>\n",
       "      <td>0.147620</td>\n",
       "      <td>-0.000733</td>\n",
       "      <td>-0.005869</td>\n",
       "      <td>0.096724</td>\n",
       "      <td>0.052872</td>\n",
       "      <td>-0.000429</td>\n",
       "      <td>0.097962</td>\n",
       "      <td>-0.009136</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.000421</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.176879</td>\n",
       "      <td>0.167998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t14</th>\n",
       "      <td>-0.013389</td>\n",
       "      <td>0.005305</td>\n",
       "      <td>0.002018</td>\n",
       "      <td>0.007190</td>\n",
       "      <td>-0.004236</td>\n",
       "      <td>-0.009796</td>\n",
       "      <td>-0.000106</td>\n",
       "      <td>-0.000521</td>\n",
       "      <td>0.000894</td>\n",
       "      <td>-0.001257</td>\n",
       "      <td>-0.002082</td>\n",
       "      <td>0.003425</td>\n",
       "      <td>-0.002317</td>\n",
       "      <td>-0.000421</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000028</td>\n",
       "      <td>0.000427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t15</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t16</th>\n",
       "      <td>0.182701</td>\n",
       "      <td>0.132429</td>\n",
       "      <td>0.237500</td>\n",
       "      <td>0.000418</td>\n",
       "      <td>0.266616</td>\n",
       "      <td>0.252479</td>\n",
       "      <td>-0.000997</td>\n",
       "      <td>0.007214</td>\n",
       "      <td>0.081060</td>\n",
       "      <td>0.052032</td>\n",
       "      <td>0.034876</td>\n",
       "      <td>0.088138</td>\n",
       "      <td>0.007481</td>\n",
       "      <td>0.176879</td>\n",
       "      <td>0.000028</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.209295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cl_t17</th>\n",
       "      <td>0.122248</td>\n",
       "      <td>0.083333</td>\n",
       "      <td>0.172109</td>\n",
       "      <td>0.010531</td>\n",
       "      <td>0.198332</td>\n",
       "      <td>0.143075</td>\n",
       "      <td>-0.000662</td>\n",
       "      <td>0.035977</td>\n",
       "      <td>0.054167</td>\n",
       "      <td>0.042026</td>\n",
       "      <td>0.010756</td>\n",
       "      <td>0.065295</td>\n",
       "      <td>-0.005626</td>\n",
       "      <td>0.167998</td>\n",
       "      <td>0.000427</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.209295</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           cl_t0     cl_t1     cl_t2     cl_t3     cl_t4     cl_t5     cl_t6  \\\n",
       "cl_t0   1.000000  0.385155  0.286471 -0.011409  0.359006  0.303135  0.004614   \n",
       "cl_t1   0.385155  1.000000  0.207322  0.002877  0.283651  0.271736  0.003016   \n",
       "cl_t2   0.286471  0.207322  1.000000 -0.005973  0.375981  0.256510  0.011878   \n",
       "cl_t3  -0.011409  0.002877 -0.005973  1.000000 -0.001758  0.000640 -0.000317   \n",
       "cl_t4   0.359006  0.283651  0.375981 -0.001758  1.000000  0.324164 -0.002104   \n",
       "cl_t5   0.303135  0.271736  0.256510  0.000640  0.324164  1.000000 -0.002396   \n",
       "cl_t6   0.004614  0.003016  0.011878 -0.000317 -0.002104 -0.002396  1.000000   \n",
       "cl_t7   0.064800  0.075600  0.036705 -0.010102  0.025576 -0.042693 -0.001046   \n",
       "cl_t8   0.135711  0.117397  0.146902 -0.000279  0.162456  0.066129 -0.000925   \n",
       "cl_t9   0.097607  0.080720  0.101063  0.000178  0.102158  0.032474 -0.000811   \n",
       "cl_t10  0.030550  0.040843  0.030295 -0.000495  0.051609  0.066918 -0.000342   \n",
       "cl_t11  0.142260  0.120563  0.150761 -0.001889  0.187061  0.085634 -0.000973   \n",
       "cl_t12  0.020644  0.031362  0.006901 -0.006905  0.024644  0.044841 -0.000381   \n",
       "cl_t13  0.139987  0.104162  0.182404  0.001694  0.209060  0.147620 -0.000733   \n",
       "cl_t14 -0.013389  0.005305  0.002018  0.007190 -0.004236 -0.009796 -0.000106   \n",
       "cl_t15       NaN       NaN       NaN       NaN       NaN       NaN       NaN   \n",
       "cl_t16  0.182701  0.132429  0.237500  0.000418  0.266616  0.252479 -0.000997   \n",
       "cl_t17  0.122248  0.083333  0.172109  0.010531  0.198332  0.143075 -0.000662   \n",
       "\n",
       "           cl_t7     cl_t8     cl_t9    cl_t10    cl_t11    cl_t12    cl_t13  \\\n",
       "cl_t0   0.064800  0.135711  0.097607  0.030550  0.142260  0.020644  0.139987   \n",
       "cl_t1   0.075600  0.117397  0.080720  0.040843  0.120563  0.031362  0.104162   \n",
       "cl_t2   0.036705  0.146902  0.101063  0.030295  0.150761  0.006901  0.182404   \n",
       "cl_t3  -0.010102 -0.000279  0.000178 -0.000495 -0.001889 -0.006905  0.001694   \n",
       "cl_t4   0.025576  0.162456  0.102158  0.051609  0.187061  0.024644  0.209060   \n",
       "cl_t5  -0.042693  0.066129  0.032474  0.066918  0.085634  0.044841  0.147620   \n",
       "cl_t6  -0.001046 -0.000925 -0.000811 -0.000342 -0.000973 -0.000381 -0.000733   \n",
       "cl_t7   1.000000 -0.013043 -0.023073 -0.016801 -0.021513 -0.021948 -0.005869   \n",
       "cl_t8  -0.013043  1.000000  0.043614 -0.008907  0.098812 -0.018302  0.096724   \n",
       "cl_t9  -0.023073  0.043614  1.000000 -0.004351  0.243086  0.011356  0.052872   \n",
       "cl_t10 -0.016801 -0.008907 -0.004351  1.000000  0.006318  0.023544 -0.000429   \n",
       "cl_t11 -0.021513  0.098812  0.243086  0.006318  1.000000 -0.008024  0.097962   \n",
       "cl_t12 -0.021948 -0.018302  0.011356  0.023544 -0.008024  1.000000 -0.009136   \n",
       "cl_t13 -0.005869  0.096724  0.052872 -0.000429  0.097962 -0.009136  1.000000   \n",
       "cl_t14 -0.000521  0.000894 -0.001257 -0.002082  0.003425 -0.002317 -0.000421   \n",
       "cl_t15       NaN       NaN       NaN       NaN       NaN       NaN       NaN   \n",
       "cl_t16  0.007214  0.081060  0.052032  0.034876  0.088138  0.007481  0.176879   \n",
       "cl_t17  0.035977  0.054167  0.042026  0.010756  0.065295 -0.005626  0.167998   \n",
       "\n",
       "          cl_t14  cl_t15    cl_t16    cl_t17  \n",
       "cl_t0  -0.013389     NaN  0.182701  0.122248  \n",
       "cl_t1   0.005305     NaN  0.132429  0.083333  \n",
       "cl_t2   0.002018     NaN  0.237500  0.172109  \n",
       "cl_t3   0.007190     NaN  0.000418  0.010531  \n",
       "cl_t4  -0.004236     NaN  0.266616  0.198332  \n",
       "cl_t5  -0.009796     NaN  0.252479  0.143075  \n",
       "cl_t6  -0.000106     NaN -0.000997 -0.000662  \n",
       "cl_t7  -0.000521     NaN  0.007214  0.035977  \n",
       "cl_t8   0.000894     NaN  0.081060  0.054167  \n",
       "cl_t9  -0.001257     NaN  0.052032  0.042026  \n",
       "cl_t10 -0.002082     NaN  0.034876  0.010756  \n",
       "cl_t11  0.003425     NaN  0.088138  0.065295  \n",
       "cl_t12 -0.002317     NaN  0.007481 -0.005626  \n",
       "cl_t13 -0.000421     NaN  0.176879  0.167998  \n",
       "cl_t14  1.000000     NaN  0.000028  0.000427  \n",
       "cl_t15       NaN     NaN       NaN       NaN  \n",
       "cl_t16  0.000028     NaN  1.000000  0.209295  \n",
       "cl_t17  0.000427     NaN  0.209295  1.000000  "
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_cl1[colClust].corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "mmm = biz_cl1[biz_cl1.columns[24:]].mean(axis = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    708.000000\n",
       "mean       0.011466\n",
       "std        0.154915\n",
       "min        0.000017\n",
       "25%        0.000017\n",
       "50%        0.000087\n",
       "75%        0.000896\n",
       "max        3.972067\n",
       "dtype: float64"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mmm.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "ll = mmm[mmm<0.0005]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "ll.to_csv('d.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "499"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(ll)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "l2 = mmm[mmm>=0.0005]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "l2.to_csv('model_cat.csv', header= True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "biz_cl2 = biz_cl1.drop(ll.index, axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(57173, 233)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_cl2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "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>Unnamed: 0</th>\n",
       "      <th>RestaurantsPriceRange2</th>\n",
       "      <th>stars</th>\n",
       "      <th>review_count</th>\n",
       "      <th>how_many_cl</th>\n",
       "      <th>num_of_categ</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>57173.000000</td>\n",
       "      <td>51111.000000</td>\n",
       "      <td>57173.000000</td>\n",
       "      <td>57173.000000</td>\n",
       "      <td>57173.000000</td>\n",
       "      <td>57173.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>28586.000000</td>\n",
       "      <td>1.676117</td>\n",
       "      <td>3.444930</td>\n",
       "      <td>63.924108</td>\n",
       "      <td>1.990712</td>\n",
       "      <td>3.972067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>16504.567807</td>\n",
       "      <td>0.625056</td>\n",
       "      <td>0.803564</td>\n",
       "      <td>162.644692</td>\n",
       "      <td>1.790387</td>\n",
       "      <td>2.016420</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>14293.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>28586.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>19.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>42879.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>59.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>57172.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>7968.000000</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>37.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Unnamed: 0  RestaurantsPriceRange2         stars  review_count  \\\n",
       "count  57173.000000            51111.000000  57173.000000  57173.000000   \n",
       "mean   28586.000000                1.676117      3.444930     63.924108   \n",
       "std    16504.567807                0.625056      0.803564    162.644692   \n",
       "min        0.000000                1.000000      1.000000      3.000000   \n",
       "25%    14293.000000                1.000000      3.000000      7.000000   \n",
       "50%    28586.000000                2.000000      3.500000     19.000000   \n",
       "75%    42879.000000                2.000000      4.000000     59.000000   \n",
       "max    57172.000000                4.000000      5.000000   7968.000000   \n",
       "\n",
       "        how_many_cl  num_of_categ  \n",
       "count  57173.000000  57173.000000  \n",
       "mean       1.990712      3.972067  \n",
       "std        1.790387      2.016420  \n",
       "min        0.000000      1.000000  \n",
       "25%        1.000000      2.000000  \n",
       "50%        2.000000      4.000000  \n",
       "75%        3.000000      5.000000  \n",
       "max       12.000000     37.000000  "
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_cl2.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "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>Unnamed: 0</th>\n",
       "      <th>business_id</th>\n",
       "      <th>RestaurantsPriceRange2</th>\n",
       "      <th>stars</th>\n",
       "      <th>review_count</th>\n",
       "      <th>cl_t0</th>\n",
       "      <th>cl_t1</th>\n",
       "      <th>cl_t2</th>\n",
       "      <th>cl_t3</th>\n",
       "      <th>cl_t4</th>\n",
       "      <th>...</th>\n",
       "      <th>Vegetarian</th>\n",
       "      <th>Venues &amp; Event Spaces</th>\n",
       "      <th>Vietnamese</th>\n",
       "      <th>Waffles</th>\n",
       "      <th>Whiskey Bars</th>\n",
       "      <th>Wine &amp; Spirits</th>\n",
       "      <th>Wine Bars</th>\n",
       "      <th>Wineries</th>\n",
       "      <th>Wraps</th>\n",
       "      <th>num_of_categ</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>1hBmklxIBhw9nu9NjOTaQA</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>240</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>22</td>\n",
       "      <td>4lCTd6-EzOuKl4YbH9gnxw</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>30</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>42</td>\n",
       "      <td>9dP5WPlF2pfjMOx2UeUFnA</td>\n",
       "      <td>3.0</td>\n",
       "      <td>3.5</td>\n",
       "      <td>69</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>53</td>\n",
       "      <td>DEBqmgxv2yhJ93LqGpKltg</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>111</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>70</td>\n",
       "      <td>HlYzk84INQaNslFT-lyBFw</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>42</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 233 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    Unnamed: 0             business_id  RestaurantsPriceRange2  stars  \\\n",
       "13          13  1hBmklxIBhw9nu9NjOTaQA                     2.0    3.0   \n",
       "22          22  4lCTd6-EzOuKl4YbH9gnxw                     2.0    4.0   \n",
       "42          42  9dP5WPlF2pfjMOx2UeUFnA                     3.0    3.5   \n",
       "53          53  DEBqmgxv2yhJ93LqGpKltg                     1.0    4.0   \n",
       "70          70  HlYzk84INQaNslFT-lyBFw                     2.0    2.0   \n",
       "\n",
       "    review_count  cl_t0  cl_t1  cl_t2  cl_t3  cl_t4      ...       Vegetarian  \\\n",
       "13           240   True   True   True  False   True      ...            False   \n",
       "22            30   True  False  False  False  False      ...            False   \n",
       "42            69   True   True  False  False  False      ...             True   \n",
       "53           111   True   True  False  False  False      ...            False   \n",
       "70            42   True   True   True  False  False      ...            False   \n",
       "\n",
       "    Venues & Event Spaces  Vietnamese  Waffles  Whiskey Bars  Wine & Spirits  \\\n",
       "13                  False       False    False         False           False   \n",
       "22                  False       False    False         False           False   \n",
       "42                  False       False    False         False           False   \n",
       "53                  False       False    False         False           False   \n",
       "70                  False       False    False         False           False   \n",
       "\n",
       "    Wine Bars  Wineries  Wraps  num_of_categ  \n",
       "13      False     False  False             9  \n",
       "22      False     False  False             8  \n",
       "42       True     False  False             9  \n",
       "53      False     False   True             8  \n",
       "70      False     False  False             8  \n",
       "\n",
       "[5 rows x 233 columns]"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \n"
     ]
    }
   ],
   "source": [
    "biz_cl2['rest_isna'] = pd.isnull(biz_cl2['RestaurantsPriceRange2'])\n",
    "biz_cl2['RestaurantsPriceRange2'][pd.isnull(biz_cl2['RestaurantsPriceRange2'])] =2\n",
    "\n",
    "mmmm = biz_cl2[biz_cl2.columns[21:] ].mean(axis = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "rev_norm = biz_cl2['review_count'].map(lambda x: x if x < 250 else 250)\n",
    "biz_cl2['st_over_am'] = biz_cl2['stars']/rev_norm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['Unnamed: 0', 'business_id', 'RestaurantsPriceRange2', 'stars',\n",
       "       'review_count', 'cl_t0', 'cl_t1', 'cl_t2', 'cl_t3', 'cl_t4',\n",
       "       ...\n",
       "       'Vietnamese', 'Waffles', 'Whiskey Bars', 'Wine & Spirits', 'Wine Bars',\n",
       "       'Wineries', 'Wraps', 'num_of_categ', 'rest_isna', 'st_over_am'],\n",
       "      dtype='object', length=235)"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_cl2.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['Acai Bowls', 'Active Life', 'Afghan', 'African', 'American (New)',\n",
       "       'American (Traditional)', 'Arabian', 'Arcades', 'Argentine',\n",
       "       'Art Galleries',\n",
       "       ...\n",
       "       'Waffles', 'Whiskey Bars', 'Wine & Spirits', 'Wine Bars', 'Wineries',\n",
       "       'Wraps', 'num_of_categ', 'rest_isna', 'st_over_am', 'stars'],\n",
       "      dtype='object', length=213)"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "colPred = biz_cl2.columns[2:4].union( biz_cl2.columns[24:])\n",
    "colPred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "213"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(colPred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('colPred213', 'wb') as f:\n",
    "    pickle.dump(colPred, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['RestaurantsPriceRange2', 'stars', 'review_count'], dtype='object')"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "biz_cl2.columns[2:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "colClust_n = ['cl_t0', 'cl_t1', 'cl_t2', 'cl_t3', 'cl_t4', 'cl_t5',  'cl_t7',\n",
    "       'cl_t8', 'cl_t9', 'cl_t10', 'cl_t11', 'cl_t12', 'cl_t13', 'cl_t14',\n",
    "        'cl_t16', 'cl_t17']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = biz_cl2[colPred]\n",
    "####\n",
    "Y = biz_cl2[colClust_n] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, Y_train, Y_test = train_test_split(  X, Y, test_size=0.33, random_state=91)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'RestaurantsPriceRange2' in X_train.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "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>cl_t0</th>\n",
       "      <th>cl_t1</th>\n",
       "      <th>cl_t2</th>\n",
       "      <th>cl_t3</th>\n",
       "      <th>cl_t4</th>\n",
       "      <th>cl_t5</th>\n",
       "      <th>cl_t7</th>\n",
       "      <th>cl_t8</th>\n",
       "      <th>cl_t9</th>\n",
       "      <th>cl_t10</th>\n",
       "      <th>cl_t11</th>\n",
       "      <th>cl_t12</th>\n",
       "      <th>cl_t13</th>\n",
       "      <th>cl_t14</th>\n",
       "      <th>cl_t16</th>\n",
       "      <th>cl_t17</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4921</th>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34910</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17982</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       cl_t0  cl_t1  cl_t2  cl_t3  cl_t4  cl_t5  cl_t7  cl_t8  cl_t9  cl_t10  \\\n",
       "4921   False  False  False  False  False  False  False  False  False   False   \n",
       "34910   True   True  False  False  False   True  False  False  False   False   \n",
       "17982   True   True  False  False  False  False  False   True  False   False   \n",
       "\n",
       "       cl_t11  cl_t12  cl_t13  cl_t14  cl_t16  cl_t17  \n",
       "4921    False   False   False   False   False   False  \n",
       "34910   False   False   False   False   False   False  \n",
       "17982   False   False   False   False   False   False  "
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y_train.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "cw = Y_train.mean(axis =0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['cl_t0', 'cl_t1', 'cl_t2', 'cl_t3', 'cl_t4', 'cl_t5', 'cl_t7', 'cl_t8',\n",
       "       'cl_t9', 'cl_t10', 'cl_t11', 'cl_t12', 'cl_t13', 'cl_t14', 'cl_t16',\n",
       "       'cl_t17'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cw.index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'cl_t0': {1: 0.45, 0: 0.55},\n",
       " 'cl_t1': {1: 0.66, 0: 0.33999999999999997},\n",
       " 'cl_t2': {1: 0.11, 0: 0.89},\n",
       " 'cl_t3': {1: 0.1, 0: 0.9},\n",
       " 'cl_t4': {1: 0.2, 0: 0.8},\n",
       " 'cl_t5': {1: 0.25, 0: 0.75},\n",
       " 'cl_t7': {1: 0.1, 0: 0.9},\n",
       " 'cl_t8': {1: 0.1, 0: 0.9},\n",
       " 'cl_t9': {1: 0.1, 0: 0.9},\n",
       " 'cl_t10': {1: 0.1, 0: 0.9},\n",
       " 'cl_t11': {1: 0.1, 0: 0.9},\n",
       " 'cl_t12': {1: 0.1, 0: 0.9},\n",
       " 'cl_t13': {1: 0.1, 0: 0.9},\n",
       " 'cl_t14': {1: 0.1, 0: 0.9},\n",
       " 'cl_t16': {1: 0.1, 0: 0.9},\n",
       " 'cl_t17': {1: 0.1, 0: 0.9}}"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weights = {}\n",
    "for cl in cw.index:\n",
    "    weights[cl] = {}\n",
    "    weights[cl][1] = max(round(cw[cl],2),0.1)\n",
    "    weights[cl][0] = 1 - weights[cl][1]\n",
    "weights\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### looked how rf worked"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf0 = RandomForestClassifier(n_estimators=400,  max_depth=6, \n",
    "                             min_samples_split=2, min_samples_leaf=1, min_weight_fraction_leaf=0.0, max_features=17, \n",
    "                             max_leaf_nodes=None, bootstrap=True, n_jobs=-1, \n",
    "                             random_state=179,   class_weight={0:1, 1:0.9})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 306,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight={0: 1, 1: 0.9},\n",
       "            criterion='gini', max_depth=6, max_features=17,\n",
       "            max_leaf_nodes=None, min_impurity_decrease=0.0,\n",
       "            min_impurity_split=None, min_samples_leaf=1,\n",
       "            min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "            n_estimators=400, n_jobs=-1, oob_score=False, random_state=179,\n",
       "            verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 306,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf0.fit(X_train, Y_train['cl_t0'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 307,
   "metadata": {},
   "outputs": [],
   "source": [
    "probs = rf0.predict_proba(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 27.3 s, sys: 1.29 s, total: 28.6 s\n",
      "Wall time: 7.91 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([-0.64195637, -0.64302025, -0.64143821])"
      ]
     },
     "execution_count": 214,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "cross_val_score(rf0,X_train,Y_train['cl_t0'],scoring= 'neg_log_loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 13.3 s, sys: 1.43 s, total: 14.7 s\n",
      "Wall time: 5.15 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([-0.64231128, -0.64421853, -0.64476238])"
      ]
     },
     "execution_count": 215,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "cross_val_score(rf0,X_test,Y_test['cl_t0'],scoring= 'neg_log_loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a199f2ac8>"
      ]
     },
     "execution_count": 195,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#feat_scores = pd.DataFrame({'Fraction of Samples Affected' : rf0.feature_importances_},\n",
    "                           index=X.columns)\n",
    "#feat_scores = feat_scores.sort_values(by='Fraction of Samples Affected')[:20]\n",
    "#feat_scores.plot(kind='barh')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 308,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8526941057534566"
      ]
     },
     "execution_count": 308,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(Y_train['cl_t0'], probs[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 309,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1a1a1d3320>"
      ]
     },
     "execution_count": 309,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "plot_roc_curve(Y_train['cl_t0'], X_train, rf0, ax,  'train-rf')\n",
    "plot_roc_curve(Y_test['cl_t0'], X_test, rf0, ax,  'test-rf')\n",
    "ax.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 310,
   "metadata": {},
   "outputs": [],
   "source": [
    "gr0 = GradientBoostingClassifier( learning_rate=0.1, n_estimators=300,\n",
    "                                  subsample=0.8,  \n",
    "                                   max_depth=7,  random_state=179, max_features='sqrt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 311,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 22.7 s, sys: 139 ms, total: 22.9 s\n",
      "Wall time: 22.9 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.1, loss='deviance', max_depth=7,\n",
       "              max_features='sqrt', max_leaf_nodes=None,\n",
       "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "              min_samples_leaf=1, min_samples_split=2,\n",
       "              min_weight_fraction_leaf=0.0, n_estimators=300,\n",
       "              presort='auto', random_state=179, subsample=0.8, verbose=0,\n",
       "              warm_start=False)"
      ]
     },
     "execution_count": 311,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "gr0.fit(X_train, Y_train['cl_t0'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 22.9 s, sys: 113 ms, total: 23 s\n",
      "Wall time: 23 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([-0.60153555, -0.60666749, -0.61542372])"
      ]
     },
     "execution_count": 227,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "cross_val_score(gr0,X_test,Y_test['cl_t0'],scoring= 'neg_log_loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 312,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.9096924382208591, 0.8773976794606934)"
      ]
     },
     "execution_count": 312,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(Y_train['cl_t0'], gr0.predict_proba(X_train)[:,1]), roc_auc_score(Y_test['cl_t0'], gr0.predict_proba(X_test)[:,1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1a1a126908>"
      ]
     },
     "execution_count": 313,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "plot_roc_curve(Y_train['cl_t0'], X_train, rf0, ax,  'train-rf')\n",
    "plot_roc_curve(Y_test['cl_t0'], X_test, rf0, ax,  'test-rf')\n",
    "plot_roc_curve(Y_train['cl_t0'], X_train, gr0, ax,  'train-gr')\n",
    "plot_roc_curve(Y_test['cl_t0'], X_test, gr0, ax,  'test-gr')\n",
    "\n",
    "ax.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Gridsearch params for RF and GRBoosting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.4min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.4min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.3min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.4min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.3min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.3min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.3min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.2min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.2min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.3min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.2min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.0min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.3min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 12 candidates, totalling 36 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  36 out of  36 | elapsed:  1.2min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 10min 35s, sys: 8.86 s, total: 10min 44s\n",
      "Wall time: 21min 39s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "param_grid = {'max_depth' : [5,8, 12 ], 'max_features':[5,10,20,25] }\n",
    "clust_model_rf ={}\n",
    "for col in colClust_n:\n",
    "    rf_c_grid =  GridSearchCV(RandomForestClassifier(n_estimators=400,  \n",
    "                             min_samples_split=2, min_samples_leaf=1,   \n",
    "                             max_leaf_nodes=None, bootstrap=True, n_jobs=-1, \n",
    "                             random_state=179,   class_weight=weights[col]),\n",
    "                             param_grid,\n",
    "                             n_jobs=-1,\n",
    "                             verbose=True,\n",
    "                             scoring='neg_log_loss')\n",
    "    rf_c_grid.fit(X_train, Y_train[col])\n",
    "    clust_model_rf[col] = rf_c_grid.best_estimator_\n",
    "    #params = \n",
    "   # print(\"fot cluster \",col, \" best parameters:\", clust_model_rf[col].)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.0min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.0min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.2min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.2min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.0min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:   57.4s finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:   50.7s finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.0min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 6 candidates, totalling 18 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  18 out of  18 | elapsed:  1.1min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5min 6s, sys: 4.29 s, total: 5min 10s\n",
      "Wall time: 22min 8s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "#colClust\n",
    "param_grid = {'max_depth' : [3,6,8 ], 'learning_rate':[0.05, 0.1] }\n",
    "clust_model_gr ={}\n",
    "for col in colClust_n:\n",
    "    gr_c_grid = GridSearchCV(GradientBoostingClassifier( n_estimators=300,\n",
    "                                  subsample=0.8,  \n",
    "                                     random_state=179, max_features='sqrt'),\n",
    "                             param_grid,\n",
    "                             n_jobs=-1,\n",
    "                             verbose=True,\n",
    "                             scoring='neg_log_loss')\n",
    "    gr_c_grid.fit(X_train, Y_train[col])\n",
    "    clust_model_gr[col] = gr_c_grid.best_estimator_\n",
    "    #print(\"fot cluster \",col, \" best parameters:\", gr_c_grid.best_params_)\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Оля ! сохрани модельки "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['cl_t0'])"
      ]
     },
     "execution_count": 271,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "2+2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 16 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(4,4, figsize = (20,10))\n",
    "for col, ax in zip(colClust_n, axs.flatten()):\n",
    "    plot_roc_curve(Y_test[col], X_test, clust_model_rf[col], ax,  'test-rf')\n",
    "    plot_roc_curve(Y_test[col], X_test, clust_model_gr[col], ax,  'test-gr')\n",
    "    \n",
    "    ax.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cl_t0 random forest  0.8704393934956147 \n",
      "\n",
      "cl_t0 gradient booster  0.8791479208713442 \n",
      "\n",
      "cl_t1 random forest  0.8913893100621495 \n",
      "\n",
      "cl_t1 gradient booster  0.8998435383091047 \n",
      "\n",
      "cl_t2 random forest  0.8510646657341068 \n",
      "\n",
      "cl_t2 gradient booster  0.8774568670753413 \n",
      "\n",
      "cl_t3 random forest  0.7946137368711995 \n",
      "\n",
      "cl_t3 gradient booster  0.8033725223090895 \n",
      "\n",
      "cl_t4 random forest  0.8852416417448351 \n",
      "\n",
      "cl_t4 gradient booster  0.9080219236497257 \n",
      "\n",
      "cl_t5 random forest  0.8621606333215235 \n",
      "\n",
      "cl_t5 gradient booster  0.8758760911238163 \n",
      "\n",
      "cl_t7 random forest  0.9613918272294678 \n",
      "\n",
      "cl_t7 gradient booster  0.9701469356328838 \n",
      "\n",
      "cl_t8 random forest  0.912061966244024 \n",
      "\n",
      "cl_t8 gradient booster  0.928685706066782 \n",
      "\n",
      "cl_t9 random forest  0.8889209701753136 \n",
      "\n",
      "cl_t9 gradient booster  0.9110886457533327 \n",
      "\n",
      "cl_t10 random forest  0.9637256862954271 \n",
      "\n",
      "cl_t10 gradient booster  0.9688024500391152 \n",
      "\n",
      "cl_t11 random forest  0.8910665736160617 \n",
      "\n",
      "cl_t11 gradient booster  0.905465020119222 \n",
      "\n",
      "cl_t12 random forest  0.9874842213819617 \n",
      "\n",
      "cl_t12 gradient booster  0.9893743328994906 \n",
      "\n",
      "cl_t13 random forest  0.9060541734010622 \n",
      "\n",
      "cl_t13 gradient booster  0.919265085045794 \n",
      "\n",
      "cl_t14 random forest  0.7365301565418221 \n",
      "\n",
      "cl_t14 gradient booster  0.4993371864892094 \n",
      "\n",
      "cl_t16 random forest  0.8554047463516948 \n",
      "\n",
      "cl_t16 gradient booster  0.8820381516735101 \n",
      "\n",
      "cl_t17 random forest  0.9244858301440795 \n",
      "\n",
      "cl_t17 gradient booster  0.9406534739358334 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "for col in colClust_n:\n",
    "    rf_auc = roc_auc_score(Y_test[col],  clust_model_rf[col].predict_proba(X_test)[:,1])\n",
    "    print(col,'random forest ',rf_auc, '\\n')\n",
    "    gr_auc = roc_auc_score(Y_test[col],  clust_model_gr[col].predict_proba(X_test)[:,1])\n",
    "    print(col,'gradient booster ',gr_auc, '\\n')\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "cl_t0     0.451821\n",
       "cl_t1     0.658139\n",
       "cl_t2     0.111134\n",
       "cl_t3     0.005717\n",
       "cl_t4     0.203159\n",
       "cl_t5     0.249445\n",
       "cl_t7     0.059601\n",
       "cl_t8     0.046809\n",
       "cl_t9     0.037175\n",
       "cl_t10    0.006788\n",
       "cl_t11    0.052343\n",
       "cl_t12    0.007858\n",
       "cl_t13    0.029709\n",
       "cl_t14    0.000731\n",
       "cl_t16    0.053988\n",
       "cl_t17    0.023887\n",
       "dtype: float64"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "colOutliers = ['cl_t7', 'cl_t10', 'cl_t12', 'cl_t14', 'cl_t17']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
       "              learning_rate=0.05, loss='deviance', max_depth=8,\n",
       "              max_features='sqrt', max_leaf_nodes=None,\n",
       "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "              min_samples_leaf=1, min_samples_split=2,\n",
       "              min_weight_fraction_leaf=0.0, n_estimators=300,\n",
       "              presort='auto', random_state=179, subsample=0.8, verbose=0,\n",
       "              warm_start=False)"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clust_model_gr['cl_t4']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf4 = RandomForestClassifier(bootstrap=True, class_weight={1: 0.2, 0: 0.8},\n",
    "            criterion='gini', max_depth=7, max_features=25,\n",
    "            max_leaf_nodes=None, min_impurity_decrease=0.0,\n",
    "            min_impurity_split=None, min_samples_leaf=1,\n",
    "            min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
    "            n_estimators=400, n_jobs=-1, oob_score=False, random_state=179,\n",
    "            verbose=0, warm_start=False)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "gr4 = GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
    "              learning_rate=0.1, loss='deviance', max_depth=5,\n",
    "              max_features='sqrt', max_leaf_nodes=None,\n",
    "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
    "              min_samples_leaf=1, min_samples_split=2,\n",
    "              min_weight_fraction_leaf=0.0, n_estimators=300,\n",
    "              presort='auto', random_state=179, subsample=0.8, verbose=0,\n",
    "              warm_start=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.44293941, -0.44327626, -0.44027319])"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cross_val_score(rf4, X_train,Y_train['cl_t4'],scoring= 'neg_log_loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.30677927, -0.30311866, -0.30227886])"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cross_val_score(gr4, X_train,Y_train['cl_t4'],scoring= 'neg_log_loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight={1: 0.2, 0: 0.8},\n",
       "            criterion='gini', max_depth=7, max_features=25,\n",
       "            max_leaf_nodes=None, min_impurity_decrease=0.0,\n",
       "            min_impurity_split=None, min_samples_leaf=1,\n",
       "            min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "            n_estimators=400, n_jobs=-1, oob_score=False, random_state=179,\n",
       "            verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gr4.fit(X_train,Y_train['cl_t4'])\n",
    "rf4.fit(X_train,Y_train['cl_t4'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "fig, ax = plt.subplots()\n",
    "plot_roc_curve(Y_test['cl_t4'], X_test, rf4, ax,  'test-rf')\n",
    "plot_roc_curve(Y_test['cl_t4'], X_test, gr4, ax,  'test-gr')\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "clust_model_gr['cl_t4'] = gr4\n",
    "clust_model_rf['cl_t4'] = rf4\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "for col in colOutliers:\n",
    "\n",
    "    clust_model_gr[col] = GradientBoostingClassifier(criterion='friedman_mse', init=None,\n",
    "              learning_rate=0.1, loss='deviance', max_depth=5,\n",
    "              max_features='sqrt', max_leaf_nodes=None,\n",
    "              min_impurity_decrease=0.0, min_impurity_split=None,\n",
    "              min_samples_leaf=1, min_samples_split=2,\n",
    "              min_weight_fraction_leaf=0.0, n_estimators=300,\n",
    "              presort='auto', random_state=179, subsample=0.8, verbose=0,\n",
    "              warm_start=False)\n",
    "    clust_model_gr[col].fit(X_train,Y_train[col])\n",
    "    clust_model_rf[col] = RandomForestClassifier(n_estimators=400, max_depth=7, max_features=25, \n",
    "                             min_samples_split=2, min_samples_leaf=1,   \n",
    "                             max_leaf_nodes=None, bootstrap=True, n_jobs=-1, \n",
    "                             random_state=179,   class_weight=weights[col])\n",
    "    clust_model_rf[col].fit(X_train,Y_train[col])\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABIQAAAJOCAYAAADGcdzeAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3XuYXHWd5/H3r6pv6XQn6XQuJOncCZAQAoEACoogIuAFREdExNHdmcFZx5nZHWWFHUVhdnbdcRXWGWV0XUcdHcHBUXEAuSuiKAQMgSRAAoGkc+3cOn2/VP32j2pCLp2kk3T36arzfj1PnnPqnFOnPpXngfrle36XEGNEkiRJkiRJ6ZFJOoAkSZIkSZKGlwUhSZIkSZKklLEgJEmSJEmSlDIWhCRJkiRJklLGgpAkSZIkSVLKWBCSJEmSJElKGQtCkiRJkiRJKWNBSNKgCSF8NITwWNI5JEmSSpltLkmDwYKQpGEXQoghhOP3ev2hEELrXn/a+645I8mckiRJxayfNldFCOHOEMIrfefO3+/6EEL4XyGE7X1//i6EEIY9uKRhYUFIUuJijN+PMda89gf4OPAy8HTC0SRJkkrNY8A1wOZ+zl0LvAc4FVgEvAv42PBFkzScLAhJOiohhOkhhH8LITT1PUH6hwG+79G+3Wf6egN9oJ/LPgJ8N8YYBy2wJElSERrMNleMsTvGeGuM8TEg18/bPgJ8KcbYGGPcAHwJ+OigfBFJI44FIUlHLISQBf4deBWYBUwDbh/Ie2OM5/XtntrXI+iO/e49EzgP+O6gBZYkSSpCQ9nmOoiTgWf2ev1M3zFJJciCkKSjcRYwFbguxtgWY+zse9I0GP4Q+FWMce0g3U+SJKlYDWWbqz81QPNer5uBGucRkkqTBSFJR2M68GqMsXcI7v2HwHeG4L6SJEnFZijbXP1pBcbs9XoM0Oowfqk0WRCSdDTWAzNCCGWDedMQwrkUnoLdOZj3lSRJKlJD0uY6hBUUJpR+zal9xySVIAtCko7GE8Am4AshhNEhhKq+Ys5AbQHm9HP8I8CPYowtgxFSkiSpyA16myuEUBlCqOp7WdF3z9eGhH0X+KsQwrQQwlTgk8C3j+0rSBqpLAhJOmIxxhzwbuB4YB3QCPS3WtjBfB74TghhVwjhSoC+hsmVOFxMkiQJGJo2F/AC0EFhgur7+vZn9p37OvAz4FngOeDuvmOSSlBwOKgkSZIkSVK62ENIkiRJkiQpZYZrcjJJKRJCeDNwb3/nYow1wxxHkiSpJNnmknQsHDImSZIkSZKUMon1EJowYUKcNWtWUh8vSZKG2FNPPbUtxjgx6Rzal20wSZJK20DbYIkVhGbNmsXSpUuT+nhJkjTEQgivJp1BB7INJklSaRtoG8xJpSVJkiRJklLGgpAkSZIkSVLKWBCSJEmSJElKGQtCkiRJkiRJKWNBSJIkSZIkKWUsCEmSJEmSJKXMYQtCIYRvhRC2hhCeO8j5EEL4SghhTQhheQjh9MGPKUmSlC62wSRJ0lAaSA+hbwOXHOL8pcC8vj/XArcdeyxJkqTU+za2wSRJ0hApO9wFMcZHQwizDnHJ5cB3Y4wR+G0IYVwIYUqMcdMgZZSk0tbaBLnupFMcu9bN0NwIhKSTHFY+Rpo7epKOMWCtXTmaWjoJQ/RXe8pbP0hZecXQ3FxHzTaYNHKt2rSbV7e3JR1D0iFUtW1g7K4VScc4qIraiZx8zjsSzXDYgtAATAPW7/W6se/YAY2REMK1FJ5gMWPGjEH4aEmDItcDXS0HHu9shufuhBj7f9+Ol6Fj1+Hv37QKejqOLeMRiEA8WGYgl4/kD3769fvESG8+HrK8kYsDu9fBjMvtOPo366hlgLqkQxyBOmD6EN6/403vtSBUnGyDSQn50Dd/x462EniYI5Ww75X/LYuzI7cgtKp8AZRAQai/fyv1+8+jGOM3gG8ALFmy5Bj+CSWVsFwP5Htff93bBVtWcJD/rF63ZUXh2u5WWPc4VNfve767Dba/tO+9X7Pr1WPLfNyiPbuFYsx+5yvGEOvmEsc29Pv29u4cHd05unpz7O7oYWtLFxVlBx/RurWli0wIB+0t0dbZz3ccIiEEairLKMseZdeNMhiT381T5adz6NJTcWjLjGZDZtqAro0RunrzTB5TyeQxVUOc7EC9+ciCKbXD/rlHa1RFGTPGVw/JvWdVDc19NeRsg6lodPbkuPOpRjp7cklHOWYxwo62bv7wjTP54FkWWKWkZTt2MGbNTwhx338D1D3bTE9+Euvf8b2Ekh1abdXopCMMSkGokX0fXDYAGwfhvtKxG2gxpT+tW2Hnq5DJHvl7t6wo9IgZyHs7dsKOtYX99m3Q1nTkn9eP7uxoOqom7XOss3wcrdX7/mP9lW1t9GZmkMuOYm121gH36Qll/KL8PDqp7Pdz8mSgo/Bvkhhh9dbWQckPMHdi//+TjNWwo72bt54wqd/zAB09Oc6ePb7fc/kIE2srmVY36rAZxo0qZ2b9of9nnc0ceyHn3GO+g6QUsg2movH4y9v5zE/6nR+9aL1xTj3zp4xJOoakx78Hv/lc/+dO/0PmLDx7ePMUkcEoCN0FfCKEcDtwNtDs2HUdlV3roGuvYsKmZyDu9xRp6yrYvga2vQiVYyAE2Pj74c15JMb2P8gjFyP5vcYZ9dROZydjyNc2UFG1m53jTqa7fOye8yHm2FZ3GlAYs15ZluHZDc00d/Ts86Tt5fwUWqimlyzdlMMRDG0/9/h6xlSV93tuysBvw8z6asZVVzB3Ys2A35OPkYm1lTTUjWLcqApmTaimumIw/vckSSXNNphGtHw+8g+PrGF7axfrdxaGjv/Ln5zNKdPGHuadI182E2yraGR54efw0kNJp0jGpmcK2796Hir2e5BbWTy9wZNw2P+LhRB+AJwPTAghNAKfA8oBYoz/CNwDvANYA7QD/2GowmqEy/VC8/p9j3XsLAxT2rkW2nfA6vug5jgOGOvz6q+P/POmLYHq8TD3Qiir3GfY0oHXHuVKvFXjYPycAV26paWDnr16KeYrx0C2gvtXbub+lVtYs7WVHW3dZDOB3P6Tzmzb72YbDhkKgLGjptCRyzF74mg+9IYZNPT1dhk7gB4te8uGQN1o5w6RpJHGNpiK3fqd7Xz5gReprshSUZZhxvhqTphcS+1BHkBJOgaP/l2hMJLWAsiU06Bm0tGN7kixgawy9sHDnI/Anw1aIo1Mvd3QsQNaNsGyf4H1v4PqCa+f79gxwJ46Adq2w5T9ijez3gy7N8J510F53zCemINJJ0PFfnNbVE848Ngg6e7Ns7m5k5WbdvPLF5v4/bqdwCqe39xCbWUZ2YPME7OrfWCrFc2ZMJoLTioMc5pUW8n0veYDyeUj8ybXUFN56P8sQwhMHVtFGKrlhiRJI4JtMCWppbOHv/7xc7R2Hf28fG197/27P1jEuxZNHaxoGulam+De66C7Pekk6bJtDcy5AK65M+kkKiL2c9S+dqyFna8U9p+7E1o2w8Zlhblt+jNtSWEbMlA/DyacAPPfve815VUwaQHUTIZR44Ys+sGs39HOq9sLP0hPvbpzn8l/e3J5Hn2xiYm1laze0srL2w4cY1VXXc7M+mpObRhHXfXBn2jtbO/hvBMmHjDDZwjwlhMmUl/T/xw8kiRJI82LW1q565nClFTHMsTrzFl1JTFETEdg0zJY8ePCvw32H76joTN+NpyU7IpVKj4WhFQY6rV9Ndz33+Clhw88X398YbvgMjjuFBg9Cea/a0gjbdndydOv7mTdjnbKsq+vNvVs4y7y8cARZ929edZua6OrN7/P8bX9FHj6U1WeYeb40YypKuOShcdxxsw6zphZx9yJNfbEkSRJRW3d9nY+8YOn6ege+ApfHX1zFH7nP57FW06YOFTRRrYY4Y5rYNvqAxufOrjX5gS94h+hYUmyWSQdkgWhNNrwFPz7XxWWH2/ZfGDvn7d8utDdEGDKqUM2PAsKS5DevXwTXb15enJ5frJsA79ft+uw75tVf2CmqvIsC6aOIbvXD/Yp08ayrbWLdy6awgmTawnASVPGUL5XL6FMCJRnD77EuSRJUjF7fvNuljc2c+7x9YwdNfD5e86eXc+pDSnu3ZPPwfP/XthfcHmyWYpN5QUwaX7SKSQdhgWhUvfSw4Uumy2boXXL6zOwA1TUwJzzC5M9TzkVZpwNJ76jMEHzEVq3vZ3fvrx9n+FYr9nd0cPyDc2MqSpnR1s3jTvbiUBvLvLshuZ+77eoYSxXLJ7GmbPG7zPPDkBNZdmgLPMtSZJUTLbu7uSdf/8YLZ0Dm7vwNa8tZvHX71jAgqlFvEz6j/8TLL8DssO1GEbfIiBv/Syc96lh+kxJGj4WhIpZZzO0by8UeUIGXvl1YTn2rhbYsLT/98x6c+F9F/9tYf8w3V9bOnt4cUsrtzzwIp09OcqzheXO27t7qSrPEoCefKR7v6FaBzN2VDnHjali0phC0ems2eM5bkwVN7zjJDIhUFmWYVy1K15JkiTtrb27l1WbW2hq6eKSk49jZj+9pQ9lzKhyTphcU3jR3Q7dAxtWP6JsWFpYdOSsPxm+z8xkYdGVw/d5kjSMLAiNdD0dhR/sXDc8/d3Xjze9ACv+7eDvm7q4MInz6Amw6AMw802QOXBYVC4f2dXezbod7azctJtNuzrpzuX5p1+vpao8S0vnvitLLJlZx/wptTS1dHHh/Ml7JlDu6s0zb3LNQceYj64sY4KTKkuSJB2xtq5ezvrbB2nrmwPoQ2+YwZvnHeW8Pt3t8KUToWv3ICYcRie/F97+N0mnkKSSYEFopOpuh/89D7pbD33dmX8MkxdCw5mQKSus8rVX4aezJ0c+RujNA4VePHct28jKTbu586lG2g8yueDYUeVMqq3kkpPHMWVsFQunjeXC+ZMdqiVJkjRAXb05Vm1qIcZ4VO8v62iioqWR3Z29zOt5gTefMIHFM+p4Q8VaaHz16EJ17CwUg06+Amaee3T3SNLctyadQJJKhgWhkSSfgwduhBU/gd2Nrx8/+z9B/dzCJNBnfBTKqsjnI4+ubuKV7e3QBY/c08RzG5qpG71lz9u27O48oIfP/moqy3jb/EmcOn0cU8eNYlHDWI4bUwXg6lqSJEnH4CsPrearj7x01O9/qOKTzM1sAuAnlcC6vj+PDUK4eW+H064ehBtJkoqVBaGktW2DfzgTcj3Q3bLPqWcmX8G/TPhLfr18J2NHlRem+3lyKc9tOHgX3/qaCuZNqgXgxMm1tHX3ctbs8fusvNXRk+O9ixuYMq7K1bUkSZKOwSvb2lh+kEUynlnfTG1VGV/54OKD3yBGxm96lLKelgOOz/ntFjbPuIwtsy6jLBM48bhayvqZAuCIZcthxjnHfh9JUlGzIJSE1q1wxzWFYtCOwlOjmKlgR91pPJs5iT/dcDGdVMKrwKsbqasuZ3trN+fMrQdg8klVNHf0cM7xE3jT8RM4flIN2Uw4omVEJUmSdOw+/aPl/G7tjoOeP2FyDRecOOngN9iyEm7/44OePm7J5Rx3yh8cS0RJkvplQWg4Nb0A9/5XePkXew49m5/Fo/lFfLH3A9D+ei+eirIMf/e+RZwzt55JfUO4JEmSNLL8bu0Olsys4wvvW7TP8epXH6Jy20pqqsrgVwdZ/RVg98bC9p1fglnn7XsuWwZ1swc5sSRJBRaEhssj/xN++YXC/sT5bK2ey1kvFMZtf/ZdC7hldDm5PJw5q44Z46udv0eSJGmEa+7oAaA7l+f4STX7nvyn/wwdB+85tI9sBUw/GyaeMMgJJUk6OAtCQ+2Fe+HeT8OuvpUgTv0guy7+Cmfd/AAA//TRM7ngpEN0I5YkSVIynr8bVt+/52U+wvLGXXT1FlZuzecj/6OsjTMq6+Bn/7rve7ta4PSPwDu+ePjPCZnCvD6SJA0jC0JDpbsdfnwtrPrZnkM7/nQ5p9/6HPyuUAw66bhai0GSJEkj1c+vh9YmqBoDFApAU9u6CQFe68s9rwzG7CiH5v0me66uLyzrXlY5vJklSRogC0JDoelF+OqZr7/+6N38e/NMPnHr8j2H/vt7FnLlkukJhJMkSUqxzt1w1ycK2700tXaxublzz+sMeU7uWsd3x32cB2rfA0BbVy9P79jF339wMe8+deqwxpYkabBZEBps37kM1v6ysF83my1X3cvSLfCJHzwNwH88dzY3vntBggElSZJSbNuLsPKnhf2Gs/Yc3t3cSndHD6MrXm8eP1N+Go9m30BbV++eY+fMrefUhnHDFleSpKFiQWgw/fy/7SkGxff8I1f+diZP3vL0ntPvXTzNYpAkSVISfv89+NWXoLer8PpDd8K8i3jq1Z18+kfL2dDewcz6an7+n/dd6eubCUSVJGk4WBAaDKt+Bnf9xesrSfyn33Ded7awfsdOoLCK2PknTmR2/egEQ0qSJKXYmoegbRuccDGUV0PDEgCebdzFmq2tXLrwON42f3LCISVJGj4WhI5V8wa445rC/ryLWX/qX/KXP9rF+h0dlGcDz3zu7VRX+NcsSZI0rDY/C9+6FHraC69jDqaezgPz/5Y/+5enyf32cQDyMQLwhfcuYmy1K31JktLDSsWx+sc3FbaX/QMvT7+Ct36pMGRs+vhR/OhPz7EYJEmSNFy6WqCzubC/4WnoboHFH4aavp4/c85n5Uu76e7N87Hz5lCeLawMNmVclcUgSVLqWK04Fi89vGeY2E/D+Xz+tt8A8L7TG/jSlacmmUySJCldervhyydDV/O+x8/5c5h44p6Xt3z9bgD+y0UnUFWeHc6EkiSNKBaEjtaKn8C/fgSAe065lb+8o7Ck/NVnz+B/XHFKkskkSZLSoXM3bHmusN/TXigGLXgPHH9h4VjVWJhwAgAxRp7bsJvybGDhtLEWgyRJqWdB6Gjc/xn4zd8X9j/8E/78/3UAkYc/+RZmOnG0JEnS8Pj59bDs+/sem/MWOP0PD7j08Ze3c/X//R0A71o0dTjSSZI0olkQOlIr73q9GPSuW/jvqyaTy6+lfnQFcybWJJtNkiQpTTYug/Fz4F23FF5nyqDhrH0uae7o4ZHnt/JM4y4APvPO+VzzhhnDnVSSpBHHgtCRiBEe/u+F/T97kgebxvLNx5YSAnz7P5x16PdKkiRpcLVvg5CFOecf9JJ/+d06/tfPnwcgE+CShcdRWeZwMUmSLAgdiY2/h20vwFs/yw/WVnHDvy0F4OnPXETd6IqEw0mSJKVIZzO0binMGXQIz2/eDcAvPnU+tVVl1NdUDkc6SZJGPAtCAxUj/N8LAHh5wnl87vsrALjpspMtBkmSJA231Q8UtjWTDnlZR3cOgJn11YQQhjqVJElFw4LQQG0urCLGnAv48we76M7l+emfncup08clm0uSJCmNYr6wPftP6ezJ8YV7n6els/eAy1Zs3M3ciaMtBkmStB8LQgMRI3z9PACWL/prVty+lavOnG4xSJIkaQRYtWk33/7NK0yoqeh3fqA3z5uYQCpJkkY2C0IDseahwrb+eP7PssLun11wfHJ5JEmS0qazGf7tY9BVmBOI1q17Tn3p/hcB+PKVp3HeCRZ/JEkaiEzSAYrC998HQPP7/5WHnt/K2xdMZvr46oRDSZIkpUjTi/DivdCxC0IGao8rTCg9tmHPkvILpo5JOKQkScXDHkKHs+PlPbvfWVGYlPCMmXVJpZEkSUqnh24qbC+6Gea9bc/hba1dtHT28uE3zGSCK4hJkjRg9hA6nA1PF7b/4V6+/EChO/K1581JMJAkSVLKxAhbVxX2p5+5z6mXm9oKh8ePGu5UkiQVNQtCh/OL/wkEGqtPBCAEXKVCkiRpOP3rR6F9G5z/36Bq7D6nenKF1cZOm24PbkmSjoQFoUNpehG2r4FZb+KXL7cCcOsHTks4lCRJUsqs/11he/of7nM4xsjm5k4AKsps1kqSdCT85TyUh28ubBdfwy0PrAbgbfMnJxhIkiQpZTY/Cy2bYO6FMGbKPqfueHI9n/zXZwCoqTxwuXlJknRwAyoIhRAuCSG8EEJYE0K4vp/zM0IIj4QQfh9CWB5CeMfgRx1mu9bDqp/BzHPpWXgl21q7eMOc8YyudB5uSZI0PFLZBttfy5bC9tSr9jmcz0ceWFk494/XnMHciTXDnUySpKJ22IJQCCELfBW4FFgAfDCEsGC/yz4D/DDGuBi4CvjaYAcdVjHCrQsL+ye9i9+8tB2Ai08+LsFQkiQpTVLZBuvP1pWF7bgZ+xx+/OXtPPT8VgDeNn+SczxKknSEBtJD6CxgTYzx5RhjN3A7cPl+10RgTN/+WGDj4EVMwO1Xv75/9sf47E+eAxwuJkmShlX62mD9ae3rIVQ3a5/D/7688FVv+9DplGWdBUGSpCM1kF/PacD6vV439h3b2+eBa0IIjcA9wJ/3d6MQwrUhhKUhhKVNTU1HEXcY7N4IL9xT2P/rLfxi9XbW7WgHYPr46gSDSZKklElXG+xgNhXmCKLi9SFhO9u6+cEThb+aU6ePSyKVJElFbyAFof7638b9Xn8Q+HaMsQF4B/DPIYQD7h1j/EaMcUmMccnEiROPPO1wePCmwvaqH0B5FTf9rNBN+ft/fHaCoSRJUgqlqw22v3wOfv0VaG6ECSdA5esFoe8+/ioA//WSE5k6blRSCSVJKmoDKQg1AtP3et3Agd2R/wj4IUCM8XGgCpgwGAGHVa4Hlt9e2D/hYjp7cqzd1sYZM+s49/ji+zqSJKmopacN1p+mF+CBz8LOtdBw1p7DbV293PLgiwCc1mDvIEmSjtZAlsx6EpgXQpgNbKAwYeHV+12zDrgQ+HYIYT6FxkiR9UcGnvu3wnbJH0Emy33LNwDw/jMaEgwlSZJSKj1tsP488Y3C9sp/hgWX8ft1O/n2b16hJ5cH4DPvnM85PrCTJOmoHbYgFGPsDSF8ArgPyALfijGuCCHcDCyNMd4FfBL4vyGE/0KhK/NHY4z7d2ke+Vo3F7bn/gUAD60qrFzx1vmTkkokSZJSKlVtsP48f3dhO6mwsNpdz2zkp8s2MnvCaOZNqnHuIEmSjtFAeggRY7yHwkSFex+7ca/9lcC5gxstAVtXQc3kPatYPLCysKrFpNqqBENJkqS0Sk0bbH8v/BzatsLiD3PL7/M8uvrXrN/RQW1VGY986vyk00mSVBIGVBBKhc7d8MwPYO6FAPTk8nT05KirLk84mCRJUsqseaCwXfAe7vrpRlo6e5k/pZZTnTNIkqRBY0HoNQ//TWFbMRqAJ1/ZAcDHzz8+qUSSJEmp89CqLcx98h4mUM35d8COtjbefepU/s9Vi5OOJklSSbEg9JotheXleW9hAsPXhotdfPJxSSWSJElKnWcbd1Gfr2JqNsfbT54MwHsXT0s4lSRJpceCEEBvF7z6GJz+ESgfBcA/P/4q5dnAjPrqhMNJkiSlx5THP8dpmZeJs87nf1xxStJxJEkqWZmkA4wIm54pbKcWuiKv39FObz4yq350gqEkSZJSZtd6ToivAhAu+nyyWSRJKnEWhADu+HBhO2URAF99ZA0AN1++MKlEkiRJ6dK4FG5dyOK4klUVp+x5UCdJkoaGQ8YAKmugFZh2Bvl85PYn1wOweIYrWUiSJA2H/NrHyAD/u+f9TD7rQ8xPOpAkSSXOHkL5PGxfA+f8OQDbWrsAeNv8SVSVZ5NMJkmSlA65XjIPfQ6Ae/NnMWmm5SBJkoaaBaF1jxe2ofBX0drVC8C7Fk1NKpEkSVK65HsA+FHuzfzFB97J2xdMTjiQJEmlz4LQ+t8WtgsuB2B3Z6EgVFvlaDpJkqRhsWUlABtjPSceV0sIIeFAkiSVPgtCv/9eYXvcqQCs2rQbgPqayqQSSZIkpcvS/wfAk/kTmVRblXAYSZLSId0Foc7dsONlGD8HsoUeQW19Q8amjrUxIkmSNOTyeVj2fQDWx0mMrnQOR0mShkO6C0KP3VLYvumv9hxaaQ8hSZKk4dHdBnd9AoAv9/wBrzCFyjILQpIkDYd0F4S2vVjYLnzvnkObmzsByGYcuy5JkjSktq6CZd+nZVQDT8YTufAkJ5OWJGm4pHvm5OfvhonzoWL0nkO/eWk79aMrEgwlSZKUEhueAuDzvR/l8fwCnnrfKQkHkiQpPdLbQyhGIEKua8+h3Z2FJU9Pmz4uoVCSJEkp0twIwKNt0/nQ2TOoq/ahnCRJwyW9PYSaXihsz/jonkOLb34AgI9fMDeBQJIkSSnTupWt2Ul0VtTxt1fYO0iSpOGU3h5CfatZcPxFAFz/o+Xk8hGA02fUJZVKkiQpPXatoyU7nvbuXNJJJElKnfQWhDqbC9tJ8/n7h1Zz+5PrAXjqM28jBCeUliRJGlIv/xLW/YZ1HZU01I1KOo0kSamT3oLQ09+BhjMhBL752FoAbvvQ6S43L0mSNBzW/w6A7+Qu5o1z6hMOI0lS+qSzINS8obANWWKMhADnnTCRS0+ZkmwuSZKkNGhtgkf+FoAV+Vm8ca4FIUmShls6C0Ibny5s33Idz25oZld7D+9eZDFIkiRpWHTsAOCfei+msm4Kly60HSZJ0nBLZ0Foy0ogwIw38n8eXA3gkylJkqTh0tsFwG/zCzhjZh0VZelskkqSlKR0/vo2PgF1s6BiNE++UnhC1VBXnWwmSZKkNOjthie+AUAXZbzv9IaEA0mSlE7pKwjFCGsehNop9Oby7O7s5fLTpiadSpIkKR3W/w5+/88AbIwTGF1ZlnAgSZLSKX0FofbthW39XJ5pLCw9v3Dq2AQDSZIkpUjfcLHvnfgPvBinM22cS85LkpSE9BWEtr1Y2C54D0+9WhgutmRWXYKBJEmSUiTmAGiNVQCMqy5PMo0kSamV3oLQhHn829OF5edPOm5MgoEkSZJSYtd6+NGfAPCzZ7cwdlQ5VeXZhENJkpRO6SsI7XwVQgbGTmdTcycAoypsiEiSJA25Vx6Drma2Vs7g1TiZKxZPSzqRJEmplb6C0LrHYVQdHb2R5o4eV7aQJEkaLt2tAHxj7lcZVVvH5y87OeFAkiSlVwoLQr+Fmsnc9UxhuNjUcVUJB5IkSUqJ7rbCJuNE0pIkJS1dBaEYgQiTT2bVphYAPvyGmclmkiRJSosVPwbg2S1d9OTyCYefhAXJAAAgAElEQVSRJCnd0lUQ6i3MGcTkk7n3uU0ATKytTDCQJElSilTWQsgysbaKzp5c0mkkSUq1dBWE+ropU1HD7o5eZtZXE0JINpMkSVJaxAgz3kAuH5k7sSbpNJIkpVq6CkJdhWFi27rL6OjJ8f4znFBakiRp+EQgsH5nO9mMD+UkSUpSugpCfT2EbvnlRgCOn+STKUmSpGETIz0RXtzSSltXb9JpJElKtQEVhEIIl4QQXgghrAkhXH+Qa64MIawMIawIIfzL4MYcJJ3NhW3FaAAuWTglwTCSJEmHVjJtsNfsfIUdbT0AXH7atITDSJKUbmWHuyCEkAW+ClwENAJPhhDuijGu3OuaecANwLkxxp0hhElDFfiYrL4PgEd3jOHy06YmHEaSJOngSqoN1ie2bGJXvvA88qTjahNOI0lSug2kh9BZwJoY48sxxm7gduDy/a75E+CrMcadADHGrYMbc5BsfwmA9XEyWSeTliRJI1vptMH6xEw5v83P55MXncBFCyYnHUeSpFQbSEFoGrB+r9eNfcf2dgJwQgjh1yGE34YQLunvRiGEa0MIS0MIS5uamo4u8bF46WE66k4C4ESfSkmSpJGtdNpgfSLQQRXzJte40qskSQkbSEGov1/ruN/rMmAecD7wQeCbIYRxB7wpxm/EGJfEGJdMnDjxSLMeu552mtrzACxqOCCeJEnSSFI6bTCAx24hk+8hR2DhtLHJZJAkSXsMpCDUCEzf63UDsLGfa34aY+yJMa4FXqDQOBlZQoYXOgttpNNnWhCSJEkjWum0wbpa4MHP01NWw/L8XCrK0rXQrSRJI9FAfo2fBOaFEGaHECqAq4C79rvmJ8AFACGECRS6L788mEGPWYwQ87xEA+845Tgqy7JJJ5IkSTqU0miDAezeBMC3Kq7mvvyZVGZth0mSlLTDFoRijL3AJ4D7gFXAD2OMK0IIN4cQLuu77D5gewhhJfAIcF2McftQhT4q+V4AWnuznDcvoa7SkiRJA1QybTCA3RsA2NFTDkBt1WEXupUkSUNsQL/GMcZ7gHv2O3bjXvsR+Ku+PyNTTwcA3ZRx1uzxCYeRJEk6vJJogwHNL/ySscAz7fVcsXgamYwTSkuSlLT0DOBuK6yoUZWF2RNGJxxGkiQpPXbuagZg7OwzeP+ShoTTSJIkGGAPoZLQ11W5c/R0lzmVJEkaRjGfY3ccxX9552LmTxmTdBxJkkSaegjtKMyvmKublWwOSZKklIl9Q/fLsz6UkyRppEhNQSi+9AgAPbV2U5YkSRpOo9c9TJY8oyrS0zldkqSRLjW/yvmuVrJATd3kpKNIkiSlys4wlt5MlmnjRiUdRZIk9UlND6HMK79iZX4mE2srk44iSZKUKjGXo3XMCUnHkCRJe0lNQSjkumiMEyjPpuYrS5IkJS7GyBw2kMlmk44iSZL2ko7qSE8nAJvieM6ZW59wGEmSpPTo6uqiMvRQ07sz6SiSJGkv6SgIdTYDsDo2MHvC6ITDSJIkpUf72icA2DTx3ISTSJKkvaWqILQ7VhOCy51KkiQNl/Cr/w3A7tFzEk4iSZL2lo5VxrpbAZg0YULCQSRJktJlU3MH6/JzqDz1vUlHkSRJe0lFQahnxyuUA9nK6qSjSJIkpUprWxtQzqwJtsMkSRpJUjFkrGX9CgCOn+typ5IkScNl7aYmzmIFteMmMGXsqKTjSJKkvaSiINS64XkAaqedlHASSZKk9Fi9chkAZeMaEk4iSZL2l4qCUG8+D8CsiTUJJ5EkSUqH36zZxrO/vhuAcWc4f5AkSSNNKgpCFT27eSY/hwk1lUlHkSRJSoVnGpvJd7UBUH/8WQmnkSRJ+0vFpNIV3bvYFWuYU5aK+pckSVLi7nng5/ys/IcAZKpqE04jSZL2l4oKSU/rdnZRQ4UFIUmSpGFxUsW2ws45fwHZ8mTDSJKkA6SiQlKbb2FXHE1lWTbpKJIkSalwOb8o7Cz+cKI5JElS/0q+IJTv7WUMrTBqfNJRJEmSUqMmt5scWZgwL+kokiSpHyVfENr2ynIAJo8fk3ASSZKkdMjnIzPDZjZWzoYQko4jSZL6UfIFoU3rXgSgtuHkhJNIkiSlQ1dPjrrQSufohqSjSJKkgyj5glBm9yYAJk6bnXASSZKkdOjeWngg54IekiSNXCX/K53vbgOgesKsZINIkiSlRM/G5wBYN/WdCSeRJEkHU/IFoboNvyAXAxVjJycdRZIkKRXi6vsA2D1qWsJJJEnSwZR+QahzHW2MYkx1RdJRJEmSUqFr9zZW5WdQO3tJ0lEkSdJBlHxBqDVWsSXWUekYdkmSpGEROnayM9YwZ+LopKNIkqSDKPkqyah8O8vycwkueSpJkjTkNqz4NdNanqGVUUwbNyrpOJIk6SBKuyCUzzOut4nyUTVJJ5EkSUqFnSsfBmDd5At9ICdJ0ghW2gWh3k4A2qLzB0mSJA2Hjhd/AcBF77s22SCSJOmQUlEQonZKsjkkSZJSop0qAKZPnpBwEkmSdCglXRDKrf01AFOqehNOIkmSVPq27O7kTd2/pik7mUzG4WKSJI1kZUkHGEot3TnGAR0zz086iiRJUslraukiMo5M1biko0iSpMMo6R5CLa2tAFSPrk04iSRJUul7Yu0OsuTpmnxa0lEkSdJhlHRBKG5+FoBshUueSpIkDbXuXJ4yctTXjk46iiRJOoySLgiVt28FYMxxcxJOIkmSVPpe2dZGlhxlFa7wKknSSDegglAI4ZIQwgshhDUhhOsPcd0fhBBiCGHJ4EU8evnO3WyPtVTaKJEkSUWo2Npg1eVZxoQOKsrLk4whSZIG4LAFoRBCFvgqcCmwAPhgCGFBP9fVAn8B/G6wQx6tUbvX0harGF1R0nNnS5KkElSMbbDRXZsKO12tyQaRJEmHNZAeQmcBa2KML8cYu4Hbgcv7ue5vgL8DOgcx3zHpLBtDJxVMqLWHkCRJKjpF1wbr3rSqsDPznGSDSJKkwxpIQWgasH6v1419x/YIISwGpscY//1QNwohXBtCWBpCWNrU1HTEYY9UzPWwhfGMKs8O+WdJkiQNsqJrg53c81xhZ/zcIfsMSZI0OAZSEAr9HIt7ToaQAW4BPnm4G8UYvxFjXBJjXDJx4sSBpzxK+d4esmXlhNDfV5AkSRrRiq4N1hsDeTLQcMaQfYYkSRocAykINQLT93rdAGzc63UtsBD4RQjhFeANwF1JT2oIMKVzDZVl9g6SJElFqejaYLl8PPxFkiRpRBhIQehJYF4IYXYIoQK4CrjrtZMxxuYY44QY46wY4yzgt8BlMcalQ5L4COwMY6kus2EiSZKKUlG1wWKMLGz9DRnySXy8JEk6QoctCMUYe4FPAPcBq4AfxhhXhBBuDiFcNtQBj0Um5tg9qiHpGJIkSUesGNtgXbjcvCRJxWJA67HHGO8B7tnv2I0Hufb8Y4917GKM1NPM6owNE0mSVJyKqQ3WuLODSOCVcW9gVpJBJEnSgAxkyFhR6u1qB6CcXMJJJEmSSl/jzg4y5J2/UZKkIlGyBaGe5i0A5GqnJJxEkiSp9G3e3cGpmZcZlXUOIUmSikHJFoR6O3YB0FYzM+EkkiRJpS+bKTQrs6NqEk4iSZIGomQLQt2bXwAgX1adcBJJkqTSV97SCECsdUEPSZKKQckWhHrbtgOwa+z8hJNIkiSVvo7mbQD0TlmccBJJkjQQJVsQKm9aAcC48ZMTTiJJklT61m5tBqCytj7hJJIkaSBKtiBUsfMlAMorqxJOIkmSVPrGZTsBGF1VmXASSZI0EGVJBxgqbaGailhOJoSko0iSJJW8sT1bCzuZkm1eSpJUUkq2h1BN61qei7MYP7oi6SiSJEklb2trb2Fn3Ixkg0iSpAEp2Uc47bGSsbRQWVWyX1GSJGnEqCrre85o72xJkopCyfYQqutYx/NxBpPGOI5dkiRpqGWIhZ1Qss1LSZJKSsn+YveSoZZ2KsuySUeRJEkqeYH8nj1JkjTylWxBKEOe1XFa0jEkSZLSIdpDSJKkYlKyv9iV+Q4y5aOSjiFJkpQSfT2ELAhJklQUSvMXu7sdgPKQP8yFkiRJGgxhzxxCDhmTJKkYlGZBqLMZgF2ZcQkHkSRJSofgkDFJkopKaf5i9xWEOjMOGZMkSRoOwSFjkiQVldL8xc51AdAZqhMOIkmSlA6ZmCvsWBCSJKkolOYvdm+hINSWqUk4iCRJUjrU924u7GTKkg0iSZIGpEQLQp0AbO1IOIckSVJKtIXawk6FD+QkSSoGpVkQ6moFYFJ9XcJBJEmS0iHjHEKSJBWV0vzFbtsKwO7s+ISDSJIkpUOgbw6hTGk2LyVJKjWl+Yvd2gRAW5k9hCRJkoZDiJFciTYtJUkqRaX5q93dQjfl5LOVSSeRJElKhUCefIk2LSVJKkWl+audz5ELZWRCSDqJJElSKmRizoKQJElFpDR/tfO9dOctBkmSJA2XQCTvhNKSJBWN0vzVzvcSQ5befD7pJJIkSakQYo6ID+QkSSoWJVsQ6iXD3Ik1SSeRJElKhRO7VhCISceQJEkDVKIFoRy9MUtFtjS/niRJ0kjTmqmlMnYlHUOSJA1QSVZM4s615AiUWRCSJEkaNqvLT0w6giRJGqCSrJjEUfVMZBetXT1JR5EkSUqFfHQGIUmSiklJFoTy+Rxr4xQa6qqTjiJJkpQKHT058k4hJElS0SjJglDM58iRpSzjcypJkqThUFmWIUYrQpIkFYvSLAjleskRKHcOIUmSpGFTXpZNOoIkSRqgkqyYFHoIZSjL2kNIkiRJkiRpfyVZEOru7iFPhq6efNJRJEmS0sHRYpIkFZUBFYRCCJeEEF4IIawJIVzfz/m/CiGsDCEsDyE8FEKYOfhRBy7GQg+h2RNHJxlDkiTpmBRbG0ySJBWPwxaEQghZ4KvApcAC4IMhhAX7XfZ7YEmMcRFwJ/B3gx30SITeLvJkqHAOIUmSVKSKrQ0W7CIkSVJRGUjF5CxgTYzx5RhjN3A7cPneF8QYH4kxtve9/C3QMLgxj0xl88uU0+sqY5IkqZgVXRsMbHtJklQsBlIQmgas3+t1Y9+xg/kj4N7+ToQQrg0hLA0hLG1qahp4yiPUXVVPjgwh2CiRJElFq6jaYPYPkiSpuAykINRfVaXf3/wQwjXAEuCL/Z2PMX4jxrgkxrhk4sSJA095pPI5NsfxjKsuH7rPkCRJGlrF1waTJElFo2wA1zQC0/d63QBs3P+iEMLbgL8G3hJj7BqceEcp9pIjQ7lzCEmSpOJVVG0w5xCSJKm4DKRi8iQwL4QwO4RQAVwF3LX3BSGExcDXgctijFsHP+aRibkcObKUZx0yJkmSilbRtcGcQ0iSpOJx2IJQjLEX+ARwH7AK+GGMcUUI4eYQwmV9l30RqAH+NYSwLIRw10FuNzzyvfTGLKMrBtIBSpIkaeQpyjaYJEkqGgOqmMQY7wHu2e/YjXvtv22Qcx2TTN+QMecQkiRJxazY2mB2EJIkqXiU5CQ7o3t2UJWNrjImSZIkSZLUj5IcU9UTKqgKuaRjSJIkpUd0UmlJkopJSfYQigR2l9UnHUOSJClVomPGJEkqGiVZEAoxTyaTTTqGJEmSJEnSiFSaBSHyZLIWhCRJkoaT/YMkSSoeJVkQypAna0FIkiRJkiSpX6VXEIqRDJG2Hic2lCRJGj62vSRJKiYlWBDKAzC+pirhIJIkSeliSUiSpOJRsgWhTKYs4SCSJEnpEpxFSJKkolF6BaF8DoCOXp9RSZIkSZIk9af0CkJ9PYTqHDImSZI0bALRZcYkSSoiJVcQyuV6AIeMSZIkDaeIcwhJklRMSq5q0tXeQjUQy+0hJEkaGj09PTQ2NtLZ2Zl0lBGhqqqKhoYGysvLk44iSZJKmG2wfR1rG6zkCkKdXd1UA525kuv8JEkaIRobG6mtrWXWrFmEkO4xMjFGtm/fTmNjI7Nnz046jiRJKmG2wV43GG2wkqua5PuGjNVWj0o4iSSpVHV2dlJfX5/6hghACIH6+nqf1KmP/01IkoaObbDXDUYbrPQKQn2rjIVsNuEkkqRSZkPkdf5dCCA4gZAkaRjY7njdsf5dlF5BKNcLQAgWhCRJkiRJkvpTcgWh2FvoIUS25KZHkiQJgF27dvG1r33tqN5766230t7ePuDrr7vuOk4++WSuu+66o/o8pYldhCRJpa3U2mAlVxDK5Qs9hDIZewhJkkrTcDRGensLv6df//rXefrpp/niF794VJ8nSZJUKkqtDVZy3WheGzKGBSFJUom6/vrreemllzjttNO46KKLmDRpEj/84Q/p6uriiiuu4KabbqKtrY0rr7ySxsZGcrkcn/3sZ9myZQsbN27kggsuYMKECTzyyCP73Pfb3/42d999N52dnbS1tVFTU0NbWxtnn302N9xwAx/4wAcS+sYqBvYPkiSVulJrg5VcQSi2bgWgLPYmnESSlAY3/WwFKzfuHtR7Lpg6hs+9++SDnv/CF77Ac889x7Jly7j//vu58847eeKJJ4gxctlll/Hoo4/S1NTE1KlTufvuuwFobm5m7NixfPnLX+aRRx5hwoQJ/d778ccfZ/ny5YwfPx6Ampoali1bNqjfT6Up4BpjkqThYxvs2JXckLGufKFnUGt2bMJJJEkaevfffz/3338/ixcv5vTTT+f5559n9erVnHLKKTz44IN8+tOf5le/+hVjxw7sd/Giiy7a0xCRjlR05RdJUkqUQhus5HoIEQuTSo+pGZ1wEElSGhzqKdJwiDFyww038LGPfeyAc0899RT33HMPN9xwA29/+9u58cYb9zn/4x//mJtuugmAb37zmwCMHu3vpyRJGvlsgx27kush1JvLA5DJltxXkyQJgNraWlpaWgC4+OKL+da3vkVraysAGzZsYOvWrWzcuJHq6mquueYaPvWpT/H0008f8N4rrriCZcuWsWzZMpYsWZLMl1EJcRYhSVJpK7U2WMn1EMrnCj2EyrJOKi1JKk319fWce+65LFy4kEsvvZSrr76aN77xjUBhvPn3vvc91qxZw3XXXUcmk6G8vJzbbrsNgGuvvZZLL72UKVOmHDChoXSsHDAmSSplpdYGCzEm8zRnyZIlcenSpYN+3+UP38GiR6/lhXf/lBPPOH/Q7y9J0qpVq5g/f37SMUaU/v5OQghPxRjtejTCDFUbbNnNb6SmqoLj/+svB/3ekiSBbbD+HEsbrOTGVe1q6wIghpL7apIkSSNatI+QJElFo+SqJq9NHTS2ujLZIJIkSWniFEKSJBWVkisINTW3A1BZXnLTI0mSJEmSJA2KkisIlfV9o9pRFckGkSRJSpFgFyFJkopKyRWEqru3AVBect9MkiRphAvOISRJUrEoubJJVxhV2KmoSTaIJElSitg/SJKk4lJyBSFib2GbdciYJKk07dq1i6997WtH9d5bb72V9vb2QU4kSZJU+kqtDVZyBaGQ7ysIZZxUWpJUmpJujORyuWN6vyRJUjEqtTaYBSFJkorM9ddfz0svvcRpp53Gddddxxe/+EXOPPNMFi1axOc+9zkA2traeOc738mpp57KwoULueOOO/jKV77Cxo0bueCCC7jgggsOuG97eztXXnklixYt4gMf+ABnn302S5cuBaCmpoYbb7yRs88+m8cff3xYv6+Kg5NKS5JKXam1wUqvapLvq5hlssnmkCSlw73Xw+ZnB/eex50Cl37hoKe/8IUv8Nxzz7Fs2TLuv/9+7rzzTp544glijFx22WU8+uijNDU1MXXqVO6++24AmpubGTt2LF/+8pd55JFHmDBhwgH3/drXvkZdXR3Lly/nueee47TTTttzrq2tjYULF3LzzTcP7ndViXFSaUnSMLENdswG1EMohHBJCOGFEMKaEML1/ZyvDCHc0Xf+dyGEWYMddKDquhoLO/YQkiSlwP3338/999/P4sWLOf3003n++edZvXo1p5xyCg8++CCf/vSn+dWvfsXYsWMPe6/HHnuMq666CoCFCxeyaNGiPeey2Szve9/7hux7qH/F1AaTJClNSqENdtiqSQghC3wVuAhoBJ4MIdwVY1y512V/BOyMMR4fQrgK+F/AB4Yi8OF0harCTsXoJD5ekpQ2h3iKNBxijNxwww187GMfO+DcU089xT333MMNN9zA29/+dm688cZ9zv/4xz/mpptuAuCb3/wmMR58yE9VVRXZrL1vh1OxtcGymUDGZeclScPFNtgxG0gPobOANTHGl2OM3cDtwOX7XXM58J2+/TuBC0NIpkUwq+lhWhgFNkgkSSWqtraWlpYWAC6++GK+9a1v0draCsCGDRvYunUrGzdupLq6mmuuuYZPfepTPP300we894orrmDZsmUsW7aMJUuW8KY3vYkf/vCHAKxcuZJnnx3kbtg6UkXVBluYW8XciT6QkySVrlJrgw1kXNU0YP1erxuBsw92TYyxN4TQDNQD2/a+KIRwLXAtwIwZM44y8qE1NryLjR3bDwgoSVKpqK+v59xzz2XhwoVceumlXH311bzxjW8EChMPfu9732PNmjVcd911ZDIZysvLue222wC49tprufTSS5kyZQqPPPLIPvf9+Mc/zkc+8hEWLVrE4sWLWbRo0YC6OWvIFFUbjLlvhXkXD829JUkaAUqtDRYO1TUJIITwfuDiGOMf973+MHBWjPHP97pmRd81jX2vX+q7ZvvB7rtkyZL42qzZkiQVk1WrVjF//vykYwy6XC5HT08PVVVVvPTSS1x44YW8+OKLVFRUHPa9/f2dhBCeijEuGaq8pc42mCRJ+7INdqBjaYMNpIdQIzB9r9cNwMaDXNMYQigDxgI7BnBvSZI0QrS3t3PBBRfQ09NDjJHbbrttQA0RDRnbYJIkpUBSbbCBFISeBOaFEGYDG4CrgKv3u+Yu4CPA48AfAA/Hw3U9kiRJI0ptbS32HBlRbINJkpQCSbXBDlsQ6huP/gngPiALfCvGuCKEcDOwNMZ4F/D/gH8OIayh8FTqqqEMLUlS0mKMJDR374hj/WFo2AaTJOlAtsFed6xtsIH0ECLGeA9wz37HbtxrvxN4/zElkSSpSFRVVbF9+3bq6+tT3yCJMbJ9+3aqqqqSjlKSbINJkvQ622CvG4w22IAKQpIk6XUNDQ00NjbS1NSUdJQRoaqqioaGhqRjSJKk/8/enYdJWZ15H/+ebhoaaEBWZQcRV0RRFIxiJLgvuMQxakziZCHqxMlkMRPnjSY6m1ecrBMFcYmJxkTNaiKJJhMTNQmrAiIugKI2ICCyQ6913j+ewiCiFNBdT9dT3891tVZ1PV1196Ftjr8659wZ5xzs7fZ2DmYgJEnSbqqqqmLo0KFplyFJklRWnIO1rIq0C5AkSZIkSVJxGQhJkiRJkiSVGQMhSZIkSZKkMhPSahUbQlgNvNJKT98LeKOVnlvv5HgXn2NeXI53cTnexdWa4z04xti7lZ5be8g5WKY43sXleBefY15cjndxpT4HSy0Qak0hhNkxxtFp11EuHO/ic8yLy/EuLse7uBxvtSR/norL8S4ux7v4HPPicryLqy2Mt1vGJEmSJEmSyoyBkCRJkiRJUpnJaiA0Ne0CyozjXXyOeXE53sXleBeX462W5M9TcTnexeV4F59jXlyOd3GlPt6ZPENIkiRJkiRJ7y6rK4QkSZIkSZL0LgyEJEmSJEmSykxJB0IhhNNDCC+EEBaHEL68k8c7hBDuzz8+I4QwpPhVZkcB4/35EMLCEML8EML/hRAGp1FnVuxqvLe77sIQQgwh2CJyLxUy5iGEi/I/58+GEO4rdo1ZUsDvlEEhhMdCCE/nf6+cmUadWRBCuCuEsCqEsOBdHg8hhO/m/yzmhxCOKnaNKi3OwYrLOVhxOQcrLudfxeccrHja/BwsxliSH0AlsATYH2gPzAMO3eGaq4Ap+dsXA/enXXepfhQ43uOBTvnbVzrerTve+eu6AI8D04HRadddyh8F/owPB54Guufv90m77lL9KHC8pwJX5m8fCixNu+5S/QBOBI4CFrzL42cCvwUCMBaYkXbNfrTdD+dgbXK8nYMVcbzz1zkHK9J4O/9KZcydg7XceLfpOVgprxA6FlgcY3wpxtgA/AQ4d4drzgV+kL/9U2BCCCEUscYs2eV4xxgfizFuyd+dDgwoco1ZUsjPN8C/A18H6opZXEYVMuafAm6JMa4FiDGuKnKNWVLIeEega/52N2B5EevLlBjj48Cb73HJucAPY2I6sE8IoW9xqlMJcg5WXM7Biss5WHE5/yo+52BF1NbnYKUcCPUHXtvufm3+czu9JsbYBKwHehaluuwpZLy39wmSpFN7ZpfjHUIYBQyMMf6mmIVlWCE/4wcCB4YQ/hJCmB5COL1o1WVPIeP9NeCyEEItMA24ujillaXd/R2v8uYcrLicgxWXc7Dicv5VfM7B2pZU52DtivVCrWBn7zLFPbhGhSl4LEMIlwGjgfe3akXZ9p7jHUKoAL4FXF6sgspAIT/j7UiWLZ9E8u7rEyGEETHGda1cWxYVMt6XAHfHGL8RQjgOuCc/3rnWL6/s+PeldodzsOJyDlZczsGKy/lX8TkHa1tS/fuylFcI1QIDt7s/gHcuZXvrmhBCO5Llbu+1XEvvrpDxJoRwMvD/gIkxxvoi1ZZFuxrvLsAI4E8hhKUk+00f8lDDvVLo75RfxRgbY4wvAy+QTFC0+woZ708ADwDEGP8GVAO9ilJd+Snod7yU5xysuJyDFZdzsOJy/lV8zsHallTnYKUcCM0ChocQhoYQ2pMcWPjQDtc8BHwsf/tC4I8xf3KTdtsuxzu/fPY2komIe3v3znuOd4xxfYyxV4xxSIxxCMl5ARNjjLPTKTcTCvmd8kuSgzsJIfQiWcL8UlGrzI5CxvtVYAJACOEQksnI6qJWWT4eAj6a73QxFlgfY1yRdlFqs5yDFZdzsOJyDlZczr+KzzlY25LqHKxkt4zFGJtCCJ8BHiE5KZ7QQ8kAACAASURBVP2uGOOzIYQbgdkxxoeAO0mWty0meVfq4vQqLm0FjvfNQA3wYP7cyFdjjBNTK7qEFTjeakEFjvkjwKkhhIVAM3BNjHFNelWXrgLH+wvA7SGEz5Esnb3c/6HcMyGEH5Mste+VPw/gq0AVQIxxCsn5AGcCi4EtwD+mU6lKgXOw4nIOVlzOwYrL+VfxOQcrrrY+Bwv+uUqSJEmSJJWXUt4yJkmSJEmSpD1gICRJkiRJklRmDIQkSZIkSZLKjIGQJEmSJElSmTEQkiRJkiRJKjMGQpIkSZIkSWXGQEiSJEmSJKnMGAhJkiRJkiSVGQMhSZIkSZKkMmMgJEmSJEmSVGYMhCRJkiRJksqMgZAkSZIkSVKZMRCSJEmSJEkqMwZCkiRJkiRJZcZASJIkSZIkqcwYCEmSJEmSJJUZAyFJkiRJkqQyYyAkSZIkSZJUZgyEJEmSJEmSyoyBkKQ9FkK4PITwZNp1SJIklRPnYJJagoGQpFYXQoghhAO2u98+hPDTEMLS/GMn7XD9NSGEBSGEjSGEl0MI1xS9aEmSpBK3u3OwHa57PoRQW7RiJRWdgZCktDwJXAa8vpPHAvBRoDtwOvCZEMLFRaxNkiQpq95rDrbNNcCq4pQjKS0GQpIKEkIYGEL4eQhhdQhhTQjhewV+3eP5m/NCCJtCCB+KMTbEGL8dY3wSaN7xa2KMX48xPhVjbIoxvgD8Cji+5b4bSZKk0lDMOVj+64aSBEb/3TLfgaS2ykBI0i6FECqB3wCvAEOA/sBPCvnaGOOJ+ZtHxBhrYoz37+ZrB2Ac8OzufJ0kSVKpS2kO9r/AvwFbd69aSaXGQEhSIY4F+gHXxBg3xxjr8u8sFcPXSH5Xfb9IrydJktRWFHUOFkI4H2gXY/xFa72GpLajXdoFSCoJA4FXYoxNxXzREMJnSM4SGhdjrC/ma0uSJLUBRZuDhRA6A18Hzmzt15LUNhgISSrEa8CgEEK7YoVCIYSPA18GTowx2uFCkiSVo2LOwYaTbEt7ItmxT3ugWwjhdWBsjHFpK7++pCJzy5ikQswEVgA3hRA6hxCqQwi7c8jzSmD/7T8RQugQQqjO322ff86Qf+zDwH8Bp8QYX2qB+iVJkkpRMedgC0hWJB2Z//hk/uuPJAmmJGWMgZCkXYoxNgPnAAcArwK1wId24ym+BvwghLAuhHBR/nMvkBxW2B94JH97cP6x/wB6ArPyXTE2hRCm7PU3IkmSVEKKOQfLd3d9fdsH8CaQy9/faUcySaUtxBjTrkGSJEmSJElF5AohSZIkSZKkMuOh0pL2WghhHPDbnT0WY6wpcjmSJEllwTmYpL3hljFJkiRJkqQyk9oKoV69esUhQ4ak9fKSJKmVzZkz540YY++069DbOQeTJCnbCp2DpRYIDRkyhNmzZ6f18pIkqZWFEF5Juwa9k3MwSZKyrdA5mIdKS5IkSZIklRkDIUmSJEmSpDJjICRJkiRJklRmbDsvSZJU5hobG6mtraWuri7tUtqE6upqBgwYQFVVVdqlSJLUagyEJEmSylxtbS1dunRhyJAhhBDSLidVMUbWrFlDbW0tQ4cOTbscSZJajVvGJEmSylxdXR09e/Ys+zAIIIRAz549XS0lScq8XQZCIYS7QgirQggL3uXxEEL4bghhcQhhfgjhqJYvU5IkqbwUew5mGPR3joUkqRwUskLobuD093j8DGB4/mMSMHnvy5IkSSp7d+McTJKkbIox7Qp2fYZQjPHxEMKQ97jkXOCHMcYITA8h7BNC6BtjXNFCNUrpePpHMGPv59b1TTmWrdvaFv57l6Sd6v+FJ6juVJN2GdpBOc3B1q1bx3333cdVV12121/77W9/m0mTJtGpU6eCrr/mmmuYNm0aZ555JjfffPNuv54kaff9et5yJv9pSdpltAkHNT3PeXW/olPnzhzzL/enWktLHCrdH3htu/u1+c+9YzISQphE8g4WgwYNaoGXVkmr2wBP/ZDX167n5Tc2p13NOxy08rd0rl/F8n2O3qvn2VjfxMr6arp3rqKywmO7JLU9A/zdVKoyMwdbt24dt9566x4HQpdddtkuA6GmpibatWvHbbfdxurVq+nQocOelitJ2k1PLFrNwhUbOPmQfdMuJRWVsYmjNj/OyRt+zrD659hS0Zl5HT+YrBJKcZtySwRCO6t+p2shYoxTgakAo0ePdr1EWhrrYOPyXV/WHFm1sfUOVOz48u/p8cRX2Q/Yr9VeZe/8rPkEvrDs03v9PF2q2/HEP49nn07tW6AqSZKAVpqD3fDrZ1m4fMPeV7edQ/t15avnHPauj3/5y19myZIlHHnkkZxyyin06dOHBx54gPr6es4//3xuuOEGNm/ezEUXXURtbS3Nzc1cd911rFy5kuXLlzN+/Hh69erFY4899rbnvfvuu3n44Yepq6tj8+bN1NTUsHnzZsaMGcO1117Lhz70oRb9PiVJ765vt2ru+NjotMsori1vwpy7Yebtyf+D9xgGE/6HTkdcwnEd0l+d3RKBUC0wcLv7A4Bdpw1Kz/2XweLf7/KyKpK3GVvbuPpvcfDwA/nepW3vPPKzK9tzdgsktu0qKqis8IBKSVKLyswc7KabbmLBggXMnTuXRx99lJ/+9KfMnDmTGCMTJ07k8ccfZ/Xq1fTr14+HH34YgPXr19OtWze++c1v8thjj9GrV6+dPvff/vY35s+fT48ePQCoqalh7ty5RfveJEllaNXzyfEj8+6Hpq2w/0lwzrfhgFOgDa3MbolA6CHgMyGEnwBjgPWluHc9DfNeW8fXfv0sTc17t1hqROMzXL7lLipirqDrBzW/Rm3lMH7V8bz3vG7lhjp61XTgpIN671V976WhQ3c+12ccxwzpQYfqwvb+S5IkoJXmYO+1kqcYHn30UR599FFGjRoFwKZNm1i0aBHjxo3ji1/8Iv/6r//K2Wefzbhx4wp6vlNOOeWtMEiSpFaTy8GS/4Ppt8KSP0JlBzjiQzDmCtg33b9b380uA6EQwo+Bk4BeIYRa4Kski0eIMU4BpgFnAouBLcA/tlaxpe4PC1eyePWmt+7PevlNnn51HSce2Jt2Bawe6dK8juM2/YEKmt/2+cMbZjKs6SWe7VjYWTfP05Mnak5nQeddTKR6wfjRAzlmRFvd0CVJUnaV6xwsxsi1117Lpz/9zi3bc+bMYdq0aVx77bWceuqpXH/99W97/Be/+AU33HADAHfccQcAnTt3bv2iJUnlq2EzzPsxTJ8CaxZBzX7wga/A0f8InXe+erWtKKTL2CW7eDwC/9RiFWVMjJHl6+vI5SKf+fFT1DW+fRXPvl07cNfHRtOuMr9srH5jss9wZ+b+Hv48ZeeP7X8SR3z0VwXXdUTBV0qSpDSU0xysS5cubNy4EYDTTjuN6667jg9/+MPU1NSwbNkyqqqqaGpqokePHlx22WXU1NRw9913v+1re/Xqxfnnn8/555//1vMuWLAgjW9HklQO1r0Gs25PzgiqWw/9RsEFt8Oh50G70jg7tiW2jOk93PHEy/zntOfeuv9P44fxmfHD37pfVRn+HgYBfPco2LzqvZ/0Sy9Du+q3f27H+5IkSSWiZ8+eHH/88YwYMYIzzjiDSy+9lOOOOw5Izvy59957Wbx4Mddccw0VFRVUVVUxefJkACZNmsQZZ5xB375933GotCRJLSpGqJ2VbAtb+BAQ4ZCJMPZKGDgm1Y5heyIkby4V3+jRo+Ps2bNTee2W9ttnVjDl8Zd2+tjydVtZvbGemy8cSUUIfODgPnSfcTMs/sPOf1iWzYGDzoSDz975i3XrnxxIJUlSGxdCmBNjLLN2Im3fzuZgzz33HIccckhKFbVNjokktZwv/XQeTyx6g79dOyHtUvZMUwMs/FUSBC1/Cjp0g6M/Bsd+CvYZlHZ171DoHMwVQi3gL4/czzFrX2RY753sUe8M+/WtZnxDPjCaR7KsbOtaOODkd14//DQY90UYUNh5QJIkSZIkqRVsXgNzvg+z7oCNK6DnAXDm/8ARl0AbaBu/twyE9lTdBhq3rGPN5ga+tOnrdK3aDOve5dp1wKs7fG7cF2DC9Tu7WpIkSZIkpWXVczB9Msy/H5rqYP/xcM53k0Udbaht/N4yENoT9ZvgGwdR1biFbf23/tr9XN53xS2FP0f70k8TJUmSJEnKhFwuOdpl+q3w0mPJOb0jP5ScD9Qnm1uIDYQKtPaBq1m9aCa5CFU0MaxpCw/FcTzZdDAXHj2Q/cecDx26pF2mJEmSJEkqVP2mpG38jCmwZjF06Zvs5jnqcujcM+3qWpWBUAEWzfsrwxf+kE253qxsP4D27Sp4ut0YntznYxx7zHEce/SAtEuUJEmSJEmFWvcqzJwKT/0w3zb+KPjgnXDouVBZlXZ1RWEgtAsxl6PyV1cA8IOKiXzumq/TuUMybKPSLEySJEmSJBUuRnhtRrIt7LlfAwEOnQhjr4IBx5Rc2/i9lZ3TkFrJX390I/vnXmFJh0O55is3vxUGSZIkqWWsW7eOW2+9dY++9tvf/jZbtmxp4YokSZnS1ADzH4Dbx8Ndp8FLf4L3XQ2fnQf/cDcMPLbswiAwENq5V/4G3z+TzZMncMDi7wPQ8cLJdGhXmXJhkiRJ2ZN2INTc3LxXXy9JaqM2vwGP3wzfPhx+/qnkvKCzvgmffw5OuRH2GZh2hakq7+UuG1bA87+BmHvbpxue+x3tX/kLr3Y8ijW5/jzX9xxOGn5kSkVKkiQV0W+/DK8/07LPud/hcMZN7/rwl7/8ZZYsWcKRRx7JKaecQp8+fXjggQeor6/n/PPP54YbbmDz5s1cdNFF1NbW0tzczHXXXcfKlStZvnw548ePp1evXjz22GNve94tW7Zw+eWX8/zzz3PIIYewdOlSbrnlFkaPHk1NTQ2f//zneeSRR/jGN77BCSec0LLfsyQpPSufzbeNfwCa62HYBDj3Fhj2gUy1jd9b5R0IzZgMf/nOOz7dHlgWe3LG2i+yT6cqpn9qQvFrkyRJKhM33XQTCxYsYO7cuTz66KP89Kc/ZebMmcQYmThxIo8//jirV6+mX79+PPzwwwCsX7+ebt268c1vfpPHHnuMXr16veN5b731Vrp37878+fNZsGABRx759zf4Nm/ezIgRI7jxxhuL9n1KklpRLgeLHk3OB3r5z9CuIxx5KYy5AvocnHZ1bVJ5B0LzHwBgdMNUmnLxbQ81Vnbk6etOoWP7Sqqr3ComSZLKxHus5CmGRx99lEcffZRRo5L2HZs2bWLRokWMGzeOL37xi/zrv/4rZ599NuPGjdvlcz355JN89rOfBWDEiBGMHDnyrccqKyv54Ac/2DrfhCSpeOo3wdz7krbxby6BLv1gwlfh6MuhU4+0q2vTyjYQ+v3Tizll4wreCD14I1fDR8YO5sD9urz1+JCenejeuX2KFUqSJJWfGCPXXnstn/70p9/x2Jw5c5g2bRrXXnstp556Ktdff/3bHv/FL37BDTfcAMAdd9xBjPEdz7FNdXU1lZW+6SdJJWvtK/m28fdA/XroPxouvAsOmVg2beP3VlkGQo2b1vDyL26ECnim42g+MLgPV084gD5dqtMuTZIkqex06dKFjRs3AnDaaadx3XXX8eEPf5iamhqWLVtGVVUVTU1N9OjRg8suu4yamhruvvvut31tr169OP/88zn//PPfet4TTjiBBx54gPHjx7Nw4UKeeaaFz0aSJBVXjPDq9GRb2PO/AQIcdh6MuRIGHpN2dSWnLAOhFb//HpMqfkUzFYy/7P8xvp8HRkuSJKWlZ8+eHH/88YwYMYIzzjiDSy+9lOOOOw6Ampoa7r33XhYvXsw111xDRUUFVVVVTJ48GYBJkyZxxhln0Ldv33ccKn3VVVfxsY99jJEjRzJq1ChGjhxJt27div79SZL2UlMDPPvzJAhaMQ+q94HjPwvHfBK6DUi7upJVfoFQ/UYGzfsmAM9cMpsj+w1LuSBJkiTdd999b7u/7eyfbYYNG8Zpp532jq+7+uqrufrqq3f6nNXV1dx7771UV1ezZMkSJkyYwODBg4HkbCJJUhu3aTXM+T7MugM2rYReB8LZ34KRF0P7TmlXV/LKLxB6/GYAHmkezcnD90+5GEmSJLWWLVu2MH78eBobG4kxMnnyZNq394xISWrzXl+QdAWf/2DSNv6Ak2HsrbC/beNbUvkEQrkc8dmfU/fU/XQEvp27hNMqQtpVSZIkqZV06dKF2bNnp12GpBI2/aU1fOWXC2jOvfsh9dq11Rvr6VK9i/ghl4NFj+Tbxj+etI0fdVnSNr73gcUptMyUTyC0aiHhZ5+gI3Bf0we48IwJaVckSZLUZsQYCcE3y4D37E4mqbzMfW0di1dt4qzD+1LpgoK9csyQ7jt/oH5j0jZ++mRY+zJ07Q8n3wBHfdS28a2sbAKhJbN+yzDgcw1XcsIFV3HJ0QPTLkmSJKlNqK6uZs2aNfTs2bPsQ6EYI2vWrKG62u6zkpKmVgDfuOgIqqsq0y0ma9YuhRlT4el7oH4DDDgWJlwPh5xj2/giKZtAaPHK9QwDDnjfRM46on/ZT3YkSZK2GTBgALW1taxevTrtUtqE6upqBgywa40ktbgY4ZW/JtvCXpgGoQIOPQ/GXgkDRqddXdkpm0Bo05atAHxqwkjam+xKkiS9paqqiqFDh6ZdhiS1ORG3kLaIpnpYkG8b//p86Ngdjv+XfNv4/mlXV7bKJhDaWlcPQPsqO0tIkiRJktTqNq2G2XclbeM3r4LeB8M534HDL7JtfBtQFoHQxrpGDto0AyqAirL4liVJkiRJe2nbGUKeOLKbVsyHGVPgmQehuQGGn5psC9t/vIPZhpRFOvKXxWs4JqygoaKa9hUVaZcjSZIkSVK25Jrhxd8l3cKWPgFVnZJOYWOugF7D065OO1EWgdD0l9ZwMpvJ1dhZTJIkSZK0ewKuanlXdRtg7o+SFUFrl0K3gXDKjUkY1PFdWs2rTSiLQGhrQzPtQo7Ya/+0S5EkSZIkqfS9+TLMnApP3QMNG2HgWDj5Bjj4bKgsi6ih5JXFn1L7dhU0xQra7Xd42qVIkiRJkkpEzB8i5LE3eTHC0ieTbWEvTIOKSjjsAhh7BfQ/Ou3qtJvKIhBqjjH5D7jSDmOSJEmSJO2WxjpY8LMkCFr5DHTsAeO+kLSN79o37eq0h8oiEKpo3EIlOaisSrsUSZIkSVKJKdsFQhtXJm3jZ98Jm1dD70PgnO/CyIugqmPa1WkvlUUgdNi6x5Ib/sBKkiRJkgq0re182VkxD6ZPgQU/zbeNPy3fNv4k989lSFkEQj1WTU9ujLw43UIkSZIkSSUnlEMIkmtOzgWaPgVeeRKqOsPRl8Oxn4ZeB6RdnVpBWQRCjVVdoQnosm/apUiSJEmSSkRZLBCqWw9P3wszboN1r0C3QXDqf8Coj0DHfdKuTq2oLAKhgVsXUh+q6ZB2IZIkSZKkkpPJ9UFrliRt45++Fxo2waDj4NR/h4POsm18mcj8n3LM5TiCRWUS7UqSJEmSWkrmzhCKEZY+kW8b/1uoaAcjLoAxV0D/o9KuTkVWUchFIYTTQwgvhBAWhxC+vJPHB4UQHgshPB1CmB9COLPlS90zDZvWADC374dSrkSSJGn3lPIcTJKypOSPEGqsS1YCTTkBfnAOvDYDTvwi/MszcMFUw6AytcsVQiGESuAW4BSgFpgVQngoxrhwu8u+AjwQY5wcQjgUmAYMaYV6d1vjsvl0AOpq+qddiiRJUsFKfQ4mSVkQS32rycaVScv4WXfCljegz2Ew8Xtw+IV24VZBW8aOBRbHGF8CCCH8BDgX2H4yEoGu+dvdgOUtWeTeaHztKQA2d/VUdEmSVFJKeg4mSVlScl3Gls9NtoUt+BnkmuDA05O28UNPzMByJ7WUQgKh/sBr292vBcbscM3XgEdDCFcDnYGTd/ZEIYRJwCSAQYMG7W6te2TLi3+kO9DYe0RRXk+SJKmFlPQcTJKyoKTOEMo1w/MPJ0HQq39N2saP/jiM+TT0HJZ2dWqDCjlDaGfx4Y7/WVwC3B1jHACcCdwTQnjHc8cYp8YYR8cYR/fu3Xv3q90DjRWdeD1255gRBxfl9SRJklpISc/BJElFUrce/vo9+O6R8MBHYEMtnPqf8PmFcObXDYP0rgpZIVQLDNzu/gDeuRz5E8DpADHGv4UQqoFewKqWKHJv7Lf6SV6NPdm3sqDzsyVJktqKkp6DSVIWtOkFQmuWwIzbYO6Pkrbxg4+H0/4LDjoTKirTrk4loJBAaBYwPIQwFFgGXAxcusM1rwITgLtDCIcA1cDqlix0T+UIdKSBLh0K+VYlSZLajJKeg0lSVrSpI3dihJf/nGwLe/GRpG384RcmbeP7HZl2dSoxu0xJYoxNIYTPAI8AlcBdMcZnQwg3ArNjjA8BXwBuDyF8jiREvTzGtrHbsiLmWLjPSQysaEv/FUuSJL23Up+DSVLaPnLnDBat3LRXz7GxrnGn+3eLrnErPPNgEgStWgidesH7vwSjPwFd9k27OpWogpbNxBinkbQx3f5z1293eyFwfMuW1jIqYxNNwdVBkiSp9JTyHEyS0vbEojc4tG9XDu/fba+eZ/i+NS1U0R7YsCJpGz/7LtiyBvYdAefeAiMuhKrq9OpSJmQ7KcnlqArNxMr2aVciSZIkSSqykw/dl8+fcmDaZey+5U/n28b/PGkbf9AZSdv4IePa2B42lbJMB0J1G9+gGujbsTntUiRJkiRJenfNTfD8b2DGFHj1b9C+Bo75JIyZBD32T7s6ZVCmA6E/P7+C04DNnQelXYokSZIkSe+0dR089UOYORXWvwb7DIbT/htGfRiq9267m/ReMh0IVa17CYARA/ZJuRJJkiRJkrbzxuJkNdDc+6BxMww+AU6/KdkeZtt4FUGmA6EO9WsAyHUflnIlkiRJkqRia3On7cQIL/0pOR9o0SNQ2T45IHrsFdD3iLSrU5nJdCBU0bgFgNhtYMqVSJIkSZLKVuNWmP9AEgStfg4694b3fxlGf9y28UpNpgOhyqbNyY0OKbYJlCRJkiQVVYwx7RISG5bDrDtg9vdh65uw3+Fw3mQY8UFo1yHt6lTmMh0IHb5kKgAVBkKSJEmSpGJZNidZDfTsLyDXDAeflbSNH3y8bePVZmQ6EMqFdqyO3WhX1THtUiRJkiRJRVbU7KW5CZ7/dRIEvTYD2neBYyclHz2GFrEQqTCZDoQ616/iZ82ncG6FCawkSZIkqRVsXZu0jZ8xFTbUQvchSbewIz8M1V3Trk56V9kNhOo2ANA5bKXSQEiSJEmSykZRjhBa/WLSNn7ej6FxCwwZB2feDAeeZtt4lYTsBkJNdQDMzw3jDPMgSZIkSdLeihGW/DHZFrb490nb+MMvStrG73d42tVJuyXzgVAj7aiqrEi5GEmSJElSsQVaaHVAwxaYf3+yImj189C5D5z0bzD6H6GmT8u8hlRk2Q2E1i4FoJIc7dwyJkmSJEnaXeuXJW3j53w/OStov5Fw3hQYcYFt41XyshsIPfdrAJ4Nwwm29ZMkSZKksrHXRwjVzoHpt8LCX0LMJW3jx1wJg99n23hlRnYDoY0rAFjUbljKhUiSJEmS2rzmRnjuIZg+BWpnQoeuMOYKOPZTSecwKWOyGwjV7AtAh3ae7i5JkiRJ5aigxTxb3oSnfgAzb4cNy6D7UDjj63DkpdChS6vXKKUlu4FQrpmNld0NhCRJkiRJ77T6heSQ6Lk/hqatMPREOOsbMPxU28arLGQ3EIrN5KigfTs7jEmSJEmSyLeN/7982/g/QGUHGHlRsjVsvxFpVycVVXYDoVwzzaGS9raclyRJkqSyEuMOx0o3bIZ5P0lWBL3xYnLEyPivJG3jO/dKp0gpZdkOhGIFHaoMhCRJkiSpHHWpfx1+fy/MuRvq1kHfI+D8qXDY+dCufdrlSanKbiAUm2kmUF3l3k9JkiRJKiu1s/jfqu9y1sxZECIcfDaMvQoGjbVtvJSX3UAo10wzFQZCkiRJklQOmhth4a9g+mTaLZvN+ys6Mbf/JRx14Zeg++C0q5PanOwGQstmk4tQ7aHSkiRJbUpDU45z/vdJVqzfmnYpkjKgGxv5IP/HJfyOfXmTpfTlR3ycH9WfwFXDRnKUYZC0U9kNhDr3oXLtS3Rs7wohSZKktmRTfRMvrNzImKE9OKRv17TLkVSiem99mTGrH+CINb+jKtazpMsx/KHPtSzuehwxVHBJRWDiEf3TLlNqs7IbCMVmFjGYtVsa065EkiRJO3Hm4X352PuGpF2GpFKSy+Xbxt8KS/6YtI0f9SEYcwXD9j2MYWnXJ5WQ7AZCuSaa6cABvWvSrkSSJEmStDcaNsO8H8P0KbBmEdTsBx/4Chxt23hpT2U4EGqmiQoqPUJIkiRJkkrTutdg1u35tvHrod8ouOB2OPQ828ZLeynDgVATTbGCClsKSpIkSVLpiBFqZyXbwhY+BEQ4ZCKMvRIGjrFtvNRCshsIbV1HQ+xF8JeFJEmSJLV9TQ35tvG3wvKnoLobHPdPcOynYJ9BaVcnZU5mA6FYt57O1LGhzkOlJUmSJKnN2rwG5nwfZt0BG1dAzwPgzP+BIy6BDp4JK7WWzAZCtOvA2vouDOzeKe1KJEmSJEk7WvUcTJ8M8++HpjrYfzyc81044GSo8DBYqbVlNhAK9Rt4k650al+ZdimSJEmSJEjaxi/+Q7It7KXHoF01HHExjLkC+hySdnVSWclmINSwGYAatkCFZwhJkiRJUqrqNyVt42dMgTWLoUtfmHA9HHU5dO6ZdnVSWcpmIJRrAuDl2JcDDIQkSZIkKR3rXoWZU2HOD6F+PfQ7Cj54Jxx6LlRWpV2dVNYyGgg1A9BMhQuEJEmSJKmYYoTXZiTbwp77NRDg0Ikw9ioYcIxt46U2oqCTukIIp4cQXgghLA4hfPldrrkohLAwhPBsCOG+li1zN+VXCDVRLLg34QAAIABJREFUSTsPI5MkSSWq5OZgkspbUwPMfwBuHw93nQYv/QnedzV8dh78w90w8FjDIKkN2eUKoRBCJXALcApQC8wKITwUY1y43TXDgWuB42OMa0MIfVqr4ILkA6FmKujTtUOqpUiSJO2JkpyDSSpPm99I2sbPvAM2vQ49h8NZ30wOi27fOe3qJL2LQraMHQssjjG+BBBC+AlwLrBwu2s+BdwSY1wLEGNc1dKF7pY3XgSgPU2uEJIkSaWq9OZgksrLymfzbeMfgOZ6GDYBzr0Fhn3AtvFSCSgkEOoPvLbd/VpgzA7XHAgQQvgLUAl8Lcb4ux2fKIQwCZgEMGjQoD2ptzD5FULP5QZxlocISZKk0lR6czBJ2ZfLwaJHk/OBXv4ztOsIR16abxt/cNrVSdoNhQRCO0tU4k6eZzhwEjAAeCKEMCLGuO5tXxTjVGAqwOjRo3d8jpaTywHQQBWDe3ZqtZeRJElqRaU3B5OUXfWbYO59Sdv4N5dAl34w4atw9OXQqUfa1UnaA4UEQrXAwO3uDwCW7+Sa6THGRuDlEMILJJOTWS1S5e6Kf+8yVtMhm43UJElS5pXeHExS9qx9JWkb/9Q9Sdv4/qPhwrvgkIm2jZdKXCFpySxgeAhhKLAMuBi4dIdrfglcAtwdQuhFsnz5pZYsdLfk287nqKBdpVvGJElSSSq9OZikbIgRXp2ebAt7/jdAgMPOgzFXwsBj0q5OUgvZZSAUY2wKIXwGeIRkb/pdMcZnQwg3ArNjjA/lHzs1hLAQaAauiTGuac3C37vov68Q8lBpSZJUikpyDiaptDU1wLM/T4KgFfOgeh84/rNwzCeh24C0q5PUwgraTxVjnAZM2+Fz1293OwKfz3+kL/f3QKjSQ6UlSVKJKrk5mKTStGl10jZ+1h2waSX0OhDO/haMvBjaeyarlFXZPGAnJodKd2zvnlZJkiRJ2qnXF8CMyTD/waRt/AEnw9hbYX/bxkvlIJuBUH6F0P77dk25EEmSJElqQ3I5WPRIvm3841DVCUZdlrSN731g2tVJKqJsBkL5M4QIlenWIUmSJEltQf1GePpHSdv4tS9D1/5w8g1w1EdtGy+VqWwGQvkVQtFASJIkKTWf/cnTTHtmxTs+H2Pyb496lIpg7VKYMRWevgfqN8CAY2HC9XDIObaNl8pcNgOhuC0Qct+rJElSWp5bsYGBPTpx+mH7veOxdpUVnD6ibwpVSWUgRnjlr8m2sBemQaiAQ8+DsVfCgNFpVyepjchmIJRfIUSFK4QkSZLSdNC+XfjS6QenXYZUHprqYcHPYPpkeH0+dOwOJ3wuaRvftV/a1UlqY7IZCOW7jAUDIUmSJElZt2kVzL4LZt0Jm1dB74PhnO/A4RfZNl7Su8pmILRthZBbxiRJkiRl1Yr5ySHRzzwIzQ0w/NRkW9j+4yF4SJek95bNQOitM4Sy+e1JkiRJKlO5Znjxd8m2sKVPJG3jj/po0ja+1/C0q5NUQrKZmOSakn9XuEJIkiRJUgbUbYC529rGL4VuA+GUG5MwqGP3tKuTVIIyGght2zKWzW9PkiRJUpl482WYORWeugcaNsLAsXDyDXDw2VDp/+9I2nPZ/A0S7TImSZIkqUTFCEufTLaFvTAt+f+awy6AsVdA/6PTrk5SRmQzEMolXcZWbmxIuRBJkiRJKlBj3d/bxq98Bjr2gHFfyLeN75t2dZIyJpuBUH6F0MF990m5EEmSJEnahY0rk7bxs++Ezauh9yFwzndh5EVQ1THt6iRlVDYDoVwzzQSCh0pLkiRJaqtWzIPpU2DBT/Nt40/Lt40/ybbxklpdNgOhtS9TSaTCX6KSJEmS2pJcc3Iu0PQp8MqTUNUZjr4cjv009Dog7eoklZFsBkJ5FeZBkiRJktqCuvXw9L0w4zZY9wp0GwSn/geM+gh09KgLScWXzUAoVPBK3JcKEyFJkiRJaVqzJGkb//S90LAJBh0Hp/47HHSWbeMlpSqbv4FyzeSocNutJEmSpOKLEZY+kW8b/1uoaAcjLoAxV0D/o9KuTpKArAZCsZlmKjxDSJIkSVLxNNbBMw/CjCmwcgF06gknXgPHfAK67Jd2dZL0NtkMhHLNNMUKzxCSJEmS1Po2vg6z7kxax295A/ocBhO/B4dfaNt4SW1WJgOhXH7LWF1jLu1SJEmSJGXV8rnJtrAFP4NcExx4etI2fuiJto2X1OZlMhCKa5aQI9C5Qya/PUmSJElpyTXD8w8nQdCrf03axo/+OIz5NPQclnZ1klSwTCYmuc770vuNZ+jRqSrtUiRJkiRlQd16eOoemHkbrHsV9hkEp/4njLrMtvGSSlImA6EYcyyN+9GusiLtUiRJkiSVsjVLYMZtMPdHSdv4wcfDaf8FB50JFZVpVydJeyyTgVBDYzORQFOzZwhJkiRJ2k0xwst/TraFvfhI0jb+8AuTtvH9jky7OklqEZkMhGJsJsZAv3080V+SJElSgRq3Jm3jp0+GVQuhUy94/5dg9Cegy75pVydJLSqjgVAkR6CmOpPfniRJUpsWY6Q5F4kx7UqkAm1YAbO3tY1fA/uOgHNvgREXQlV12tVJUqvIZGISczlyBKo9Q0iSJKnozv7fJ3l2+QYADtyvS8rVSO9h+dP5tvE/T9rGH3RG0jZ+yDjbxkvKvEwGQhu3NhAJdKzykDdJkqRie2n1ZkYP7s77D+zNyYe6zUZtTHMTPP8bmDEFXv0btK+BYz4JYyZBj/3Trk6SiiaTgVAFkUgFh/XrmnYpkiRJZemowd25esLwtMuQ/m7rOnjqhzBzKqx/DfYZDKf9N4z6MFR3S7s6SSq6TAZCjc3NEKoILvOUJEmSytsbi5PVQHPvg8bNMPgEOP2mZHuYbeMllbFMBkIh5miKhkGSJElSWYoRXvpTcj7Qokegsj0c/g9J2/i+I9OuTpLahGwGQkSqPT9IkiRJKi+NW2H+/TB9Cqx+Djr3hpOuhdEfh5o+aVcnSW1KJgMhiBDsMCZJkiSVhQ3LYdYdMPv7sPVN2O9wOG8yjPggtOuQdnWS1CZlMhAKMUf0/CBJkiQp25bNSbaFPfsLyDXDwWclbeMHH2/beEnahUwGQsRIxL8AJElSaQshnA58B6gE7ogx3vQu110IPAgcE2OcXcQSpeJrboLnf50EQa/NgPZd4NhJyUePoWlXJ0klo6BAqNQmI41NTeQqDIQkSVLpCiFUArcApwC1wKwQwkMxxoU7XNcF+GdgRvGrlIpo69qkbfyMqbChFroPSbqFHflhqO6adnWSVHJ2GQiV4mSkQ0UzTaEq7TIkSZL2xrHA4hjjSwAhhJ8A5wILd7ju34GvA18sbnlSkax+MWkbP+/H0LgFhoyDM2+GA0+zbbwk7YVCVgiV3GQkNjfRobOHx0mSpJLWH3htu/u1wJjtLwghjAIGxhh/E0J41zlYCGESMAlg0KBBrVCq1MJihCV/TLaFLf59vm38RTD2iuTAaEnSXiskECq5yUgVTWxqcsuYJEkqaTubzMS3HgyhAvgWcPmunijGOBWYCjB69Oi4i8ul9DRsSdrGz5gCq5+Hzn3gpH+D0f9o23hJamGFBEIlNxmpoomunTq11tNLkiQVQy0wcLv7A4Dl293vAowA/hSSbkr7AQ+FECZ6sLRKzvplSdv4Od9PzgrabyScNwVGXGDbeElqJYUEQiU3GenOBnIVniEkSZJK2ixgeAhhKLAMuBi4dNuDMcb1QK9t90MIfwK+aBikklI7B6bfCgt/CTGXtI0fcyUMfp9t4yWplRUSCJXkZKRz8/o0X16SJGmvxBibQgifAR4h6fR6V4zx2RDCjcDsGOND6VYo7aHmRnjuIZg+BWpnQoeuMOYKOPZTSecwSVJR7DIQKsXJSDMVrG3fN+0yJEmS9kqMcRowbYfPXf8u155UjJqkPbblTXjqBzDzdtiwDLoPhTO+DkdeCh26pF2dJJWdQlYIldxkpJIcwSWmkiRJUvpWv5AcEj33x9C0FYaeCGd9A4afatt4SUpRQYFQSYn5s6pDRbp1SJIkSeUqRljyf/m28X+Ayg4w8qJka9h+I9KuTpJEpgMhVwhJkiRJRdWwGeb9JFkR9MaLULMvjP9K0ja+c69df70kqWiyFwgR8/90hZAkSZJUFOtrk7OB5twNdeug7xFw/lQ47Hxo1z7t6iRJO5G9QCjmADxDSJIkSWptr83Kt43/FRDh4LNh7FUwaKwr9iWpjctgIOSWMUmSJKnVNDcmAdD0ybBsNnToBmOvhGMnQffBaVcnSSpQBgOhZIVQ9FBpSZIkqeVseTPZEjbzdti4HHoMgzP/B464BDrUpF2dJGk3ZTYQCrhCSJIkqdi2NjQT82c6KiNWPQ8zJsO8+5O28fufBOd8Gw44BSp8E1aSSlX2AqFth0q7QkiSJKmonqldz3m3/oXmXKRdhW/OlbRcLt82/lZY8sekbfwRH4IxV8K+h6ZdnSSpBWQuEIq5ZgIeKi1JklRsr2+oozkXmXTi/nz0uCFpl6M90bAZ5v0Ypk+BNYugZj/4wFfgaNvGS1LWZC4QamhupgOwpdGlypIkSWmYeEQ/9utWnXYZ2h3rXoNZ29rGr4d+o+CC2+HQ82wbL0kZlblAKOaSIKhbJ//ikiRJKqYYfUOupMQIr81MtoU992sgwiETk45hA8fYtVeSMi6zgVDwDCFJkiTpnZoa8m3jb4XlT0F1Nzjun+DYT8E+g9KuTpJUJNkLhBq3JDd8R0OSJKmoXB/Uxm1eA3O+D7PugI0roOcBto2XpDKWuUCIrW8CUBmbUi5EkiRJagNWPQfTJ8P8+6GpDvYfD+d8Fw442bbxklTGMhcI5XLNAGzsNDDlSiRJksqTC7XbgFwOFv8h2Rb20mPQrhqOuBjGXAF9Dkm7OklSG5C5QCjmA6GIMxFJkqRi8kzpNqB+U9I2fsYUWLMYuvSFCdfDUZdD555pVydJakMyFwiFXC65UVGZbiGSJElSsax7FWZOhTk/hPr10O8o+OCdcOi5UFmVdnWSpDYoc4FQrrkxuWGXMUmSpFQEV2oXR4zw2ozt2sYHOHQijL0KBhzj3j1J0nvKXCAUGjYBUJWrT7kSSZKkcuOesaJoaoCFv8y3jX8aqveB9/1z0ja+24C0q5MklYjMBUK5imRJbEP7bilXIkmSVJ5cmNJKNr+RtI2feQdseh16DoezvpkcFt2+c9rVSZJKTOYCobjtNEO3jEmSJBWVh0q3kpXP5tvGPwDN9TBsApx7Cwz7gG3jJUl7LIOBUHKotG9MSZIkqWTlcrDo0WRb2Mt/hnYd4chL823jD067OklSBmQuENpS30R3YFN9Lu1SJEmSypJbxvZC/SaYex/MmAxvvgRd+sGEr8LRl0OnHmlXJ0nKkMwFQtsOM+zTtTrlOiRJksqLO8b2wtpXkrbxT92TtI3vPxou/AocMtG28ZKkVpG9QCi/eT1U+NaUJElSGmw7X6AY4dW/JdvCnn8YCHDYeTDmShh4TNrVSZIyLnOBUG5bIORaZUmSJLVFTQ3w7M+TIGjFvKRt/PGfhWM+adt4SVLRZC4Q2tZlzHemJEmSissuY7uwaXXSNn7WHbBpJfQ6EM7+Foy8GNp3Srs6SVKZyW4g5AohSZKkVDgN28HrC5JDouc/mLSNP+BkGHsr7G/beElSegyEJEmS1CKix0r/Xa4ZXnwk2Ra29Amo6gSjLkvaxvc+MO3qJEnKXiD01lrl4LstkiRJKrL6jfD0j2DGFFj7MnTtDyffAEd91LbxkqQ2JXOBUIw5AGwyJkmSlI6ynIatXQozpsLT90D9BhhwLEy4Hg45x7bxkqQ2KYOBUPLv4AohSZKkoiq7Q6VjhFf+mmwLe2FaskL90PNg7JUwYHTa1UmS9J4yGAglK4Q8Q0iSJCkdmZ+GNdXDgp/B9Mnw+nzo2B1O+FzSNr5rv7SrkySpIJkNhOzYIEmSVFyZXyC0aRXMvgtm3QmbV0Hvg+Gc78DhF9k2XpJUcjIXCPFWl7GU65AkSVI2rJifHBL9zIPQ3ADDT022he0/3kmnJKlkZS4QyuXfmqrwDCFJklTiQginA98BKoE7Yow37fD454FPAk3AauDjMcZXil7oO2QgJMk1w4u/S7aFbWsbf9RHk7bxvYanXZ0kSXutoEColCYjdQ2N26pK4+UlSZJaRAihErgFOAWoBWaFEB6KMS7c7rKngdExxi0hhCuBrwMfKn61iZiFU6XrNsDcbW3jl0K3gXDKjUkY1LF72tVJktRidhkIldpkZNuq3aYMzEckSVJZOxZYHGN8CSCE8BPgXOCtOViM8bHtrp8OXFbUCt9FSe6ievNlmDkVnroHGjbCwLFw8g1w8NlQmblF9ZIkFbRCqKQmIyH/zlRNdVVaJUiSJLWE/sBr292vBca8x/WfAH67swdCCJOASQCDBg1qqfpKX4yw9MlkW9gL06CiEg67AMZeAf2PTrs6SZJaVSGBUElNRrYtVa4oybemJEmS3rKzycxO10CHEC4DRgPv39njMcapwFSA0aNHu466se7vbeNXPgMde8C4L+TbxvdNuzpJkoqikECopCYjube6jHmotCRJKmm1wMDt7g8Alu94UQjhZOD/Ae+PMdYXqbb31Gbfltu4MmkbP/tO2Lwaeh8C53wXRl4EVR3Trk6SpKIqJBAqqclI7q0VQmlVIEmS1CJmAcNDCEOBZcDFwKXbXxBCGAXcBpweY1xV/BLfrs2eKb1iHkyfAgt+mm8bf1q+bfxJJXrgkSRJe6+QQKi0JiPbVgiZCEmSpBIWY2wKIXwGeISk0+tdMcZnQwg3ArNjjA8BNwM1wIMhCTZejTFOTK3otiTXnJwLNH0yvPIXqOoMR18Ox34aeh2QdnWSJKVul4FQqU1GcjEHQIVbxiRJUomLMU4Dpu3wueu3u31y0YsqQEhz1U3denj6XphxG6x7BboNglP/A0Z9BDruk15dkiS1MQX10CylyUj0DCFJkqRUxJ0fM1kca5YkbeOfvhcaNsGg4+DUf4eDzrJtvCRJO5G5vx1DcxMAFW4ZkyRJSkXRZmExwtIn8m3jfwsV7WDEBTDmCuh/VLGqkCSpJGUuEGrXsA4A1wdJkiRlVGMdPPMgzJgCKxdAp55w4jVwzCegy35pVydJUknIXCD0el0VRwC075R2KZIkSWWl1buMbXwdZt2ZtI7f8gb0OQwmfg8Ov9C28ZIk7abMBULV+e9on84d0i1EkiSpTLX4mdLL5ybbwhb8DHJNcODpSdv4oSfaNl6SpD2UuUCIXDMAFRWVKRciSZJUXlp0hVCuGZ5/OAmCXv1r0jZ+9MdhzKeh57AWfCFJkspT5gKhmG87HwyEJEmSUhH25ljpuvXw1D0w8zZY9yrsMwhO/U8YdZlt4yVJakGZC4RCPhCqrDQQkiRJKhlrlsCM22Duj5K28YOPh9P+Cw46E3yjT5KkFpe5QGjDlnrALWOSJEnFtts7xmKEl/+cbAt78ZGkbfzhFyZt4/sd2RolSpKkvMwFQp2qkiXKFZU2npckSUrDLs95btyatI2fPhlWLYROveD9X4LRn4Au+xalRkmSyl3mAqGQf28qBFcISZIkFVPc1anSG1bA7G1t49fAviPg3FtgxIVQVV2cIiVJEpDBQGhblzGCK4QkSZLahGVPJauBnv1F0jb+oDOStvFDxtk2XpKklGQuEOra8HpywzOEJEmtpLGxkf/f3v0HWVWeBxz/PuDiBndDdDck4EYhNkzAlVmSrcZR2jCWIhmLw8RoQmmlQ8pMHBmnFkZ2WptI/yiNo7GdCtE2jEkcm1BmbElJxp1p6NBkNAbJFgE1Lgbisi2sRKjsDr+Wt3/cKyI/r7h7zt6z38/MHc7Z+95zn/tw5+6zz33Pe7q6ujh06FDeoQwJtbW1NDU1UVNTk3coytmhY8ff2TneDy/9oNQIev05GFUHv/1luG4RXPbx/IKUJElAARtCR0d8oLQx8uJ8A5EkFVZXVxf19fVMmDCBGOazG1JK7Nu3j66uLiZOnJh3OMrZoz/uBODii0aUrhj2TBt86EqY9Tcw7Q+hdkzOEUqSpLcVriGUCI6kkYwa4SljkqTBcejQIZtBZRFBQ0MDPT09eYeiIeBDo2s40n+csR+shV0/hcuugrt/7sxtSZKGoAJ2TRJggS5JGlw2g95hLvS2iODTV15a2tmzDT7abDNIkqQhqngNoVSaJSRJkqScHOmFN3fC2KvzjkSSJJ1F8RpCnOdyp5IkVbn9+/ezcuXKC3rsI488Ql9fX8Xjly5dytVXX83SpUsv6Pk0TPW8DCT4yJS8I5EkSWdRvIZQSraEJEmFlkVD6NixYwA89thjbN68mQcffPCCnk/D1J7tpX/H2hCSJGmoKtyi0pA8ZUySlJkHfrCN7d3/N6DHnDL+g3z1D85+qs2yZcvYsWMHLS0tzJw5k7Fjx7JmzRoOHz7M3LlzeeCBB+jt7eX222+nq6uL/v5+7r//fvbs2UN3dzczZsygsbGRDRs2vOu4TzzxBOvXr+fQoUP09vZSV1dHb28v1113HW1tbdxxxx0D+jpVYHu3Q81ouNQrz0mSNFQVsCHkGkKSpGJbsWIFW7dupaOjg/b2dtauXcvzzz9PSok5c+awceNGenp6GD9+POvXrwfgwIEDjBkzhocffpgNGzbQ2Nh4xmM/++yzbNmyhcsuuwyAuro6Ojo6MnttKog92+DDnwSv+ipJ0pBVvIZQSl5kTJKUmXPN5MlCe3s77e3tTJs2DYCDBw/y6quvMn36dJYsWcJ9993HLbfcwvTp0ys63syZM080g6QLtnc7TJqVdxSSJOkcCvi1jSsISZKGj5QSbW1tdHR00NHRQWdnJwsXLmTSpEm88MILXHPNNbS1tbF8+fLTHvv000/T0tJCS0sLmzZtAuCSSy7J+iWoYOr734TeHq8wJknSEFe4hlC4hpAkqeDq6+t56623AJg1axarV6/m4MGDAOzevZu9e/fS3d3N6NGjmT9/PkuWLGHz5s2nPXbu3LknGkmtra35vBgVzuVHflXaGDs530AkSdI5Fe6UsZRcQ0iSVGwNDQ3ccMMNNDc3M3v2bObNm8f1118PlNb8efLJJ+ns7GTp0qWMGDGCmpoaVq1aBcCiRYuYPXs248aNO21RaWkgnGgIfcQZQpIkDWWFawh5ypgkaTh46qmn3rV/zz33vGv/qquuYtas09dwWbx4MYsXLz7jMRcsWMCCBQve9bO3Zx5JlWo68hqMboS6sXmHIkmSzqFwp4yRki0hSZKknFx+5FfwkSl5hyFJks6jeA0hEl5mTJIkKXuRjjP+yC4XlJYkqQoUsCHkGkKSJEl5+OjxPVycDjlDSJKkKlDIhpAkSZKyN6F/Z2nDGUKSJA15hWsIRfKy85IkSXmY0L+T4wSM/WTeoUiSpPMoXEMIXFRakiQpDxP6d/HGReNg1CV5hyJJks6jgA0hcFFpSVKR7d+/n5UrV17QYx955BH6+voGOCKpZEL/TnaPmph3GJIkqQLFawgl5wdJkoot74ZQf3//+3q8CuroIS4/3m1DSJKkKnFR3gEMPNcQkiRl6EfL4H9fHNhjfvQamL3irHcvW7aMHTt20NLSwsyZMxk7dixr1qzh8OHDzJ07lwceeIDe3l5uv/12urq66O/v5/7772fPnj10d3czY8YMGhsb2bBhw7uO29fXx4IFC3j55ZeZPHkyO3fu5NFHH6W1tZW6ujruvfdennnmGR566CFuvPHGgX3Nqn5vvMJIjtsQkiSpShS0ISRJUnGtWLGCrVu30tHRQXt7O2vXruX5558npcScOXPYuHEjPT09jB8/nvXr1wNw4MABxowZw8MPP8yGDRtobGw87bgrV67k0ksvZcuWLWzdupWWlpYT9/X29tLc3Mzy5csze52qMnu2A9BlQ0iSpKpQvIZQAtcQkiRl5hwzebLQ3t5Oe3s706ZNA+DgwYO8+uqrTJ8+nSVLlnDfffdxyy23MH369PMe6yc/+Qn33HMPAM3NzUydOvXEfSNHjuTzn//84LwInVVE3Az8HTAS+KeU0opT7r8Y+A7waWAfcEdKaWfWcQKwdxtHqGFvTVMuTy9Jkt6bitYQioibI+KViOiMiGVnuP/iiPh++f6fRcSEgQ60UmOO7mUEx/N6ekmSMpVSoq2tjY6ODjo6Oujs7GThwoVMmjSJF154gWuuuYa2trYzzux5+umnaWlpoaWlhU2bNpHOsQ5fbW0tI0eOHMyXolNExEjgUWA2MAX4UkRMOWXYQuDNlNJvAd8A/jbbKN/R+/oWdo1o4nj4PpEkqRqctyFUbcXIJf0H6OMDeT29JEmDrr6+nrfeeguAWbNmsXr1ag4ePAjA7t272bt3L93d3YwePZr58+ezZMkSNm/efNpj586de6KR1Nrayo033siaNWsA2L59Oy++OMBrI+m9uhboTCm9llI6AnwPuPWUMbcC3y5vrwVuiohcpkr3vv4iW45ezgdqbAhJklQNKjll7EQxAhARbxcj208acyvwtfL2WuAfIiLSub5qHCRHZn2dPUdGMD7rJ5YkKSMNDQ3ccMMNNDc3M3v2bObNm8f1118PQF1dHU8++SSdnZ0sXbqUESNGUFNTw6pVqwBYtGgRs2fPZty4cactKn3XXXdx5513MnXqVKZNm8bUqVMZM2ZM5q9PJ1wOvH7Sfhdw3dnGpJSORcQBoAF44+RBEbEIWARwxRVXDHykKdH92YcYN6KOv5w2eeCPL0mSBlycr2cTEbcBN6eUvlze/yPgupTS3SeN2Voe01Xe31Eec65i5NO7du0ayNciSVImXnrpJSZPLt4fvf39/Rw9epTa2lp27NjBTTfdxC9/+UtGjRp13seeKScR8UJKqXWw4i26iPgCMOuUGuzalNLik8ZsK485uQa7NqW072zHbW1tTZs2bRrc4CVJUm4qrcEqmSF0pmnHp3aRKhlDSukcqL45AAAGvUlEQVRx4HEoFSMVPLckScpIX18fM2bM4OjRo6SUWLVqVUXNIA2aLuBjJ+03Ad1nGdMVERcBY4DfZBOeJEmqZpU0hCxGJEkaBurr63HmyJDyc+ATETER2A18EZh3yph1wJ3As8BtwI/zOGVfkiRVn0quMnaiGImIUZSKkXWnjHm7GAGLEUnSMOCvuXeYi8GRUjoG3A08A7wErEkpbYuI5RExpzzsW0BDRHQC9wKnXQ1WkiTpTM47Q6i8QOHbxchIYPXbxQiwKaW0jlIx8t1yMfIbSk0jSZIKqba2ln379tHQ0EBOF3QaMlJK7Nu3j9ra2rxDKaSU0g+BH57ys786afsQ8IWs45IkSdWvklPGLEYkSTpJU1MTXV1d9PT05B3KkFBbW0tTU1PeYUiSJOk9qKghJEmS3lFTU8PEiRPzDkOSJEm6YJWsISRJkiRJkqQCsSEkSZIkSZI0zNgQkiRJkiRJGmYir0vFRkQPsGuQDt8IvDFIx9bpzHf2zHm2zHe2zHe2BjPfV6aUPjxIx9YFsgYrFPOdLfOdPXOeLfOdrdxrsNwaQoMpIjallFrzjmO4MN/ZM+fZMt/ZMt/ZMt8aSL6fsmW+s2W+s2fOs2W+szUU8u0pY5IkSZIkScOMDSFJkiRJkqRhpqgNocfzDmCYMd/ZM+fZMt/ZMt/ZMt8aSL6fsmW+s2W+s2fOs2W+s5V7vgu5hpAkSZIkSZLOrqgzhCRJkiRJknQWNoQkSZIkSZKGmapuCEXEzRHxSkR0RsSyM9x/cUR8v3z/zyJiQvZRFkcF+b43IrZHxJaI+I+IuDKPOIvifPk+adxtEZEiwktEvk+V5Dwibi+/z7dFxFNZx1gkFXymXBERGyLiF+XPlc/lEWcRRMTqiNgbEVvPcn9ExN+X/y+2RMSnso5R1cUaLFvWYNmyBsuW9Vf2rMGyM+RrsJRSVd6AkcAO4OPAKOC/gSmnjLkL+GZ5+4vA9/OOu1pvFeZ7BjC6vP0V8z24+S6Pqwc2As8BrXnHXc23Ct/jnwB+AVxa3h+bd9zVeqsw348DXylvTwF25h13td6A3wE+BWw9y/2fA34EBPAZ4Gd5x+xt6N6swYZkvq3BMsx3eZw1WEb5tv7KJefWYAOX7yFdg1XzDKFrgc6U0msppSPA94BbTxlzK/Dt8vZa4KaIiAxjLJLz5jultCGl1FfefQ5oyjjGIqnk/Q3w18DXgUNZBldQleT8T4FHU0pvAqSU9mYcY5FUku8EfLC8PQbozjC+QkkpbQR+c44htwLfSSXPAR+KiHHZRKcqZA2WLWuwbFmDZcv6K3vWYBka6jVYNTeELgdeP2m/q/yzM45JKR0DDgANmURXPJXk+2QLKXU6dWHOm++ImAZ8LKX071kGVmCVvMcnAZMi4qcR8VxE3JxZdMVTSb6/BsyPiC7gh8DibEIblt7rZ7yGN2uwbFmDZcsaLFvWX9mzBhtacq3BLsrqiQbBmb5lShcwRpWpOJcRMR9oBX53UCMqtnPmOyJGAN8AFmQV0DBQyXv8IkrTlj9L6dvX/4qI5pTS/kGOrYgqyfeXgCdSSg9FxPXAd8v5Pj744Q07/r7Ue2ENli1rsGxZg2XL+it71mBDS66/L6t5hlAX8LGT9ps4fSrbiTERcRGl6W7nmq6ls6sk30TE7wF/AcxJKR3OKLYiOl++64Fm4D8jYiel803Xuajh+1LpZ8q/pZSOppR+BbxCqUDRe1dJvhcCawBSSs8CtUBjJtENPxV9xktl1mDZsgbLljVYtqy/smcNNrTkWoNVc0Po58AnImJiRIyitGDhulPGrAPuLG/fBvw4lVdu0nt23nyXp88+RqkQ8dze9+ec+U4pHUgpNaaUJqSUJlBaL2BOSmlTPuEWQiWfKf9KaeFOIqKR0hTm1zKNsjgqyfevgZsAImIypWKkJ9Moh491wB+Xr3TxGeBASul/8g5KQ5Y1WLaswbJlDZYt66/sWYMNLbnWYFV7ylhK6VhE3A08Q2ml9NUppW0RsRzYlFJaB3yL0vS2TkrfSn0xv4irW4X5fhCoA/6lvG7kr1NKc3ILuopVmG8NoApz/gzw+xGxHegHlqaU9uUXdfWqMN9/DvxjRPwZpamzC/yD8sJExD9TmmrfWF4P4KtADUBK6ZuU1gf4HNAJ9AF/kk+kqgbWYNmyBsuWNVi2rL+yZw2WraFeg4X/r5IkSZIkScNLNZ8yJkmSJEmSpAtgQ0iSJEmSJGmYsSEkSZIkSZI0zNgQkiRJkiRJGmZsCEmSJEmSJA0zNoQkSZIkSZKGGRtCkiRJkiRJw8z/A92HT0chByAJAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1440x720 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(2,2, figsize = (20,10))\n",
    "for col, ax in zip(colOutliers, axs.flatten()):\n",
    "    plot_roc_curve(Y_test[col], X_test, clust_model_rf[col], ax,  'test-rf')\n",
    "    plot_roc_curve(Y_test[col], X_test, clust_model_gr[col], ax,  'test-gr')\n",
    "    ax.set_title(col)\n",
    "    ax.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "for col in colClust_n:\n",
    "    filename_gr = 'grboost/mgr_'+col+'.pkl'\n",
    "    with open(filename_gr, 'wb') as f:\n",
    "        pickle.dump(clust_model_gr[col], f)\n",
    "    filename_rf = 'rforest/mrf_'+col+'.pkl'\n",
    "    with open(filename_rf, 'wb') as f:\n",
    "        pickle.dump(clust_model_rf[col], f)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## unpickle, looking for threshold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "clust_model_gr = {}\n",
    "clust_model_rf = {}\n",
    "for col in colClust_n:\n",
    "    filename_gr = 'grboost/mgr_'+col+'.pkl'\n",
    "    with open(filename_gr,'rb') as f:\n",
    "        clust_model_gr[col]= pickle.load( f)\n",
    "    filename_rf = 'rforest/mrf_'+col+'.pkl'\n",
    "    with open(filename_rf,'rb') as f:\n",
    "        clust_model_rf[col] =pickle.load( f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calc_thres(y_true, X, model,  thres = 0.7,label = ''):\n",
    "    \"\"\"Plot roc curve, y_true & model.predict_proba(X), ax --- axes of subplot\"\"\"\n",
    "    fpr, tpr, thr = roc_curve(y_true, model.predict_proba(X)[:,1])\n",
    "    idx = np.abs(tpr - thres).argmin()\n",
    "    print(label, ':  ', col,' ', tpr[idx], ' ', fpr[idx], ' ', thr[idx])\n",
    "    return thr[idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "for col in colClust_n:\n",
    "    calc_thres(Y_test[col], X_test, clust_model_rf[col], 0.7,  'test-rf')\n",
    "    calc_thres(Y_test[col], X_test, clust_model_gr[col], 0.7,  'test-gr')\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "th_gr72 ={}\n",
    "th_rf72 = {}\n",
    "for col in colClust_n:\n",
    "    th_rf72[col] = max(calc_thres(Y_test[col], X_test, clust_model_rf[col], 0.72,  'test-rf'), 0.01)\n",
    "    th_gr72[col] = max(calc_thres(Y_test[col], X_test, clust_model_gr[col], 0.72,  'test-gr'),0.01)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('thres_72_gr', 'wb') as f:\n",
    "    pickle.dump(th_gr72,f)\n",
    "with open('thres_72_rf','wb') as f:\n",
    "    pickle.dump(th_rf72,f )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test-rf :   cl_t0   0.3498466619485728   0.021174205967276226   0.7145996009074023\n",
      "test-gr :   cl_t0   0.35043642368483136   0.015591915303176131   0.8779382033821507\n",
      "test-rf :   cl_t1   0.34997177646963956   0.013298283593629194   0.9251774972382272\n",
      "test-gr :   cl_t1   0.34835900330618497   0.0029379928869645894   0.9706098379912487\n",
      "test-rf :   cl_t2   0.3504148365056125   0.028479695582377073   0.09672319260149016\n",
      "test-gr :   cl_t2   0.34992679355783307   0.026695998573042393   0.4776026600525889\n",
      "test-rf :   cl_t3   0.35185185185185186   0.07430703624733476   0.0016454277748129452\n",
      "test-gr :   cl_t3   0.35185185185185186   0.04893390191897654   0.014083656099112751\n",
      "test-rf :   cl_t4   0.350863213811421   0.012845130106601337   0.3525910562359607\n",
      "test-gr :   cl_t4   0.350066401062417   0.010991193802555784   0.7589383689834613\n",
      "test-rf :   cl_t5   0.34980323567993005   0.023226528613404226   0.3448134487670461\n",
      "test-gr :   cl_t5   0.35002186270223   0.021267664754442424   0.7117587094833544\n",
      "test-rf :   cl_t7   0.3481071098799631   0.0051728985099803206   0.10512960889284231\n",
      "test-gr :   cl_t7   0.34995383194829177   0.0047230812482429016   0.7778628551805029\n",
      "test-rf :   cl_t8   0.3501144164759725   0.012615316216516617   0.09037281891030602\n",
      "test-gr :   cl_t8   0.3489702517162471   0.007224630432366344   0.49469297358954595\n",
      "test-rf :   cl_t9   0.349768875192604   0.024260387507547065   0.051410468261200035\n",
      "test-gr :   cl_t9   0.349768875192604   0.013502387617322576   0.3310141033729966\n",
      "test-rf :   cl_t10   0.35   0.0006400682739492212   0.16208938583548552\n",
      "test-gr :   cl_t10   0.35   0.0014401536163857478   0.4986321134433481\n",
      "test-rf :   cl_t11   0.3497854077253219   0.013325156110615521   0.09687667183805175\n",
      "test-gr :   cl_t11   0.35085836909871243   0.008920606601248885   0.4670956493578571\n",
      "test-rf :   cl_t12   0.34911242603550297   0.0014439274827530884   0.06795537384334152\n",
      "test-gr :   cl_t12   0.3609467455621302   0.001283491095780523   0.6950609061737456\n",
      "test-rf :   cl_t13   0.3503521126760563   0.01005464480874317   0.10770567432929509\n",
      "test-gr :   cl_t13   0.34683098591549294   0.007213114754098361   0.38795605801273597\n",
      "test-rf :   cl_t14   0.3333333333333333   0.0034466302561111405   0.00869064632234338\n",
      "test-gr :   cl_t14   0.1111111111111111   0.009491489474521449   1.0\n",
      "test-rf :   cl_t16   0.3492063492063492   0.04232922732362822   0.039693625572722824\n",
      "test-gr :   cl_t16   0.3501984126984127   0.027715565509518477   0.28655708842245764\n",
      "test-rf :   cl_t17   0.3506224066390041   0.018057217448058306   0.024528574099973412\n",
      "test-gr :   cl_t17   0.3506224066390041   0.011095398672903297   0.2985290350036436\n"
     ]
    }
   ],
   "source": [
    "th_gr35 ={}\n",
    "th_rf35 = {}\n",
    "for col in colClust_n:\n",
    "    th_rf35[col] = max(calc_thres(Y_test[col], X_test, clust_model_rf[col], 0.35,  'test-rf'), 0.01)\n",
    "    th_gr35[col] = max(calc_thres(Y_test[col], X_test, clust_model_gr[col], 0.35,  'test-gr'),0.01)\n",
    "with open('thres_35_gr', 'wb') as f:\n",
    "    pickle.dump(th_gr35,f)\n",
    "with open('thres_35_rf','wb') as f:\n",
    "    pickle.dump(th_rf35,f )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}