{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>.container { width:100% !important; }</style>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Full width\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(\"<style>.container { width:100% !important; }</style>\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "import os\n",
    "import subprocess\n",
    "import pickle \n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from IPython.display import display\n",
    "\n",
    "# \n",
    "from lib_modeling import *\n",
    "from lib_feature_engineering import *\n",
    "\n",
    "# some settings for displaying Pandas results\n",
    "pd.set_option('display.width', 2000)\n",
    "pd.set_option('display.max_rows', 500)\n",
    "pd.set_option('display.max_columns', 500)\n",
    "pd.set_option('display.precision', 4)\n",
    "pd.set_option('display.max_colwidth', -1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load combined features + label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of features: 1042\n",
      "pdf_features_label: (356255, 1045)\n",
      "pdf_train: (216948, 1045)\n"
     ]
    }
   ],
   "source": [
    "pdf_features_label = pd.read_csv(os.path.join(\"features\", \"pdf_features_label.csv.bz2\"), compression=\"bz2\")\n",
    "pdf_train = pdf_features_label.query(\"tvt_code == 'train'\")\n",
    "meta_cols = [\"SK_ID_CURR\", \"TARGET\", \"tvt_code\"]\n",
    "ls_features = [cname for cname in pdf_features_label.columns if cname not in meta_cols]\n",
    "\n",
    "# \n",
    "print(\"Number of features: {}\".format(len(ls_features)))\n",
    "print(\"pdf_features_label: {}\".format(pdf_features_label.shape))\n",
    "print(\"pdf_train: {}\".format(pdf_train.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load feature important"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['auc',\n",
       " 'ls_tracked_auc',\n",
       " 'ls_curr_features',\n",
       " 'imp',\n",
       " 'ls_tracked_imp',\n",
       " 'model',\n",
       " 'features']"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# read model\n",
    "version = \"v07\"\n",
    "with open(\"../05_modeling/models/xgb_model_{}.mod\".format(version), \"rb\") as input_file:\n",
    "    res_model = pickle.load(input_file)\n",
    "res_model.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>F-score</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feat_name</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>prev_app_NAME_PRODUCT_TYPE_walk_in_max</th>\n",
       "      <td>162.8774</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>baseline_extend_EXT_SOURCE_2</th>\n",
       "      <td>126.3244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>baseline_extend_EXT_SOURCE_3</th>\n",
       "      <td>117.8964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bureau_CREDIT_DAY_OVERDUE_TO_YEARS_median</th>\n",
       "      <td>97.5458</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NAME_EDUCATION_TYPE_Higher_education</th>\n",
       "      <td>77.9346</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            F-score\n",
       "feat_name                                          \n",
       "prev_app_NAME_PRODUCT_TYPE_walk_in_max     162.8774\n",
       "baseline_extend_EXT_SOURCE_2               126.3244\n",
       "baseline_extend_EXT_SOURCE_3               117.8964\n",
       "bureau_CREDIT_DAY_OVERDUE_TO_YEARS_median  97.5458 \n",
       "NAME_EDUCATION_TYPE_Higher_education       77.9346 "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load most important feature set\n",
    "pdf_imp = pd.DataFrame(res_model[\"imp\"])\n",
    "pdf_imp.rename(columns= {0: \"feat_name\", 1: \"F-score\"}, inplace=True) \n",
    "pdf_imp.set_index(\"feat_name\", inplace=True)\n",
    "pdf_imp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# analyse tracking important features\n",
    "ls_imp = []\n",
    "for track in res_model[\"ls_tracked_imp\"]:\n",
    "    imp = pd.DataFrame(track)\n",
    "    imp.rename(columns= {0: \"feat_name\", 1: \"imp\"}, inplace=True) \n",
    "    ls_imp.append(imp)\n",
    "pdf_analysis = pd.concat(ls_imp).groupby(\"feat_name\").agg({\"imp\": [\"max\", \"min\", \"mean\", \"std\"], \"feat_name\": \"count\"})\n",
    "\n",
    "# rename columns\n",
    "name01 = pdf_analysis.columns.get_level_values(0)\n",
    "name02 = pdf_analysis.columns.get_level_values(1)\n",
    "rename_cols = [\"{}_{}\".format(tpl[0], tpl[1]) for tpl in zip(name01, name02)]\n",
    "rename_cols[-1] = \"num_chosen_by_model\"\n",
    "pdf_analysis.columns = rename_cols\n",
    "pdf_analysis = pdf_analysis.sort_values(by=\"imp_max\", ascending=False)"
   ]
  },
  {
   "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>imp_max</th>\n",
       "      <th>imp_min</th>\n",
       "      <th>imp_mean</th>\n",
       "      <th>imp_std</th>\n",
       "      <th>num_chosen_by_model</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feat_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>prev_app_NAME_PRODUCT_TYPE_walk_in_max</th>\n",
       "      <td>162.8774</td>\n",
       "      <td>71.1525</td>\n",
       "      <td>117.0150</td>\n",
       "      <td>64.8593</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>baseline_extend_EXT_SOURCE_2</th>\n",
       "      <td>127.9498</td>\n",
       "      <td>124.9303</td>\n",
       "      <td>126.4015</td>\n",
       "      <td>1.5112</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>baseline_extend_EXT_SOURCE_3</th>\n",
       "      <td>123.9631</td>\n",
       "      <td>108.6327</td>\n",
       "      <td>116.8307</td>\n",
       "      <td>7.7206</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bureau_CREDIT_DAY_OVERDUE_TO_YEARS_median</th>\n",
       "      <td>97.5458</td>\n",
       "      <td>13.5619</td>\n",
       "      <td>55.5538</td>\n",
       "      <td>59.3855</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NAME_EDUCATION_TYPE_Higher_education</th>\n",
       "      <td>78.8431</td>\n",
       "      <td>73.9365</td>\n",
       "      <td>76.9047</td>\n",
       "      <td>2.6104</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            imp_max   imp_min  imp_mean  imp_std  num_chosen_by_model\n",
       "feat_name                                                                                            \n",
       "prev_app_NAME_PRODUCT_TYPE_walk_in_max     162.8774  71.1525   117.0150  64.8593  2                  \n",
       "baseline_extend_EXT_SOURCE_2               127.9498  124.9303  126.4015  1.5112   3                  \n",
       "baseline_extend_EXT_SOURCE_3               123.9631  108.6327  116.8307  7.7206   3                  \n",
       "bureau_CREDIT_DAY_OVERDUE_TO_YEARS_median  97.5458   13.5619   55.5538   59.3855  2                  \n",
       "NAME_EDUCATION_TYPE_Higher_education       78.8431   73.9365   76.9047   2.6104   3                  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get top features\n",
    "pdf_analysis.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>imp_max</th>\n",
       "      <th>imp_min</th>\n",
       "      <th>imp_mean</th>\n",
       "      <th>imp_std</th>\n",
       "      <th>num_chosen_by_model</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feat_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>bureau_balance_lt1year_STATUS_Sold_sum_median</th>\n",
       "      <td>7.3072</td>\n",
       "      <td>7.3072</td>\n",
       "      <td>7.3072</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pos_cash_gt3year_is_DPD_DEF_max</th>\n",
       "      <td>5.9348</td>\n",
       "      <td>5.9348</td>\n",
       "      <td>5.9348</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is_REG_REGION_NOT_LIVE_REGION</th>\n",
       "      <td>5.1233</td>\n",
       "      <td>5.1233</td>\n",
       "      <td>5.1233</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ORGANIZATION_TYPE_Other</th>\n",
       "      <td>5.1182</td>\n",
       "      <td>5.1182</td>\n",
       "      <td>5.1182</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bureau_balance_STATUS_Sold_sum_std</th>\n",
       "      <td>3.5234</td>\n",
       "      <td>3.5234</td>\n",
       "      <td>3.5234</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               imp_max  imp_min  imp_mean  imp_std  num_chosen_by_model\n",
       "feat_name                                                                                              \n",
       "bureau_balance_lt1year_STATUS_Sold_sum_median  7.3072   7.3072   7.3072   NaN       1                  \n",
       "pos_cash_gt3year_is_DPD_DEF_max                5.9348   5.9348   5.9348   NaN       1                  \n",
       "is_REG_REGION_NOT_LIVE_REGION                  5.1233   5.1233   5.1233   NaN       1                  \n",
       "ORGANIZATION_TYPE_Other                        5.1182   5.1182   5.1182   NaN       1                  \n",
       "bureau_balance_STATUS_Sold_sum_std             3.5234   3.5234   3.5234   NaN       1                  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get tail features\n",
    "pdf_analysis.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_distribution_over_target(pdf_input, cname):\n",
    "    fig, axes = plt.subplots(nrows=1, ncols=3, figsize=(15, 5))\n",
    "    \n",
    "    #\n",
    "    pdf_input.groupby(\"TARGET\")[cname].plot(kind='kde', ax=axes[0], rot=45)\n",
    "    pdf_input.groupby(\"TARGET\")[cname].hist(bins=100, ax=axes[1], xrot=45)\n",
    "    pdf_input.boxplot(column=cname, by='TARGET', ax=axes[2])\n",
    "    \n",
    "    #\n",
    "    plt.suptitle(\"Distribution of {} (0: blue, 1: red)\".format(cname))\n",
    "    plt.show()\n",
    "    \n",
    "def my_auc(y_score, y_true, flexible_sign=True):\n",
    "    # filter NaN\n",
    "    idx = np.isfinite(y_score)\n",
    "    xxx = y_score[idx]\n",
    "    yyy = y_true[idx]\n",
    "\n",
    "    # if label not only 1s/0s\n",
    "    if yyy.std() > 0.0:\n",
    "        auc = metrics.roc_auc_score(y_score=xxx, y_true=yyy)\n",
    "    else:\n",
    "        auc = 0.5\n",
    "\n",
    "    # for evaluation only\n",
    "    if (auc < 0.5) & (flexible_sign):\n",
    "        auc = 1.0 - auc\n",
    "    return auc\n",
    "\n",
    "def feature_evaluate(pdf_feat_label, ls_feat=None):\n",
    "    out_res = {\n",
    "        \"feat_name\": [],\n",
    "        \"auc\": [],\n",
    "        \"corr\": [],\n",
    "        \"coverage\": []\n",
    "    }\n",
    "\n",
    "    # calculate correlation\n",
    "    pdf_corr = pdf_feat_label[[\"TARGET\"] + ls_feat].corr()\n",
    "\n",
    "    for feat in ls_feat:\n",
    "        out_res[\"feat_name\"].append(feat)\n",
    "        out_res[\"auc\"].append(my_auc(pdf_feat_label[feat], pdf_feat_label[\"TARGET\"]))\n",
    "        out_res[\"corr\"].append(pdf_corr.loc[feat, \"TARGET\"])\n",
    "        out_res[\"coverage\"].append((~pdf_feat_label[feat].isna()).mean())\n",
    "\n",
    "    #\n",
    "    pdf_res = pd.DataFrame(out_res)\n",
    "    pdf_res = pdf_res[[\"feat_name\", \"auc\", \"corr\", \"coverage\"]].sort_values(by=\"auc\", ascending=False)\n",
    "    pdf_res.set_index(\"feat_name\", inplace=True)\n",
    "    return pdf_res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>imp_max</th>\n",
       "      <th>imp_min</th>\n",
       "      <th>imp_mean</th>\n",
       "      <th>imp_std</th>\n",
       "      <th>num_chosen_by_model</th>\n",
       "      <th>auc</th>\n",
       "      <th>corr</th>\n",
       "      <th>coverage</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feat_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>prev_app_NAME_PRODUCT_TYPE_walk_in_max</th>\n",
       "      <td>162.8774</td>\n",
       "      <td>71.1525</td>\n",
       "      <td>117.0150</td>\n",
       "      <td>64.8593</td>\n",
       "      <td>2</td>\n",
       "      <td>0.5468</td>\n",
       "      <td>0.0595</td>\n",
       "      <td>0.9467</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>baseline_extend_EXT_SOURCE_2</th>\n",
       "      <td>127.9498</td>\n",
       "      <td>124.9303</td>\n",
       "      <td>126.4015</td>\n",
       "      <td>1.5112</td>\n",
       "      <td>3</td>\n",
       "      <td>0.6559</td>\n",
       "      <td>-0.1602</td>\n",
       "      <td>0.9978</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>credit_card_balance_AMT_DRAWINGS_CURRENT_min</th>\n",
       "      <td>10.4587</td>\n",
       "      <td>6.8901</td>\n",
       "      <td>8.2056</td>\n",
       "      <td>1.9604</td>\n",
       "      <td>3</td>\n",
       "      <td>0.5124</td>\n",
       "      <td>0.0158</td>\n",
       "      <td>0.2824</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               imp_max   imp_min  imp_mean  imp_std  num_chosen_by_model     auc    corr  coverage\n",
       "feat_name                                                                                                                         \n",
       "prev_app_NAME_PRODUCT_TYPE_walk_in_max        162.8774  71.1525   117.0150  64.8593  2                    0.5468  0.0595  0.9467  \n",
       "baseline_extend_EXT_SOURCE_2                  127.9498  124.9303  126.4015  1.5112   3                    0.6559 -0.1602  0.9978  \n",
       "credit_card_balance_AMT_DRAWINGS_CURRENT_min  10.4587   6.8901    8.2056    1.9604   3                    0.5124  0.0158  0.2824  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ls_feat = [\"prev_app_NAME_PRODUCT_TYPE_walk_in_max\", \"baseline_extend_EXT_SOURCE_2\", \"credit_card_balance_AMT_DRAWINGS_CURRENT_min\"]\n",
    "pdf_eval01 = feature_evaluate(pdf_train, ls_feat=ls_feat)\n",
    "pdf_eval01 = pdf_analysis.join(pdf_eval01, how=\"inner\")\n",
    "pdf_eval01"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# AUC\n",
    "\n",
    "Seperate càng cao thì càng tốt "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pdf0 = pdf_train.query(\"TARGET == 0\")\n",
    "pdf1 = pdf_train.query(\"TARGET == 1\")\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "pdf0[\"prev_app_NAME_PRODUCT_TYPE_walk_in_max\"].value_counts().plot(kind=\"bar\", ax=ax, color=\"blue\")\n",
    "pdf1[\"prev_app_NAME_PRODUCT_TYPE_walk_in_max\"].value_counts().plot(kind=\"bar\", ax=ax, color=\"red\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>imp_max</th>\n",
       "      <th>imp_min</th>\n",
       "      <th>imp_mean</th>\n",
       "      <th>imp_std</th>\n",
       "      <th>num_chosen_by_model</th>\n",
       "      <th>auc</th>\n",
       "      <th>corr</th>\n",
       "      <th>coverage</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>baseline_extend_EXT_SOURCE_2</th>\n",
       "      <td>127.9498</td>\n",
       "      <td>124.9303</td>\n",
       "      <td>126.4015</td>\n",
       "      <td>1.5112</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.6559</td>\n",
       "      <td>-0.1602</td>\n",
       "      <td>0.9978</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               imp_max   imp_min  imp_mean  imp_std  num_chosen_by_model     auc    corr  coverage\n",
       "baseline_extend_EXT_SOURCE_2  127.9498  124.9303  126.4015  1.5112   3.0                  0.6559 -0.1602  0.9978  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "attr = \"baseline_extend_EXT_SOURCE_2\"\n",
    "display(pdf_eval01.loc[attr].to_frame().T)\n",
    "plot_distribution_over_target(pdf_train, attr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>imp_max</th>\n",
       "      <th>imp_min</th>\n",
       "      <th>imp_mean</th>\n",
       "      <th>imp_std</th>\n",
       "      <th>num_chosen_by_model</th>\n",
       "      <th>auc</th>\n",
       "      <th>corr</th>\n",
       "      <th>coverage</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>credit_card_balance_AMT_DRAWINGS_CURRENT_min</th>\n",
       "      <td>10.4587</td>\n",
       "      <td>6.8901</td>\n",
       "      <td>8.2056</td>\n",
       "      <td>1.9604</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.5124</td>\n",
       "      <td>0.0158</td>\n",
       "      <td>0.2824</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              imp_max  imp_min  imp_mean  imp_std  num_chosen_by_model     auc    corr  coverage\n",
       "credit_card_balance_AMT_DRAWINGS_CURRENT_min  10.4587  6.8901   8.2056    1.9604   3.0                  0.5124  0.0158  0.2824  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA8wAAAFnCAYAAABzbMBqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzs3Xuc1VW9//HXO1As7ylhXhJLKpCyjGNWdIIoRSuxcywdLdGmzJNS51hep5+aNaVdNCPNo415SQfJtChRM2VKyht2tMDRxEuKeEFRFFMM/Pz+WGvDdzZ779kjDDPDfj8fj3mw9/qu7/qu77582eu71vosRQRmZmZmZmZm1tVr+roCZmZmZmZmZv2RG8xmZmZmZmZmFbjBbGZmZmZmZlaBG8xmZmZmZmZmFbjBbGZmZmZmZlaBG8xmZmZmZmZmFbjBbGZmZmZmZlaBG8xmBoCkcyX9v7VU1pskLZU0KD/vkPT5tVF2Lu8aSZPXVnk9OO63JD0l6fF1fNzhkkLS4Py8T86/FkmHSppdR76QtPOrPMZDkj7yavY1s/pImidp3Brs3y5pv7VUl6rXi3qvOQOdpHGSFhSe3yZpl76sk1mjcYPZrAHkhsaLkp6X9KykP0s6QtLKa0BEHBER36yzrJqNloh4OCI2iYgVa6Hup0j6eVn5e0fERWtadg/rsQPwVWBURGyzLo9drnj+jfKjcSDI70VI+nRZ+ricfmVZ+q45vaNwk6n0F5JeKDz/YI3jXijp5fz9fl7SXEnfkbR5Wd1W5LKek3SXpI9XKGvjnGdmWfoJFdLuq5J2YH68srGTv8ch6VOFvINz2vBC2hhJv5X0TL5W3S2pVdKWefuGkn4gaUGu54OSzqz22pTV7SBJc/J+j+UbT2MLr+G3yvKX36gqXUeXSno877NJhfdhqaTFkq6X9PYq70Hxb9tC+U9I2riwz+fX9PPRUxGxS0R0vJp9Jb0T2BX4dSHtIEn/yPX9laTXr6WqrhOS3ihphqSF5Z/XPvJ94NQ+roNZQ3GD2axxfCIiNgV2BE4DjgPa1vZBSj8u10M7Ak9HxJNrUoiSAX3tVR45YKuZDCzO/5ZbBLxf0lZl+f8OXW4ybRIRpUbYroW0m7o59nfz93socBiwB/CnYuMLuDmXvQVwDjBN0hZl5ewPLAP2lPTGQvofgQ9o1aiRbYANgN3K0nbOeStZDJxa7fMj6f1AB/An4O0RsQUwEVhOaoQBnACMAXYHNgXGA/9X7UUplH008EPg28Aw4E35NZjU3b5lPpFfw3cB7871Kfpu3r4d8CirX2NvLr7P+W9hYftg4CvlB10Ln4915YvApRERAEo9of8LfJb0uv+T9LoPJK8A1wL/uaYFraX/H2cA48u+n2bWiwb0jzYz67mIWBIRM4ADgMmSRkPXHhZJW+denmdzT8lNkl4j6RLSD83f5F6NYwu9MM2SHgZuLO+Zyd6iNJRsiaRfl3oZVDbcLKc9JOkjkiYCJwIH5OPdlbevHOKd6/X13IPxpKSLlXvWCvWYLOlhpeHULdVeG0mb5/0X5fK+nsv/CHA9sG2ux4VV9p8k6U6lHrz7c/1L9W2V9CfSD8Y352O1KfV0Pao03LvU8Bgk6fu5vg8AHys7TkfueRoJnAu8L9fr2VrvvaTXKvXO/SO/D7MlvTZv+4VSr9kSSX9UYchf/mz8RNJMSS+QfqxtlXtdnpN0G/CWWscus4+kB/L5fa90A0HSWyTdKOnpvO3SCg26Up12l3Rz/ow+JunHkjYsbA+lURT3KfVWni1Jhe1fkNSp1CN7t6Tdcvq2kn6ZPwMPSvpyPSckaUfgQ8DhwF6ShpVleRn4FVDqfR0EfBq4tN4XrR4R8VJE3A7sC2xFajyX53kFuATYGBhRtnky6TP1V+DgQvrtpAbyu/LzfwdmAfeWpd1f1gAsupb0OnymyvbvAj+LiO9ExBO5rg9HxMmFHs9/A66KiIWRPBQRF1cpD0jfa1KP3JERcWVEvBAR/4qI30TEMbX2rSYiHgeuY9W5l29/EZhebXsN3wO+Vu1z/2rk7+85Sj3qSyX9SdI2kn6Yvxv3SHp3If/KUURKIwOm5+vi80rDtcfUONzewB8Kzw8GfhMRf4yIpcD/A/5D0qY9OIWK14uyc1zt/xyVTQWS9Ln8nX9G0nX5O9utiHgiIs4hfQd6LL+ex0n6K/CC0siKqteZfJ2+MNfzbtJnvlifl4A7gD1fTX3MrOfcYDZrUBFxG7AAqDSU76t521BSr8CJaZf4LPAwuZclIr5b2OdDwEhgryqHPAT4HLAtqcfoR3XU8VpSj9Dl+Xi7Vsh2aP4bD7wZ2AT4cVmescDbgAnASbmhWclUYPNczodynQ+LiN+TfgguzPU4tHxHSbsDFwPHkHrw/h14qJDls6TG1KbAP4CLSK/DzqSeqj2B0o+7LwAfz+ljSL1+q4mITuAIVvVadfcj+/vAe4D3A68HjiX1ngBcQ2o8vQH4C6s35A4CWnP9ZwNnAy8BbyS9r5/r5thFn8zntRuph6+0r4DvkD4jI4EdgFOqlLEC+B9ga+B9pPf2S2V5Pk76sbkrqXG6F4DSsOBTSO/vZqTG5dP5h/hvgLtIPYQTgP+WVO0zXXQIMCcifgl00rWxWXJxzkeuyzygWuNyjUTE86SbPKt9v3Nj/TDgX6TPYin9TcA40nt/aaGuRMTLwK2kzzX535tIn4ViWrXeZYAgNZhOlrRBWZ02Jr2Pv+zm1G4Bjpb0JUnvKN4EqeF9wEbAVXXkrYuk7UnXhPlVtm8MNFXbXsMcUi/719akfhV8Gvg66fuyDLiZ9D3fGrgCOKPGvvsC00jXtRmsfn0FVp7zTqSbKCW7kL5PAETE/aSbJm/N+5wjqbse52rXi7opzak+EfgP0v9rNwHtPS2nStnHS/ptN9maSDc+tyBdc2tdZ04m3YB8C+k6UWnESierRl2YWS9zg9mssS0kNZzK/YvUENox98TcVBpiV8MpuefmxSrbL4mIuRHxAulH86e1dob2HgycEREP5B6ME4AD1bV3+xsR8WJE3EX6kbLaD41clwOAEyLi+Yh4CPgBqaFbj2bggoi4PiJeiYhHI+KewvYLI2JeRCwnveZ7A/+dX7MngTPJvY+kH7c/jIhHImIxqRG5RnJj8HPAV3LdVkTEnyNiGUBEXJDPexmpMbmrCnNggV9HxJ9y7+S/SMMTT8r1n0u6AVCv0yNicUQ8TBom25TrMD+/fssiYhHpR/yHKhUQEXdExC0RsTy/V/9bIe9pEfFsPs4sVvX2fZ40dPb23Es5PyL+QWpcD42IUyPi5Yh4ADifVe9LLYcAl+XHl1HhR25E/Bl4vaS35fw1e0bXgvLv9x5KoxBeIt08+Ux0nWJwCPDXiLib1JjYpdjzSOo5LDWOP0hqdNxUllbsXVxNHt2yiFU3h0q2JP0mWRlQT9J38wiCFyR9PSd/Bzid9L2fAzyq7gPgbQU8lb97a+pXkp4HHgGeJDVuir6WX+PnSTfqyq8fe+RzKv3dX+EYJwFTJA1dC/UtuSp/Z14i3Th4KSIujhRn4nLSzblqZkfEzJz3Eqo31Eo37J4vpG0CLCnLt4R0442I+FJElN/oKlfxetFDXwS+ExGd+XPwbeBd9fYy1xIRp0XEavEAyvwoX89fpPvrzKeB1nzOj1D55vLzrHq9zayXucFs1ti2I80rLPc9Us/I7/JQuOPrKOuRHmz/B2l459Z11bK2bSn0kuXHg0k94yXFqNb/JP2IK7c1sGGFsrarsx47AJV+/JYUz39H0vk/VvrhTGrwvSFv35bVX681tTWpl221OioNAT9NaRj5c6zqGS++P8X6DCW9xq+2juX7lYIevUHSNKUh6s8BP6fKZ0TSW5WmDTye8367Qt5q73u192pH0rD7Zwvvy4l0/SxVqssHSD1r03LSZcA7JFUajnsJcBRpRMRa6/Gsovz7fUsehbAlqaewvPf5EPLIgkjDqv9A14b/H4GxSgG4hkbEfcCfSXOztwRGU7uHueTrQAvp81jyDKnnbeW8zIg4Ntf3KtLnjXyj5+yI+ACpwdAKXFBj1AjA08DWqj1/dDnpO1m0Qa7TK4W0/SLNFR8HvJ3VP3Pfz3UeDrxIGtlSdEtEbFH4W20qQ74B9VugnutuvZ4oPH6xwvNK18SS8u/RRlVey9KUkOJw66WkURxFm9G1Ud2diteLHtoROKvwvV5MGtFS7/V9TZVf/2tdZ+q5/m/KqtfbzHqZG8xmDUrSv5F+LKwWYTn3NH41It4MfII0BHJCaXOVIrvrgd6h8PhNpF7Kp4AXgNcV6jWI1CCrt9yFpB8gxbKX0/UHYT2eynUqL+vROvd/hNrzeIvn8QhpWOTWhR/Om0VEad7wY6z+etVTbi1PkXoWK9XxINJQx4+QhqQPz+nF4a7F4ywivcb11rFc+X6lYcnfycd5Z0RsRprrWm3I7U+Ae4AROe+JNfKWq/ZePQI8WNag2TQi9ummvMn52HcqLTl2a04/pELeS0hDx2dGxD/rrG+PKUVv/gipB7iLPBLjS8BnSz3ISgG3RgAn5JsQjwPvBZoKjaObSZ+Pw0mBuYiI50jv3+GkKQsPdle3iLiedEPuS4W0F0iv23/Ue4551MjZpMb2qBpZbyZ99mstdfQwqz73JTsBj+RRFeXH/gNwIamnvlLdHiYF7zpLOU5AD51Mmpqxrhp0ayy/h/eTh1tn8yj0SEt6MzCEHOyuTtWuF0Uv5H9fV0grrmbwCPDFsu/2a/Ooj3Wh/Ppf6zpTz/V/JIWh7mbWu9xgNmswkjZTWk5mGvDziPhbhTwfl7Rznh/4HGm+aGmJqCdIc3x76jOSRkl6HSkAzxV5iN/fST0WH8vzGr9O+kFV8gQwvFKgl6wd+B9JO+VGQmnOc4+GX+a6TAdaJW2ah+odTerlrEcbcJikCUqBwrZTYUmZsmM9BvwO+EF+P16jFPCqNKR4OvBlSdvnnrtaPU1PANurEPCqyjFfAS4AzsgBZwZJep+kIaTeimWknrjXkV7DWmWtAK4ETpH0OkmjqDzPrppjJG2ptFTXV0hDQsn1WAo8K2k70nzwajYlfTaX5tf5v3pw/J+Shs6+R8nO+f2+DXhOKUDPa/NrNDrfXKpI0kakIZSHk4Z8l/6mAAeX98TlBuWHSD2sa52kIZLeQwow9gzws0r5IuJp0utwUk6aTJrzPKpwDqNJn4e98z4vkoZBH03XhvjsnFZP73JJC2kOfdGxwOfynNA35PPZntRwLZ3ffysFCnytUvCkyaTPQtVI2RGxJJ/n2ZL2y5/ZDSTtLakUh+GXwMck7Znf921J16Jp1colDQ/+aJWRBKUbA6WbCT0SEfNJ34u6gs71IzPpOjXiUuATkj6oNMf5VODKSHPs61XterFSnsLxKOn/mUGSPkfXm2Lnkm4G7QIrAzwWlzjrkHRKtQrk73np/6Uh+fmr1d11Znqu65b58z+lrC5DSLEorl+DOphZD7jBbNY4fqNVc+9aSPNDV4ugm40Afk9qvNwMnBOrotR+B/h6HkrWk8A0l5B6ZB4nDcX8Mqz8Mfsl0o/3R0k9BcWo2b/I/z4t6S8Vyr0gl/1H4EFST9KUCvnqMSUf/wFSI+CyXH63IgVRO4w0F3kJaThrrflxh5CGgN9NathcwarhqOeTIvDeRQrMc2WlArIbSb04j0t6qptqfg34Gyna62LSXNDXkObS/oP0+t9NCqzUnaNIwzgfJ72vFRtmVfyaFOX1TuBqVi298w1SYJ8lOb3WeX+N1DP+POn1Wu1HdDUR8QvSUN7L8v6/Al6fbwR8gtRYfJDUK/9TUq9qNfuRhrReHBGPl/7yOQ0iLYtUfvzZUT2S9Kt1bP5+Lya9n3cA78+9ftX8kBSB+J2kRv/U4jnkxv0ldL0Z8gfS1IHiyJSbclrdDeaI+BOp4VBMmw18mDQn+u95qOq1pCBYU3O2F0mxBR4nvT9HAv+Z54HWOt4ZpEb910kjJB4hfYZ/lbfPI82N/Q7pNbyZ1OP9jRplLiK91v+vxqG/R3pvSo2tUkT74l+1GzKnkiKZDyTnkW4UCVa+rkeQGs5Pkm5urBxZIOlcSed2U2a160W5L5Busj1NCja2svc4Iq4iXe+mKU3hmEu+EZTtQB41UcWLpP8PIY1sWRmrQ9KJkq7p5hxWquM68w3S9fhB0o3VS8qK2Bfo6IVriJlVoeg2jo+ZmZmZWfckXQZMj4hf9XVd6pF7cX8REe/r67rUQ9KtQHOe625m64AbzGZmZmZmZmYVeEi2mdl6RNK8CkM+l0qqtC5wbxz/g1WOv7T7vfsvSddUOa8T19Hx+/R97e8kvana505pfen1mj8fZma9xz3MZmZmZrbOSOogBZ38aW7UT46IPfu4Wl1IClIU/vk18lwILIiIr1fLU2PfU4CdI+Izr7qSZlZTvqn95oj4/JqU4x5mMzMzM+sTEXFpsbEsKSTt3Jd1MpC0cR6lMLPCtockvSxp67L0O/P7N7xsVM6/cv7S86qB3nIU/FcKeRdIml4eHC8f54Wc51FJZygtS1le3oWSlitFvi+lvTHvP6yQ1lIl7dpCOd/Kj4fnvFeXHevnKkRbV1px44z8er0g6WFJV0javZBnUn7dnpP0lKQbJA2v9voU9nurpF/kfZZI+quko5Wiro+TtKDCPh2SPp8fn5Lfl6VKQVz/LOl9hbzF9+F5SfdKOqysvOJ7UPo7tlB+qGs0+sFr+vnoqYj49po2lsENZjMzMzN7lVS2dNpAM9Dr34v2Jy03uKekN1bY/iApsjwAkt4BrFxzPCL2johNImITUpT075aeR8QR3Rx7Yd5vU2APUmTymyRNKMu3a873IeAA4HPFjUpLmf0naeWFldMT8tKO80kR+Uv+PR+nPK1W9P89JH2g0galyPg3Au8APg5sRlo/exqwT86zMynS/ldJUdJ3As4BVlv7vazst5Ci+D8CvCMiNgc+BYwhvWb1ujy/flsDs1i1KklJ6X3YDPgf4HxJbyvLs2vhfd0kIr5b2LYYOLXSjYw1/Hysc24wm5mZmdlqJO0g6UpJiyQ9LenHkg6V9CdJZ0paDJyS835OUqekZyRdp7S2eamcj0q6J/eE/RhQYduhkmbnx6XGyV25p+mAbupX7J27X9LEnH5Yrsvzkh6Q9MXCPuOUei2Pk/Q4eUk8ScdIekzSQqV1nOu1taTr87H+UHbeZ0l6JNfvDkkfrHEuv5D0eH6N/qi8ZnTedqGksyVdnY9za240lbbvkuuwWNITyrEVJL1GaV3z+/P7N13S6+s8r8mk9av/SqGxWXAJaXnEYv6L6yy7LpEsiIiTSEtvnV4l33zSsmDla6L/J/AsaYm2yWXb/khuHOcG3buBs8rS3kftBvN3gW9V2fZZYHtgv4iYGxErIuKFiLgiIk7Jed4FPBgRN+RzfT4ifhkRD9c4JqSlx/4cEUfnxj8RcW9EHBQRz3az72oiYjmp0bqdpKEVtkdEzCQ1gN/Zg6KvBV4G1tq0g8L391hJT+bv7H6S9pH09/wdOLGQ/xRJP8+PSyMDJiv19j8lqaWe47rBbGZmZmZd5AbDb0lrAg8HtiP1jgG8l7Re/RuAVkn7AScC/wEMJa3N3Z7L2Rr4JWkN7K2B+4GKvXIRUerdK/VaVV1fXWlY68WktZe3IDV0Hsqbn2RVr95hwJmSdivsvg3wemBH4PDc0P4a8FFgBPCRWq9NmYOBb+Zzu5PU8Ci5ndQoej1p3fdfSNqoSjnX5GO/AfhLWTmQenO/AWxJ6h1thTTsF/g9qXGyLbAzcEPe58ukteI/lLc9A5zd3QkpBcobl+twKV0bxiW3AJtJGpk/KwcAP++u7DVwJbBb7jXuQtLbgQ+SXpeiyaTP4TTg7WWfgZUNZlJj+R7S61ZM24Cy9eLLnA28VVKlz8tHgOsi4oUa+/8l1+tMSeMlbVIjb3nZV9SZt1uSNiS9x0+TPiPl218jaV/SZ7zqnP4KgrRO/cmSNlgbdc22ATYiXZNOAs4nNcrfQ/ocnCTpzTX2Hwu8DZiQ847s7oBuMJuZmZlZud1Jjaxjcs/YSxExO29bGBFTI2J5RLwIfBH4TkR05t6qbwPvyr2t+wB35561fwE/BB5fC/VrBi6IiOsj4pWIeDQi7gGIiKsj4v7cM/YH4HekH9IlrwAnR8SyXP9PAz/LPYEvkHvN63R1RPwxIpYBLcD7JO2Q6/HziHg6v04/AIaQfqivJiIuyD2My/Lxd5W0eSHLlRFxW6E3sNSb+nHg8Yj4QX6Pno+IW/O2LwItuZe2VO7+6n4Y+iHAXyPiblKDcxdJ766Qr9TL/FFSg/PRbspdEwtJIxO2KKT9RdILQCfQQRrODKxs9I8HLouIJ0iN4WIv8x+A0ZK2JH02boqI+0gjBkppt0TEyzXq9BLpxkWlXuatKXzOJb1Laa7wc5LuBYiIB0g3JrYDpgNP5dEE3TWctwIe6yZPPT4t6VngReALwP7581WybWH7VcDREfF/ZWX8JZ9X6W+v4saImAEsAtZ4HnHBv4DWfD2ZRnqtz8qf/XnAPGr3hH8jIl6MiLuAu4BduzugG8xmZmZmVm4H4B9lP6BLHil7viNwVulHM2nopkgNgW2L+SMiKuz/aut3f6UNkvaWdEsenvksqdFeDFC1KCJeKjzvUkdSr3q9iue2lHTu2+Z6fDUPDV+S67F5WT1K9R0k6bQ8dPo5VvWUF/MWbzL8Eyg1qqq+DqT35arC+9IJrACGVclfcgi5hzsiFpIal+VDmiE1mA8CDmUtD8euYDtSj2VxyPFupNfhANKoh2Lv82eBzoi4Mz+/FDio1NMZEQ8BC0i9jf9OGhUBcHMhrdZw7JLzgWGSPlGW/jSwcu53RNwZEVuQRmEMKaTfEhGfjoihpEb6v5NuvNTSpewKlpN6x8ttQGpslkzPdRoGzCX10BYtzNs3A34EfLhCmbtFxBaFv+sq5Pk66Zyqja7oqacjYkV+/GL+94nC9hdZ9f2opNp3qSo3mM3MzMys3CPAm6r0RpavSfoI8MWyH86vjYg/k3rCdihllKTi8zWs31vKE5WCLf0S+D4wLP/gn0lh3nSF+nepI9CTtbuL57YJafj1wjxf+ThS7/WWuR5LyupRchAwiTTUdnPSEHiq5C1X8XUobNu77H3ZKCKq9gRLej9paPgJeU7146TGaFP5ZyEi/kEK/rUPach0b/ok8JfyIc55FMF0UkP3pMKmQ4A3F87hDNINiL0LeW4iNVDfB/y5LG0sdTSYcy/nN0jD8ovv1w2kgGmrDSGvUdbtpNdxdDdZf0+an13Nw6Se8pUNwfy925EKN4Mi4inSaIRTVCHAWx6dcBzwjjz9okci4nrSUO4v9XTf/sINZjMzMzMrdxupIXma0hJDG6lKRGBScKgTSoGqJG2uVcvJXE0a0vsfucH1ZdIcxGqeAGrNPyxpAw6TNCHPsdwuz2XdkNSDtwhYLmlvoLs1nqcDh0oaJel1wMl1HL9kH0lj8zzQbwK3RsQjpGjFy3M9Bks6idRTV8mmpIjUTwOvIw1pr9dvgW0k/bekIUpLGb03bzuXNMd8RwBJQyVN6qa8ycD1wCjSsO93kRpwr6NrY7OkGfhwN3N1XxUl20k6mTSk98Qa2U8jzUffRml5pLeQphUUz+EyuvaU/5HUsF4YEc/ltNk5bXNSI7wel5A+cxMLaReTvj9XSRqdRxFsRIpkXTq/sZK+IOkN+fnbgX1J88NrORl4v6TvSdom77uz0rJWW+SgYbcCp0vaJN9EOob0eaxYdp7OcB1wbJXtLwM/oOtNiZ5oqVb2QOAGs5mZmZl1kYc8foIUROph0vDVilGrI+IqUgTjaXlI8Vxy4yr3Xn2K1KB5mtR7+acahz4FuCgPI/50jfrdRg7oReq5/QOwY0Q8T2qUTycFMDoImNHNuV5Dmlt9I6kn7MZa+ctcRmrALCYNaS1FlL6OFMjr76RevZeoPhT94pznUeBuum8wFev+PGkO8SdIQ03vI83dhRT1eQbwO0nP53LfW6kcgNyg+zQwNSIeL/w9SGoUrjYsO88Vn1Nvfeu0raSlwFJS4LR3AOMi4nfVdoiIv5E+A8fkev46Iv5WPA/S6/FxrYoU/gdSkLXZhaLuJC2PdUdE/LOeyubvysmk0QWltJdI78PdpJtGzwH3Av9Geo0hDS/fF/hbPt9rSXOFi0szVTre/aRe8eHAPElLSKMq5gDP52wH5HObT/pcTQD2KZuKUO57pJsOb6iy/QLSqJPi8PNSRPvS3w+r1PlP1A6g1q8pTSUxMzMzMzMzsyL3MJuZmZmZmZlV4AazmZmZmfU7kk4sG+5Z+rtmHdZhXpU6HNz93v2TpIOrnNO8dXT8Pn9f+ztJ11R5jWrN414v9MfPh4dkm5mZmZmZmVXQ3cLlDWnrrbeO4cOH93U1zGwduuOOO57K6yAOeK/2GvbCCy+w8cZ1r4DRa1yP/leP/lAH16N2PWpdwyS9Dbi8kPRmUrTbi3P6cNLav5+OiGfyEjRnkZYL+idwaET8JZc1mbSuKsC3IuKinP4e4EJSwKSZwFciInKApdWOUeu8Gv13WH/5fNm618jvfX/+HeYGcwXDhw9nzpy1HfDPzPozSautTThQvdprWEdHB+PGjVv7FXI9Bnw9+kMdXI/a9ah1DYuIe0lL6yBpEClq7lXA8cANEXGapOPz8+NIEa5H5L/3Aj8B3psbvyeTlsYJ4A5JM3ID+CfA4aRIzDNJS+xcU+MYVTX677D+8vmyda+R3/v+/DvMc5jNzMzMGscE4P6I+AcwCbgop18E7JcfTwIujuQWYAtJbwT2Aq6PiMW5kXw9MDFv2ywibo401+/isrIqHcPMbEBwD7OZmZlZ4zgQaM+Ph0XEYwAR8Vhh/dXt6Lpm8IKcVit9QYX0WsfoQtLhpB5qhg0bRkdHx6s6ufXB0qVLG/r8G5nf+/7JDWYzMzOzBiBpQ2Bf4ITuslZIi1eRXreIOA84D2DMmDHRqMNSobGH5TY6v/f9k4dkm5mZmTWGvYG/RMQT+fkTeTg1+d8nc/oCYIfCftsDC7tJ375Ceq1jmJkNCG4wm5mZmTWGJlYNxwaYAUzOjycDvy6kH6JkD2BJHlYycr7jAAAgAElEQVR9HbCnpC0lbQnsCVyXtz0vaY8cYfuQsrIqHcPMbEDo1QazpImS7pU0P0dGLN8+RNLlefutkoYXtp2Q0++VtFcPypwqaWk9xzAzMzNrBJJeB3wUuLKQfBrwUUn35W2n5fSZwAPAfOB84EsAEbEY+CZwe/47NacB/Bfw07zP/aQI2bWOYWY2IPTaHOa8bMHZpIvjAuD2vPTA3YVszcAzEbGzpAOB04EDJI0iBaXYBdgW+L2kt+Z9qpYpaQywRVlVKh6jF07ZzMzMrF+KiH8CW5WlPU2Kml2eN4Ajq5RzAXBBhfQ5wOgK6RWPYWY2UPRmD/PuwPyIeCAiXgamkZYWKCouNXAFMCEP5ZkETIuIZRHxIOlu5e61yswN9O8Bx9Z5DDMzMzOzfqG9vZ3Ro0czYcIERo8eTXt7e/c7mVmv680o2ZWWHnhvtTwRsVzSEtLdz+1IC98X9y0tT1CtzKOAGXnJgnqO8VQxU3E5gze96U11n6SZmZmZ2Zpob2+npaWFtrY2VqxYwaBBg2hubgagqampj2tn1th6s4e5niUG1sqyBZK2BT4FTH2V9SAizouIMRExZujQoRV2sfsXLe0+k5mZmZn1SGtrK21tbYwfP57Bgwczfvx42traaG1t7euqmTW83mwwV1t6oGIeSYOBzYHFNfatlv5uYGdgvqSHgNdJmt/NMawHfnPXQib84A/8/u4nus9sZmZmZnXr7Oxk7NixXdLGjh1LZ2dnH9XIzEp6s8F8OzBC0k6SNiQF8ZpRlqe41MD+wI050MQM4MAc4XonYARwW7UyI+LqiNgmIoZHxHDgnxGxczfHsB64+7HnALj3ief7uCZmtrYNP/5qhh9/dV9Xw8ysYY0cOZLZs2d3SZs9ezYjR47soxqZWUmvzWHO84WPIq3ZNwi4ICLmSToVmBMRM4A24JLcG7yY1AAm55sO3A0sB46MiBUAlcrspioVj2FmZmZm1h+0tLTQ3Ny8cg7zrFmzaG5u9pBss36gN4N+EREzSWv5FdNOKjx+iTT3uNK+rcBqV4lKZVbIs0k9x7D6lSaCu3PezMzMbO0qBfaaMmUKnZ2djBw5ktbWVgf8MusHerXBbGZmZmZm3WtqaqKpqYmOjg7GjRvX19Uxs6w35zCbmZmZmZmZDVhuMFtdSktbe0S2mZmZmZk1CjeYzczMzMzMzCpwg9nMzMzMzMysAjeYrS7KcbI9ItvMzMzMzBqFG8xWl9IcZjMzMzMzs0bhBrOZmZmZmZlZBW4wW484SraZmZmZmTUKN5itLh6RbWZmZmZmjcYNZjMzMzMzM7MK3GC2HgnHyTYzMzMzswbhBrPVx2GyzczMzMyswbjBbD3ioF9mZmZmZtYo3GC2urh/2czMzMzMGo0bzGZma0jSQ5L+JulOSXNy2uslXS/pvvzvljldkn4kab6kv0rarVDO5Jz/PkmTC+nvyeXPz/v6HpaZmZnZOuAGs/WIR2SbVTU+It4VEWPy8+OBGyJiBHBDfg6wNzAi/x0O/ARSAxs4GXgvsDtwcqmRnfMcXthvYu+fjpmZmZm5wWxm1jsmARflxxcB+xXSL47kFmALSW8E9gKuj4jFEfEMcD0wMW/bLCJujogALi6UZWZmZma9aHBfV8DMbD0QwO8kBfC/EXEeMCwiHgOIiMckvSHn3Q54pLDvgpxWK31BhfQuJB1O6oVm2LBhdHR01F35r75jOQBLly7t0X69xfXof/XoD3VwPfpvPczM1mduMFtdVs6YdJhss0o+EBELc6P4ekn31Mhbaf5xvIr0rgmpkX4ewJgxY2LcuHHdVrrk0OOvBuDCiRvTk/16S0dHh+vRz+rRH+rgevTfepiZrc88JNvqIsfJNqsqIhbmf58EriLNQX4iD6cm//tkzr4A2KGw+/bAwm7St6+QbmZWN0lbSLpC0j2SOiW9b10EJ6x2DDOzgcINZjOzNSBpY0mblh4DewJzgRlA6cfkZODX+fEM4JD8g3QPYEkeun0dsKekLfMPyj2B6/K25yXtkX+AHlIoy8ysXmcB10bE24FdgU7WTXDCascwMxsQerXBLGmipHvz3cbVLpCShki6PG+/VdLwwrYTcvq9kvbqrkxJbZLuyndCr5C0SU4/VNKivNzLnZI+35vnvL7zgGyz1QwDZku6C7gNuDoirgVOAz4q6T7go/k5wEzgAWA+cD7wJYCIWAx8E7g9/52a0wD+C/hp3ud+4Jp1cF5mtp6QtBnw70AbQES8HBHPsm6CE1Y7hpnZgNBrc5glDQLOJv1QXADcLmlGRNxdyNYMPBMRO0s6EDgdOEDSKOBAYBdgW+D3kt6a96lW5v9ExHP52GcAR7HqB+rlEXFUb51rI/Cqr2aVRcQDpN6a8vSngQkV0gM4skpZFwAXVEifA4xe48qaWaN6M7AI+JmkXYE7gK+wboITVjtGF2sSuHB942Bujcvvff/Um0G/dgfm5x+TSJpGustYbDBPAk7Jj68AfpyHHE4CpkXEMuBBSfNzeVQrs9BYFvBa3BlqZmZmBun33m7AlIi4VdJZ1B4a3SvBCWtZk8CF6xsHc2tcfu/7p94ckl3tLmTFPBGxHFgCbFVj35plSvoZ8DjwdmBqId9/FoZqF4PqUNj3cElzJM1ZtGhR3SfZaBwk28zMbMBZACyIiFvz8ytIDeh1EZyw2jHMzAaE3mww13O3ca3ewYyIw0hDuDuBA3Lyb4DhEfFO4PesmkfTtZCI8yJiTESMGTp0aKUsDc0jss3MzAamiHgceETS23LSBNKIv3URnLDaMczMBoTeHJJd7S5kpTwLJA0GNgcWd7NvzTIjYoWky4FjgJ/leYQl55PmSdurFB7pbmZmNhBNAS6VtCEp8OBhpI6T6ZKagYeBT+W8M4F9SIEG/5nzEhGLJZWCE8LqwQkvJE2Lu4ZVwQlPq3IMM7MBoTcbzLcDIyTtBDxKCuJ1UFme0l3Hm4H9gRsjIiTNAC7Lwbu2JS1PcBupo3O1MvPdzLdExPz8+BPAPZCG/5SCTQD7knqfrYcc9MvMzGzgiog7gTEVNvVqcMJqARDNzAaKXmswR8RySUeRhu8MAi6IiHmSTgXmRMQM0vIGl+SgXotJDWByvumk4ULLgSMjYgVAlTJfA1yUl00QcBfpTifAlyXtm8tZDBzaW+dsZmZmZmZm64/e7GEmImaShvUU004qPH6JKkNzIqIVaK2zzFeAD1Qp5wTghJ7W3Spz0C8zMzMzM2sUvRn0y9YjymOy3V42MzMzM7NG4QazmZmZmZmZWQVuMJuZmZmZmZlV4Aaz9YjnMJuZmZmZWaNwg9nMzMzMzMysAjeYzczMzMzMzCpwg9nqkoNkm5mZmZmZNQw3mK1HwgtLmZmZmZlZg3CD2eoi3MVsZmZmZmaNxQ1mMzMzMzMzswrcYLae8YhsMzMzMzNrEG4wW10c9MvMzMzMzBqNG8xmZmZmZmZmFbjBbD3iEdlmZmZmZtYo3GC2upRGZEe4yWxmZmZmZo3BDWYzMzMzMzOzCtxgNjMzMzMzM6vADWarSylKtkdkm5mZmZlZo3CD2czMzMzMzKwCN5jNzMzMzMzMKnCD2eqiHCfbI7LNzMzMzKxRuMFsdSnNYTYzMzOzta+9vZ3Ro0czYcIERo8eTXt7e19XycyAwb1ZuKSJwFnAIOCnEXFa2fYhwMXAe4CngQMi4qG87QSgGVgBfDkirqtVpqQ2YAxpyeC/A4dGxNJaxzAzMzMz62vt7e20tLTQ1tbGihUrGDRoEM3NzQA0NTX1ce3MGluv9TBLGgScDewNjAKaJI0qy9YMPBMROwNnAqfnfUcBBwK7ABOBcyQN6qbM/4mIXSPincDDwFG1jmGvjqNkm5mZma1dra2ttLW1MX78eAYPHsz48eNpa2ujtbW1r6tm1vB6c0j27sD8iHggIl4GpgGTyvJMAi7Kj68AJkhSTp8WEcsi4kFgfi6vapkR8RxA3v+1rJpuW+0YZmZmZmZ9rrOzk7Fjx3ZJGzt2LJ2dnX1UIzMr6c0G83bAI4XnC3JaxTwRsRxYAmxVY9+aZUr6GfA48HZgajfH6ELS4ZLmSJqzaNGinpxnQwmH/TIzMxtwJD0k6W+S7pQ0J6e9XtL1ku7L/26Z0yXpR5LmS/qrpN0K5UzO+e+TNLmQ/p5c/vy8r2odw7oaOXIks2fP7pI2e/ZsRo4c2Uc1MrOS3mwwV+rFLW9tVcvT0/T0IOIwYFugEzigB/UgIs6LiDERMWbo0KEVdjEzMzMb0MZHxLsiYkx+fjxwQ0SMAG7IzyFNfRuR/w4HfgKp8QucDLyXNOrv5EID+Cc5b2m/id0cwwpaWlpobm5m1qxZLF++nFmzZtHc3ExLS0tfV82s4fVm0K8FwA6F59sDC6vkWSBpMLA5sLibfWuWGRErJF0OHAP8rMYxzMzMzBrZJGBcfnwR0AEcl9MvjogAbpG0haQ35rzXR8RiAEnXAxMldQCbRcTNOf1iYD/gmhrHsIJSYK8pU6bQ2dnJyJEjaW1tdcAvs36gN3uYbwdGSNpJ0oakIF4zyvLMAErDefYHbswX5xnAgZKGSNqJdKfytmpl5qFDO8PKOcyfAO7p5hjWA6Vp337lzMzMBqQAfifpDkmH57RhEfEYQP73DTm9p1PjtsuPy9NrHcPKNDU1MXfuXG644Qbmzp3rxrJZP9FrPcwRsVzSUcB1pCWgLoiIeZJOBeZExAygDbhE0nxSr++Bed95kqYDdwPLgSMjYgVAlTJfA1wkaTPSEOy7gP/KVal4DDMzM7MG8oGIWCjpDcD1ku6pkXetTpmrR27EHw4wbNgwOjo6erL7emXp0qUNff6NzO99/9Sr6zBHxExgZlnaSYXHLwGfqrJvK7BaLP0qZb4CfKBKOVWPYWZmZtYIImJh/vdJSVeR5iA/IemNEfFYHnL9ZM5ebWrcAlYNry6ld+T07Svkp8Yxyut3HnAewJgxY2LcuHGVsjWEjo4OGvn8G5nf+/6pN4dk23rE63CZVZfXif8/Sb/Nz3eSdGuOCnt5nkJCnmZyeY4ie6uk4YUyTsjp90raq5A+MafNl+RgOWbWY5I2lrRp6TGwJzCXrtPWJgO/zo9nAIfkKW97AEvycOrrgD0lbZmDfe0JXJe3PS9pjzw17pCysiodw8xsQHCD2erilavNavoKKTp/yenAmTkq7DNAc05vBp6JiJ2BM3M+JI0iTRfZhRRZ9pzcCB8EnE2KWDsKaMp5zcx6YhgwW9JdpJgwV0fEtcBpwEcl3Qd8ND+HNJLvAWA+cD7wJYAc7OubpJgytwOnlgKAkabC/TTvcz8p4Bc1jmFmNiD06pBsM7P1naTtgY+RppAcnXtXPgwclLNcBJxCWnJlUn4McAXw45x/EjAtIpYBD+aYC7vnfPMj4oF8rGk57929fFpmth7J15BdK6Q/DUyokB7AkVXKugC4oEL6HGB0vccwMxso3MNsPeIA42ar+SFwLPBKfr4V8GxELM/Pi9FiV0aYzduX5Pw9jUhrZmZmZuuAe5itLh6RbbY6SR8HnoyIOySNKyVXyBrdbKuWXummZsW7VmsSYfar70ht+/4SndP16H/16A91cD36bz3MzNZnbjCbmb16HwD2lbQPsBGwGanHeQtJg3MvcjFabCny7AJJg4HNScvdVYtIS430LtYkwuyhx18NwIUTN+4X0Tn7S5RQ16N/1cH16L/1MDNbn3lItvWIB2SbrRIRJ0TE9hExnBS068aIOBiYBeyfs5VHni1Fi90/54+cfmCOor0TMIIUmOd2YESOur1hPsaMdXBqZmZmZoZ7mK1Ocphss544Dpgm6VvA/wFtOb0NuCQH9VpMagATEfMkTScF81oOHBkRKwAkHUVaymUQcEFEzFunZ2JmZmbWwNxgth5xzC+zyiKiA+jIjx9gVZTrYp6XgE9V2b+VFGm7PH0maYkXMzMzM1vHPCTb6uIOZjMzMzMzazRuMJuZmZmZmZlV4Aaz9Ug47JeZmZmZmTUIN5itLh6RbWZmZmZmjcYNZjMzMzMzM7MK3GC2HnGUbDMzMzOztae9vZ3Ro0czYcIERo8eTXt7e19XyQq8rJTVx2GyzczMzMzWqvb2dlpaWmhra2PFihUMGjSI5uZmAJqamvq4dgbuYTYzMzMzM+sTra2ttLW1MX78eAYPHsz48eNpa2ujtbW1r6tmmRvM1iMekW1mZmZmtnZ0dnYyduzYLmljx46ls7Ozj2pk5dxgtrp4QLaZmZmZ2do1cuRIZs+e3SVt9uzZjBw5so9qZOXcYDYzMzMzM+sDLS0tNDc3M2vWLJYvX86sWbNobm6mpaWlr6tmmYN+WY84SraZmZmZ2dpRCuw1ZcoUOjs7GTlyJK2trQ741Y+4wWx1cZBsMzMzM7O1r6mpiaamJjo6Ohg3blxfV8fK9OqQbEkTJd0rab6k4ytsHyLp8rz9VknDC9tOyOn3StqruzIlXZrT50q6QNIGOX2cpCWS7sx/J/XmOa//3MVsZmZmtrZ5LV6z/qnXepglDQLOBj4KLABulzQjIu4uZGsGnomInSUdCJwOHCBpFHAgsAuwLfB7SW/N+1Qr81LgMznPZcDngZ/k5zdFxMd761wbgRz2y8zMzKxXeC1es/6rN3uYdwfmR8QDEfEyMA2YVJZnEnBRfnwFMEGScvq0iFgWEQ8C83N5VcuMiJmRAbcB2/fiuZmZmZmZrRVei9es/+rNBvN2wCOF5wtyWsU8EbEcWAJsVWPfbsvMQ7E/C1xbSH6fpLskXSNpl0qVlXS4pDmS5ixatKi+M2xADvplZmZmtnZ5LV6z/qs3G8yVxvCWN7eq5elpetE5wB8j4qb8/C/AjhGxKzAV+FWlykbEeRExJiLGDB06tFKWhuagX2ZmZma9w2vxmvVfvdlgXgDsUHi+PbCwWh5Jg4HNgcU19q1ZpqSTgaHA0aW0iHguIpbmxzOBDSRtvSYnZmZmZma2tngtXrP+qzeXlbodGCFpJ+BRUhCvg8ryzAAmAzcD+wM3RkRImgFcJukMUtCvEaR5yapWpqTPA3sBEyLildIBJG0DPJHL3Z10k+DpXjpnMzMzM7Me8Vq8Zv1Xr/Uw5znJRwHXAZ3A9IiYJ+lUSfvmbG3AVpLmk3qFj8/7zgOmA3eT5iIfGRErqpWZyzoXGAbcXLZ81P7AXEl3AT8CDsyBwczMzMzM+oWmpibmzp3LDTfcwNy5c91YbiBeUqx/680e5tIQ6JllaScVHr8EfKrKvq3AaqEBK5WZ0yueS0T8GPhxjypuZmZmtp7JS37OAR6NiI/nEXvTgNeTYr58NiJeljQEuBh4D2lU3gER8VAu4wTSsqArgC9HxHU5fSJwFjAI+GlEnJbTKx5jHZ2yWb/nJcX6v96cw2xmZmZm/cdXSCP0Sk4HzoyIEcAzpIYw+d9nImJn4MycD0mjSNPhdgEmAudIGpQb4mcDewOjgKact9YxzAwvKTYQuMFsPeLB7GZmZgOPpO2BjwE/zc8FfBi4Ime5CNgvP56Un5O3T8j5JwHTImJZRDwIzAd2z3/zI+KB3Hs8DZjUzTHMDC8pNhC4wWx18apSZmZmA9oPgWOBUmDUrYBnc3wYSCuRbJcfbwc8Aitj0izJ+Veml+1TLb3WMcwMLyk2EPTqHGZbf7hj2czMbGCS9HHgyYi4Q9K4UnKFrNHNtmrplTpgauWvVMfDgcMBhg0bRkdHR6VsDWHp0qUNff6N5pOf/CQHH3wwxxxzDDvttBNnnnkm3/ve92hubvbnoJ9wg9nMzMxs/fYBYF9J+wAbAZuRepy3kDQ49wBvDyzM+RcAOwALJA0GNgcWF9JLivtUSn+qxjG6iIjzgPMAxowZE+PGjVujEx7IOjo6aOTzbzTjxo1j1KhRtLa2rlxS7Ac/+IEDfvUjHpJtdfGQbDMzs4EpIk6IiO0jYjgpaNeNEXEwMIu0/CbAZODX+fGM/Jy8/ca8JOcM4EBJQ3L06xHAbcDtwAhJO0naMB9jRt6n2jHMLPOSYv2bG8xmZmZmjek44GhJ80nzjdtyehuwVU4/GjgeICLmAdOBu4FrgSMjYkXuPT4KuI4UhXt6zlvrGGZmA4KHZJuZmZk1iIjoADry4wdIEa7L87wEfKrK/q3AauvdRMRMYGaF9IrHMDMbKNzDbD0SDv9lZmZmZmYNwg1mq4s8idnMzMzMzBpMXQ1mSb+U9DFJbmA3qHDHspmZmZmZNZh6G8A/AQ4C7pN0mqS392KdzMzMzMzMzPpcXQ3miPh9Xn5gN+Ah4HpJf5Z0mKQNerOC1j94SLaZmZmZmTWauodYS9oKOBT4PPB/wFmkBvT1vVIzMzMzMzMzsz5U7xzmK4GbgNcBn4iIfSPi8oiYAmzSmxU0MzMzM1vftbe3M3r0aCZMmMDo0aNpb2/v6yqZGfWvw/zTvL7eSpKGRMSyiBjTC/UyMzMzM2sI7e3ttLS00NbWxooVKxg0aBDNzc0ANDU19XHtrLe1t7fT2tpKZ2cnI0eOpKWlxe97P1LvkOxvVUi7eW1WxAYGR8s260rSRpJuk3SXpHmSvpHTd5J0q6T7JF0uacOcPiQ/n5+3Dy+UdUJOv1fSXoX0iTltvqTj1/U5mplZ72ptbaWtrY3x48czePBgxo8fT1tbG62trX1dNetlpZslU6dO5brrrmPq1Km0tLR4hEE/UrPBLGkbSe8BXivp3ZJ2y3/jSMOzrUG4oWxW1TLgwxGxK/AuYKKkPYDTgTMjYgTwDNCc8zcDz0TEzsCZOR+SRgEHArsAE4FzJA2SNAg4G9gbGAU05bxmZrae6OzsZOzYsV3Sxo4dS2dnZx/VyNYV3yzp/7rrYd4L+D6wPXAG8IP8dzRwYu9Wzcys/4tkaX66Qf4L4MPAFTn9ImC//HhSfk7ePkGScvq0PNXlQWA+sHv+mx8RD0TEy8C0nNfMzNYTI0eOZPbs2V3SZs+ezciRI/uoRrau+GZJ/1ezwRwRF0XEeODQiBhf+Ns3Iq5cR3U0M+vXck/wncCTpJUD7geejYjlOcsCYLv8eDvgEYC8fQmwVTG9bJ9q6WZmtp5oaWmhubmZWbNmsXz5cmbNmkVzczMtLS19XTXrZb5Z0v/VDPol6TMR8XNguKSjy7dHxBm9VjMzswEiIlYA75K0BXAVUOl/udLEhkqrmkeN9Eo3NlebJCHpcOBwgGHDhtHR0dF9xbOvviO165cuXdqj/XqL69H/6tEf6uB69N962JorBXiaMmXKysBPra2tDvzUAEo3S0oB30o3Szwku//oLkr2xvlfLx3V4DyF2ax7EfGspA5gD2ALSYNzL/L2wMKcbQGwA7BA0mBgc2BxIb2kuE+19OKxzwPOAxgzZkyMGzeu7nofevzVAFw4cWN6sl9v6ejocD36WT36Qx1cj/5bD1s7mpqaaGpq8vvaYHyzpP/rbkj2/+Z/v1Hpr7vCu4vsujajxUq6NKfPlXSBpA1yuiT9KOf/q6Td6nlhzMzqIWlo7llG0muBjwCdwCxg/5xtMvDr/HhGfk7efmNERE4/MF8XdwJGALcBtwMjctTtDUmBwWb0/pmZmdm6NGXKFDbaaCPGjx/PRhttxJQpU/q6SraONDU1MXfuXG644Qbmzp3rxnI/U9eyUpK+K2kzSRtIukHSU5I+080+9UR2XZvRYi8F3g68A3gt8Pmcvjfph+cI0nDFn9RzzlaZe5rNVvNGYJakv5Iat9dHxG+B44CjJc0nzVFuy/nbgK1y+tHA8QARMQ+YDtwNXAscGRErcg/1UcB1pIb49JzXzMzWE1OmTOHcc8/l29/+Ntdccw3f/va3Offcc91oNusHuhuSXbJnRBwr6ZOkYYOfIvWe/LzGPisjuwJIKkV2vbuQZxJwSn58BfDj8mixwIP5h+XuOV/FMiNiZqlQSbeRhi2WjnFx7sG5RdIWkt4YEY/Vee6Gl5UyqyYi/gq8u0L6A6y6bhXTXyJdQyuV1QqsNmkpX99mrr6HmZmtD84//3xOP/10jj76aDo6Ojj66BQ66MQTT2Tq1Kl9XDuzxlZXDzNpmRSAfYD2iFhcxz71RHZd69Fi81Dsz5J6aOqtB5IOlzRH0pxFixbVcXqNqVJUIjMzMzN79ZYtW8YRRxzRJe2II45g2bJlfVQjMyupt8H8G0n3AGOAGyQNBV7qZp9qEV/rydPT9KJzgD9GxE09qAcRcV5EjImIMUOHDq2wi4GHZJuZmZmtbUOGDOHcc8/tknbuuecyZMiQPqqRmZXUNSQ7Io6XdDrwXESskPQCaahzLbUivpbnWSvRYiWdDAwFvtjDepiZmZmZ9YkvfOELHHfccQCMGjWKM844g+OOO261XmczW/fqncMMaV3R4blhW3JxjfwrI7sCj5KCeB1UlqcULfZmCtFiJc0ALpN0BrAtq6LFqlqZkj4P7AVMiIhXyo5xVJ7v/F5giecv91y4b9nMzMysV5TmKZ944oksW7aMIUOGcMQRR3j+slk/UFeDWdIlwFuAO4EVOTmo0WCOiOWSSpFdBwEXRMQ8SacCcyJiBila7CU5qNdiUgOYnK8ULXY5OVpsrstqZeZDngv8A7g5xQ3jyog4lRQoZx9gPvBP4LB6ztnMzMzMbF2ZOnUqU6dO9TrMZv1MvT3MY4BROdJ03SpFdo2IkwqP11q02IioeC65zkf2pN5mZmZmZutSe3s7ra2tdHZ2MnLkSFpaWrwer1k/UG+DeS6wDeChzA3Ky0qZmZmZ9Y729nZaWlpoa2tjxYoVDBo0iObmZgA3mu3/s3fncXJVZf7HP9/u7CGEVZaEJT8F7RDHYciAaNSEyBIRgiMIEQQkToSBGBkYDQaXYewRGBEhMijQGSKDAcZBDJKIGBIwsggCapYBIrIEkDWEELLQ6ef3xzkdqjvdSYX07ap0f9+vV7266tSte5+61XXrnnvOeY5VWLlZsncCFkm6XdLM5luRgZmZmZmZdQf19fU0NHcicpEAACAASURBVDQwatQoevTowahRo2hoaKC+foPOlmbWycptYf5WkUHY1sMtzWZmZmYda/HixYwYMaJF2YgRI1i8eHGFIjKzZmW1MEfEXcCTQM98/wHgoQLjMjMzMzPrFurq6pg/f36Lsvnz51NXV1ehiMysWVkVZkn/CPwU+FEuGgTcUlRQVn3csGxmZrZ1ktRH0u8k/UHSQkn/msuHSLpf0uOSbpTUK5f3zo+X5Of3LlnXebn8UUmHl5QfkcuWSJpcUt7mNqylKVOmMH78eObOnUtjYyNz585l/PjxTJkypdKhmXV75XbJPhM4ELgfICIel/SuwqIyMzMzs46yBjgkIt6Q1BOYL2k28M/ApRFxg6QfAuOBK/PfZRHxHkknABcBx0saSpoCdD9gd+DXkvbN27gCOBRYCjwgaWZELMqvbWsbVqI5sdfEiRPXZ8mur693wi+zKlBu0q81EbG2+YGkHrjR0czMzKzqRfJGftgz3wI4hNSDEGA6cEy+PzY/Jj8/WpJy+Q0RsSYi/gIsITWoHAgsiYgn8vniDcDY/Jr2tmGtjBs3jgULFjBnzhwWLFjgyrJZlSi3wnyXpK8BfSUdCvwPcGtxYVnVcbYvMzOzrZakWkmPAC8CdwB/Bl6LiMa8yFLSkDvy32cA8vPLgR1Ly1u9pr3yHTeyDTOzrUK5XbInk7rQ/An4IjALuKaooMzMzMys40TEOuBvJW0H/AxoK5tU89VxtfNce+VtNcBsbPkNSJoATADYZZddmDdvXluLdQtvvPFGt37/3Zk/++pUVoU5Ipok3QLcEhEvFRyTVbFwT3wzM7OtVkS8Jmke8EFgO0k9cgvwYOC5vNhSYA9gaR6GNxB4taS8Welr2ip/eSPbaB3XVcBVAMOHD4+RI0du4Tvdes2bN4/u/P67M3/21WmjXbKVfEvSy8D/AY9KeknSNzonPKsWanqLb/T4MQMaX6l0KGZmZrYZJO2cW5aR1Bf4OLAYmAscmxc7Bfh5vj8zPyY/f2dERC4/IWfRHgLsA/yONN3oPjkjdi9SYrCZ+TXtbcPMbKuwqRbmLwMfBv4+J3dA0v8DrpR0dkRcWnSAVh0GvTyfUT1+ycN/XQOMrnQ4ZmZmVr7dgOmSakmNJTdFxC8kLQJukPRt4GGgIS/fAFwnaQmpZfkEgIhYKOkmYBHQCJyZu3oj6SzgdqAWmBYRC/O6vtrONszMtgqbqjCfDBwaES83F0TEE5JOAn4FuMLcTdSuWwOA3CXbzMxsqxIRfwT2b6P8CVKG69blq4Hj2llXPVDfRvksUo6bsrZhZra12FSW7J6lleVmeRxzz2JCsmqknOByXdl54szMzMzMzLZum6owr32Hz1kXU9OUK8xyhdnMzMzMzLqHTVWYPyDp9TZuK4D3d0aAVh2UhijRpNoKR2JmZmbW9cyYMYNhw4YxevRohg0bxowZMyodkpmxiTHMEeHakWUeu2xmZmZWhBkzZjBlyhQaGhpYt24dtbW1jB8/HoBx48ZVODqz7m1TLcxmWfpXUbjibGZmZtaR6uvraWhoYNSoUfTo0YNRo0bR0NBAff0G+dXMrJO5wmxlaUIAiKYKR2JmZmbWtSxevJgRI0a0KBsxYgSLFy+uUERm1swVZiuPVOkIzMzMzLqkuro65s+f36Js/vz51NXVVSgiM2vmCrOVJdzCbGZmZlaIKVOmMH78eObOnUtjYyNz585l/PjxTJkypdKhmXV7niPIyhIew2xmZmZWiObEXhMnTmTx4sXU1dVRX1/vhF9mVaDQFmZJR0h6VNISSZPbeL63pBvz8/dL2rvkufNy+aOSDt/UOiWdlctC0k4l5SMlLZf0SL59o7h33HWFmluYXWE2MzMz62jjxo1jwYIFzJkzhwULFriybFYlCqswS6oFrgDGAEOBcZKGtlpsPLAsIt4DXApclF87FDgB2A84AvhPSbWbWOdvgY8DT7URzm8i4m/z7YKOfJ9mZmZmZlvK8zCbVaciW5gPBJZExBMRsRa4ARjbapmxwPR8/6fAaEnK5TdExJqI+AuwJK+v3XVGxMMR8WSB76dbU3jsspmZmVkRZsyYwaRJk1i5ciUAK1euZNKkSa40m1WBIivMg4BnSh4vzWVtLhMRjcByYMeNvLacdbblYEl/kDRb0n5tLSBpgqQHJT340ksvlbHKbsZjl83MzMwK8ZWvfIUePXowbdo0br/9dqZNm0aPHj34yle+UunQzLq9IivMbc1D1LrW1d4ym1u+MQ8Be0XEB4CpwC1tLRQRV0XE8IgYvvPOO29ild2Ps2ObmZmZFWPp0qVMnz6dUaNG0aNHD0aNGsX06dNZunRppUMz6/aKrDAvBfYoeTwYeK69ZST1AAYCr27kteWss4WIeD0i3sj3ZwE9S5OCWZncwmxmZmZmZt1MkdNKPQDsI2kI8CwpiddnWy0zEzgFuBc4FrgzIkLSTOAnkr4H7A7sA/yO1MK8qXW2IGlX4IW83gNJFwle6aD32G283cLsirOZmZlZRxo8eDDHHXcc22+/PU8//TR77rkny5YtY/DgwZUOzazbK6yFOY9JPgu4HVgM3BQRCyVdIOnovFgDsKOkJcA/A5PzaxcCNwGLgF8CZ0bEuvbWCSDpS5KWklqd/yjpmryNY4EFkv4AXA6cEOHm0s3mXWZmZmZWiGOOOYYVK1awatUqmpqaWLVqFStWrOCYY46pdGhm3V6RLczNXaBntSr7Rsn91cBx7by2HqgvZ525/HJShbh1+Q+AH2xu7NaS5182MzMzK8bcuXM5+uijmT17NgCvvfYaRx99NHPnzq1wZGZWaIXZuo7maaXklmYzMzOzDrVo0SJWrlzJ7NmzWbduHbW1tZx22mk89dRTlQ7NrNsrMumXdSHuxW5mZmZWjF69etGrVy9Gjx7NoYceyujRo9eXWdc3Y8YMhg0bxujRoxk2bJjn364yrjBbWdYn/WprYi+zbkzSHpLmSlosaaGkSbl8B0l3SHo8/90+l0vS5ZKWSPqjpL8rWdcpefnHJZ1SUn6ApD/l11wuyd9EM7MuZM2aNTz22GMcddRR/OxnP+Ooo47iscceY82aNZUOzQo2Y8YMpkyZwtSpU7n99tuZOnUqU6ZMcaW5irjCbGVZ3xXbDc1mrTUC50REHfBB4ExJQ0lJDOdExD7AnPwYYAwp8/8+wATgSkgVbOCbwEHAgcA3myvZeZkJJa87ohPel5mZdaKddtqJW2+9lU996lPceuut7LSTZ0HtDurr62loaGgxB3dDQwP19RukcrIKcYXZyvL2tFJmVioino+Ih/L9FaQM/oOAscD0vNh0oDnV6Vjgx5HcB2wnaTfgcOCOiHg1IpYBdwBH5Oe2jYh7c4b/H5esy8zMuoiXX36Z7bbbjpqaGrbbbjtefvnlSodknWDx4sWMGDGiRdmIESNYvHhxhSKy1pz0y8rkpmWzTZG0N7A/cD+wS0Q8D6lSLeldebFBwDMlL1uayzZWvrSN8tbbnkBqhWaXXXZh3rx5Zcd9zvsbAXjjjTc263VFcRzVF0c1xOA4qjcO6xg9evRg4MCBvPbaawwcOJAVK1bQ2NhY6bCsYHV1dcyfP59Ro0atL5s/fz51dXUVjMpKucJsZXF2bLONk7QN8L/AlyPi9Y0MM27riXgH5S0LIq4CrgIYPnx4jBw5soyok1Mn3wbAtUf0Z3NeV5R58+Y5jiqLoxpicBzVG4d1jMbGRpYvX05EsHz5cleWu4kpU6Ywfvx4GhoaWLduHXPnzmX8+PHukl1FXGG2srzdJdsVZ7PWJPUkVZavj4ibc/ELknbLrcu7AS/m8qXAHiUvHww8l8tHtiqfl8sHt7G8mZl1ITU1NSxbtgyAZcuWUVNTQ1OTh8R1dePGjQNg4sSJLF68mLq6Ourr69eXW+V5DLOVxy3MZm3KGasbgMUR8b2Sp2YCzZmuTwF+XlJ+cs6W/UFgee66fTtwmKTtc7Kvw4Db83MrJH0wb+vkknWZmVkXUFtbC8All1zC7NmzueSSS1qUW9c2btw4FixYwJw5c1iwYIEry1XGLcxWnvAVTrN2fBj4HPAnSY/ksq8BFwI3SRoPPA0cl5+bBXwCWAK8CXweICJelfRvwAN5uQsi4tV8/wzgWqAvMDvfzMzKImkPUsLAXYEm4KqIuCxn578R2Bt4EvhMRCzLF+cuIx2r3gRObU5umKe8Oz+v+tsRMT2XH8Dbx6lZwKSIiPa2UfBb3uo0NTXRo0cPzjnnnPVlPXv2dLdssyrgCrOVRe6KbdamiJhP+zOUj25j+QDObGdd04BpbZQ/CAzbgjDNrHtrnv7uIUkDgN9LugM4lTT93YWSJpOmv/sqLae/O4g0td1BJdPfDSeN0fq9pJm5Atw8/d19pArzEaSLe5Pb2YaV2H777Vm2bBm77rorL774Iu9617t44YUX2H777Tf9YjMrlLtkW1nkFmYzM7OtUoWnv2tvG1bi9ddfp1+/fvTp0weAPn360K9fP15//fUKR2ZmrjBbmdzCbGZmtrXb2PR3QBHT37W3DSvR2NhIv379AIicN6Zfv37ukm1WBdwl28riFmYzM7OtW6Wnv9tEbO94Lvmu4uCDD+bss8/mjTfeYJtttuHSSy9l5syZ3XJfdDdz5szhv//7v3n66afZc889Oemkkxg9eoNRXVYhrjBbmdLvnscym5mZbX0qOP1de9toYUvmku8KJHHbbbfxsY99jKFDh/LQQw9x2223IclzbXdxM2bM4Prrr2fatGmsW7eO2tpaxo8fz9ChQ50tu0q4S7aVJ3cPckuzmZnZ1qXC09+1tw0rMXToUPbff3/OPfdcxowZw7nnnsv+++/P0KFDKx2aFay+vp6GhgZGjRpFjx49GDVqFA0NDdTX11c6NMtcYbayiKb81y3MZmZmW5nm6e8OkfRIvn2CNP3doZIeBw7NjyFluX6CNP3d1cA/QZr+Dmie/u4BNpz+7pr8mj/z9vR37W3DSowaNYpHHnmE7373u8yePZvvfve7PPLII4waNarSoVnBFi9ezIgRI1qUjRgxgsWLF1coImvNXbKtLGpuYcYtzGZmZluTSk5/FxGvtLUNa2nu3Ln07du3xTzMAwYMYO7cuRWMyjpDXV0d8+fPb3FxZP78+dTV1VUwKivlCrOVZX0Lc7iF2czMzKwjLVy4EIAPfehDnH322Vx66aXcc88968ut65oyZQrHH388/fv356mnnmKvvfZi5cqVXHbZZZUOzTJ3ybbyuIXZzMzMrDC77747y5cv5/jjj2f58uXsvvvulQ7JOtlGMtdbBbnCbGVylmwzMzOzojz//POcdtpp3HbbbZx22mk8//zzlQ7JOkF9fT0TJkygf//+APTv358JEyY46VcVcZdsK0tzduwaZ8k2MzMz63C7774706ZNY/HixdTV1bH77rvz7LPPVjosK9iiRYt48803aWhoaDGt1JNPPlnp0CwrtIVZ0hGSHpW0RNLkNp7vLenG/Pz9kvYuee68XP6opMM3tU5JZ+WykLRTSbkkXZ6f+6OkvyvuHXddb7csu4XZzMzMrKM9++yzDBw4kBtvvJGBAwe6stxN9OrVi7POOqvFtFJnnXUWvXr1qnRolhXWwiypFriCNIXAUuABSTMjYlHJYuOBZRHxHkknABcBx0saCpwA7AfsDvxa0r75Ne2t87fAL4B5rUIZA+yTbwcBV+a/tjnCXbLNzMzMirDffvvx9NNPc88993DPPfcAKUv2nnvuWeHIrGhr165l6tSp7L///qxbt465c+cydepU1q5dW+nQLCuyhflAYElEPBERa4EbgLGtlhkLTM/3fwqMzhPejwVuiIg1EfEX0px+B25snRHxcEQ82UYcY4EfR3IfsJ2k3Tr0nXYDnofZzMzMrBijRo1i1apVXHLJJcyePZtLLrmEVatWeR7mbmDo0KGceOKJTJw4kcMPP5yJEydy4oknMnTo0EqHZlmRY5gHAc+UPF7Khi2765eJiEZJy4Edc/l9rV47KN/f1DrLiWMQ0CKTgqQJwATAV/Pa0DyGWR7DbGZmZtah5s6dyyc/+Um+9rWvsWbNGnr37s0nP/lJz8PcDUyZMoUpU6ZsMIbZSb+qR5EV5rbyordunmxvmfbK22oR31STZzlxEBFXAVcBDB8+3M2oG3CXbDMzM7MiLFq0iJUrVzJ79uz1labTTjuNp556qtKhWcHGjRsHwMSJE9cnfKuvr19fbpVXZJfspcAeJY8HA8+1t4ykHsBA4NWNvLacdb6TOGwTlMcw13geZjMzM7MO1atXLyZOnNgi8dPEiROd+KmbGDduHAsWLGDOnDksWLDAleUqU2SF+QFgH0lDJPUiJfGa2WqZmcAp+f6xwJ0REbn8hJxFewgpYdfvylxnazOBk3O27A8CyyPCE9tttuYu2W5hNjMzM+tIa9eu5cILL2TIkCEccsghDBkyhAsvvNCJn8yqQGFdsvOY5LOA24FaYFpELJR0AfBgRMwEGoDrJC0htSyfkF+7UNJNwCKgETgzItZBmj6q9Tpz+ZeArwC7An+UNCsivgDMAj5BShz2JvD5ot5zV7Z+DLNbmM3MzMw61KBBg3j11Vd57bXXiAieffZZevbsyaBBgzb9YjMrVJFjmImIWaQKa2nZN0rurwaOa+e19cAGo93bWmcuvxy4vI3yAM7c3NitbR7DbGZmZtax3nzzTVatWsUuu+zCiy++yI477sgLL7zAm2++WenQrBPMmDGD+vr69WOYp0yZ4m7ZVaTQCrN1HW5hNjMzMyvGq6++ysCBA+nTpw8RQZ8+fdh222159dVXKx2aFWzGjBltZskGXGmuEkWOYbYuRB7DbGZmZlaYI488kv79+yOJ/v37c+SRR1Y6JOsE9fX1NDQ0tEj41tDQ4GmlqohbmG2zuEu2mZmZWcf7yU9+sv7+woULWbhwYQWjsc6yePFiRowY0aJsxIgRLF68uEIRWWuuMFtZ3CXbzMzMrFi1tbXru+WuW7eu0uFYJ6irq+Mzn/kMs2fPZs2aNfTu3ZsxY8ZQV1dX6dAsc5dsK4u7ZJuZmZkVRxJ77LFHi7/W9Q0aNIhbbrmFfv36AdCvXz9uueUWZ0ivIq4wW1maK8puYTYzMzPreKeffnqLMcynn356pUOyTnDnnXfSv39/Bg4ciCQGDhxI//79ufPOOysdmmWuMFuZmivMbmE2MzMz62g/+tGPWLlyJQArV67kRz/6UYUjss7Q2NjImWee2eJiyZlnnkljY2OlQ7PMFWYry/oxzOEWZjMzM7OO1Lt3b5qamnjyySdb/O3du3elQ7NOcMUVV7By5UoigpUrV3LFFVdUOiQr4QqzlaW5K3aNW5jNzMzMOtTOO++8WeXWdUhi5cqVjBkzhpkzZzJmzBhWrlzpMexVxBVmK0+uJ3sMs5mZmVnHWrp0KZMnT2a//fajpqaG/fbbj8mTJ7N06dJKh2YFiwj69u3LNddcw1FHHcU111xD3759CSfarRquMFtZ1mfJdguzmZmZWYe7+OKLWbhwIU1NTSxcuJCLL7640iFZJ5k0aRL77rsvNTU17LvvvkyaNKnSIVkJz8Ns5Wkeu+yrXWZmZmYdrqmpib59+7J69Wr69OnDqlWrKh2SdYLBgwczffp0rr/++vVzcJ944okMHjy40qFZ5gqzlaW5ZdljmM3MzMyK0VxJdmW5+7j44ouZNGkSp512Gk8//TR77rknjY2NXHLJJZUOzTJ3ybbyeB5mMzMzM7MONW7cOC677DL69+8PQP/+/bnssssYN25chSOzZq4wW1maW5jlLtlmZmZbHUnTJL0oaUFJ2Q6S7pD0eP67fS6XpMslLZH0R0l/V/KaU/Lyj0s6paT8AEl/yq+5XDnFb3vbsLadccYZ3HrrrZxxxhmVDsU60bhx41iwYAFz5sxhwYIFrixXGVeYrSzr52F2C7OZmdnW6FrgiFZlk4E5EbEPMCc/BhgD7JNvE4ArIVV+gW8CBwEHAt8sqQBfmZdtft0Rm9iGteG6667jqKOO4rrrrqt0KGaWucJsZVnfwuwxzGZmZludiLgbeLVV8Vhger4/HTimpPzHkdwHbCdpN+Bw4I6IeDUilgF3AEfk57aNiHsjzYXz41bramsb1oY33nijxV8zqzxXmK08HsNsZmbW1ewSEc8D5L/vyuWDgGdKlluayzZWvrSN8o1tw0rkHuxll5tZ53GWbCvL+nmYPYbZrAVJ04BPAi9GxLBctgNwI7A38CTwmYhYlsf0XQZ8AngTODUiHsqvOQU4P6/22xExPZcfQOpK2ReYBUzKLThmZkVpq5YW76C8/A1KE0hdutlll12YN2/e5rx8q7fNNtuwYsWKNsu7277oji677DJuu+023nrrLXr27MmRRx7puZiriCvMVpbmMcw1bmE2a+1a4AekLojNmsfsXShpcn78VVqOCzyINObvoJJxgcNJJ5m/lzQzd3lsHhd4H6nCfAQwuxPel5l1fS9I2i0ins/dql/M5UuBPUqWGww8l8tHtiqfl8sHt7H8xrbRQkRcBVwFMHz48Bg5cmRbi3VZbVWWm8u7277obiZOnMgvfvELLrroIoYOHcqiRYv46le/yuDBg5k6dWqlwzPcJdvK5jHMZm2p4LhAM7MtNRNoznR9CvDzkvKTc7bsDwLLc3fq24HDJG2fk30dBtyen1sh6YO5J83JrdbV1jbMDLj66qs5/vjjmTZtGkceeSTTpk3j+OOP5+qrr650aJa5hdnK0tzCvJk9rMy6qxZj9iQVMS6whS3pznjO+xuBlGSmGrr+OY7qi6MaYnAcWxaHpBmk1uGdJC0l9Wq5ELhJ0njgaeC4vPgs0tCRJaThI58HiIhXJf0b8EBe7oKIaL5geAZvDx+Zzds9YdrbhrWhpqaGpqam9X+t61uzZg2//OUvGTBgAAArV67kl7/8JWvWrKlwZNas0AqzpCNI4/VqgWsi4sJWz/cmtZgcALwCHB8RT+bnzgPGA+uAL0XE7Rtbp6QhwA3ADsBDwOciYq2kU4H/AJ7Nm/1BRFxT1HvuqprHMNeED95mW6CwcYFb0p3x1Mm3AXDtEf2rouvfvHnzHEeVxVENMTiOLYsjItqb2HV0G8sGcGY765kGTGuj/EFgWBvlr7S1DWtbcyXZleXuZeXKlQwYMIDmFCUrV66scERWqrAu2ZJqgStIY/aGAuMkDW212HhgWUS8B7gUuCi/dihwArAfabzef0qq3cQ6LwIuzfP8LcvrbnZjRPxtvrmy/A68PQ+zW5jNyvBC7k7NZowLbK+8vXGBZmZm1gWsXr2aMWPGMHPmTMaMGcPq1asrHZKVKHIM84HAkoh4IiLWklp/x7ZapnSc30+B0Xnsy1jghohYExF/IXUJOrC9debXHJLXAZ7nr8Mp1qW/TvplVo7OGBdoZmZmXcDBBx/MtGnTOOqoo5g2bRoHH3xwpUOyEkVWmNsbk9fmMhHRCCwHdtzIa9sr3xF4La+jrW19WtIfJf1UUmkrznqSJkh6UNKDL730UvnvsptY38Ls2WzMWsjjAu8F3itpaR6ndyFwqKTHgUPzY0jjAp8gXQS8GvgnSOMCgeZxgQ+w4bjAa/Jr/owzZJuZmXUpixYtYrfddqOmpobddtuNRYsWVTokK1HkGOZyxt5t7ri9tir4mxrndyswIyLWSDqd1Pp8yAYLtxr/18b6ujW3MJu1rVLjAs3MzGzrt8MOO/Daa6/Rp08fmpqaWLVqFStWrGCHHXaodGiWFdnC3N6YvDaXkdQDGEianmVzx/m9TJqepUerciLilYhoTjN3NSnBmG2m5oqyxzCbmZmZmXWMfv36MWDAAPr27UtNTQ19+/ZlwIAB9OvXr9KhWVZkhfkBYB9JQyT1IiXxmtlqmdJxfscCd+YWmJnACZJ65+zX+wC/a2+d+TVz8zqgZMxgc+Kd7GhgcQe/z26huYW5hgB3yzYzMzMz22LPPfccU6dOpX///gD079+fqVOn8txzzvFZLQrrkh0RjZLOIiWzqQWmRcRCSRcAD0bETKABuE7SElLL8gn5tQsl3QQsAhqBMyNSja2tdeZNfhW4QdK3gYfzugG+JOnovJ5XgVOLes9dmUqnk4oAtdUL3szMzMzMylVXV8fNN9/MkiVLaGpqYsmSJdx8883U1dVVOjTLCp2HOSJmkZLclJZ9o+T+atqZwD4i6oH6ctaZy58gZdFuXX4ecN7mxm4ttZx/2S3MZmZmZmZbatCgQdxyyy2cccYZfOITn2DWrFlceeWVHHbYYZUOzbIiu2RbF9Ii2Vc48ZeZmZmZ2Za66667OPHEE7n77rsZO3Ysd999NyeeeCJ33XVXpUOzzBVmK0vLLtmuMJuZmZmZbak1a9YwenTLiTVGjx7NmjVr2nmFdbZCu2Rb19Gc9AtwhdnMzMzMrAP06NGD008/nYigqamJxx57jNNPP50ePVxNqxZuYbay1LhLtpmZmZlZh6qtrWXt2rX07t0bgN69e7N27Vpqa2srHJk1c4XZyuIu2WZmZmZmHau56/Ubb7zR4q+7ZFcPV5itLDW4S7aZmZmZWRG23377Fn+terjCbGVRNNEUee5lV5jNzMzMzDrM+eefz+zZszn//PMrHYq14tHkVpaaaOItaulNI4TnYTYzMzMz6yjnnHNOpUOwdriF2coimlhHTj7gFmYzMzMzM+sGXGG2siiaaGz+d3ELs5mZmZmZdQOuMFtZamii0S3MZmZmZmbWjbjCbJsWQQ1NrKVnetz0VmXjMTMzMzMz6wSuMNum5RblVdErPV63toLBmJmZmZmZdQ5XmG3TmtIczKvpnR43eiJ1sy7p+UcqHYGZmZlZVXGF2TYtUoX5TVeYzczMzMysG3GF2TYttzC7S7aZmZmZmXUnrjDbpuUW5lVuYTYzMzMzs27EFWbbtPVjmJtbmF1hNjMzMzOzrs8VZtu03KK8Ivrlx+6SbWZmZmZmpah6vwAAIABJREFUXZ8rzLZpjasBeJ1cYXYLs5mZmZVJ0hGSHpW0RNLkSsdjZrY5XGG2TXMLs5mZmb0DkmqBK4AxwFBgnKShlY3KzKx8rjDbprmF2czMzN6ZA4ElEfFERKwFbgDGVjgmM7OyFVph3lQXHEm9Jd2Yn79f0t4lz52Xyx+VdPim1ilpSF7H43mdvTa1DSvT2jcAeD52TI/feLGCwZiZmdlWZBDwTMnjpbnMzGyr0KOoFZd0wTmUdHB8QNLMiFhUsth4YFlEvEfSCcBFwPG5q84JwH7A7sCvJe2bX9PeOi8CLo2IGyT9MK/7yva2UdT77pLefAWApbEzf43t6f1/d9N3v8/QZ8CO0LMfSBUO0MzMzKpUWycJscFC0gRgAsAuu+zCvHnzCg6r8018amK7zw27dli7z71/+vvbLJ+619Qtjsk6hz/7rVthFWZKuuAASGruglNaYR4LfCvf/ynwA0nK5TdExBrgL5KW5PXR1jolLQYOAT6bl5me13tle9uIiA0O1u/EEd+/uyNW08LlK86mZ7yFSn5PtOFvS5vPt/xVeifPl5TluzvHSxBiTf/dmb/u7zn2uV/B1PQFXksPVrAN66ghEE2qZR21BFp/22CLJUU1BMofRZNq8jJtv7bjtL3e7lLtv6fnQVzX96RKh9EpTvj7PTj1w0MqHYaZWXe2FNij5PFg4LnWC0XEVcBVAMOHD4+RI0d2SnCd6U/8qd3ntJHGhw46ZbUK8me/dSuywtxWF5yD2lsmIholLQd2zOX3tXptc/edtta5I/BaRDS2sXx723i5NJDSK5t77rln2W9yrx37lb1suZY3DqY20tzHLb8mymXaoAwg8t12n2+jShibel7iz8CSPh/gO2NGUPeuw3ng/ttY9uzjaPVr9Fj7On0aX6eHmogIomkdtdG4vsrbemu0+uIHInJFuSa/5/S6kmjaPVa0fxDx4aV9NX3fxV4DO/7/thpt379XpUMwM+vuHgD2kTQEeJbUg/CzG3+JmVn1KLLCXE4XnPaWaa+8rTHXG1u+3Dg2uLLZxmva9KPPDS930c3w8wLWuWU+VHL/70d/umJx2JbbHxhX6SDMzKxbyI0VZwG3A7XAtIhYWOGwqs5+++3HwoUb7pb99tuvAtFYZ4qINluZ3bpcPYpM+lVOF5z1y0jqAQwEXt3Ia9srfxnYLq+j9bba24aZmZmZFSwiZkXEvhHx7oior3Q81WjKlCkMGTKEO++8kzvuuIM777yTIUOGMGXKlEqHZp0gIogI5s6du/6+VY8iW5jL6YIzEzgFuBc4FrgzIkLSTOAnkr5HSvq1D/A7UmvxBuvMr5mb13FDXufPN7aNgt6zmZmZmdlmGTcu9f2aOHEiixcvpq6ujvr6+vXlZlY5hVWY2+uCI+kC4MGImAk0ANflpF6vkirA5OVuIiUIawTOjEgDXDfSreerwA2Svg08nNdNe9swMzMzM6sW48aNY9y4ccybN4+umPTMbGtVZAszETELmNWq7Bsl91cDx7Xz2npgg247ba0zlz/B25m0S8vb3YaZmZmZmZlZe4ocw2xmZmZmZma21XKF2czMzMzMzKwNrjCbmZmZmZmZtcEVZjOzrYCkIyQ9KmmJpMlFbWfvybcVtWozMzOzrY4rzGZmVU5SLXAFMAYYCoyTNLSyUZmZmZl1fYVmyTYzsw5xILAkzwaApBuAsaSp9zpcaSvzkxceWcQmzMzMzLYKiohKx1B1JL0EPJUf7gS8XMFwmjmO6ooBHEdrW3sce0XEzh0dTEeQdCxwRER8IT/+HHBQRJxVsswEYEJ++F7g0Xewqa39M+xojqO6YgDH0VppHFV7DNtcrc7DuqNq+f+yztedP/uqPYa5hbkNpR+WpAcjYngl43Ec1ReD43AcnUxtlLW42hkRVwFXbdFGqmTfOY7qi6MaYnAc1RtHR6vWk+bO0lU/V9s0f/bVyWOYzcyq31Jgj5LHg4HnKhSLmZmZWbfhCrOZWfV7ANhH0hBJvYATgJkVjsnMzMysy3OX7E3boi6OHchxvK0aYgDH0ZrjKEhENEo6C7gdqAWmRcTCAjZVLfvOcbRUDXFUQwzgOFqrljisY/lz7b782VchJ/0yMzMzMzMza4O7ZJuZmZmZmZm1wRVmMzMzMzMzsza4wmxdjqS2puDpdI6jpWqJw6wjVMv/s+NoyXG0VC1xWHkkHSHpUUlLJE2udDzWeSRNk/SipAWVjsU25ApzF9YRP5SS3ivpYEk9JdV2RFydoCpOEKJ6EgR4f1ibtsbvd+lxrcKVAX+vSlRLHFTJ54LjsM2Uj8FXAGOAocA4SUMrG5V1omuBIyodhLXNWbI7SJ7qpSkiGiscxzakZG4rtvQERtI/AP8OPJtvD0q6NiJe74BQCyHpCOBzkh4G/hIR/5vL1ZkndJJGA8cA9wNPRcRvOmvbreLw/ihDZ++ParA1fr+z3SX9FegZEasl1UREU2cG4O9V1cZRLZ+L47B34kBgSUQ8ASDpBmAssKiiUVmniIi7Je1d6TisbW5h7gCSPgXcDFwv6fMVjONY0tyst0maKGn3LVhXT+B4YHxEjAZ+DuwBfEXSth0ScAeT9EFgKnAH8Arw7eYuTZ18kvIxoAH4M/Bu4FuSvthZ2y+Jw/uj/Zg+Kmls/u5WU+tYp9gav98Ako4EZgM/AK6RtEdENEnqtN8yf6+qNo5q+Vwch71Tg4BnSh4vzWVmVmFuYd5CkuqAC4CJpPlRL5G0E2me1Fc6MY69gK8B/5iLJgM7SvpZRPzhHa52W2Af4LfAz4CXgSOBz0r6URX+6G4D3BkR1wJIuhu4O19Mv6gT43gXcF1EfF9SP9IJyxWSmiLi6k6Mw/ujDZIOAX4CfB84UdKHgYsj4sXOjqXCtqrvdz7GXQycSaqcnQbMl/TRiHiqE1ua/b2qzjiq5XNxHPZOtdV9vqqOw2bdlVuYt1xf4K8RMS8i5gDjgBHAqZ0cR2/gLeCxiPg98BVgB2CspJ1h88b7RcRbwPeAf5D0kXwiOh94hPT+qtEKYFdJ/QEi4s/AR4GTJB3ViXGsAz6UY3gzIu4BzgKOzVf9O4v3Ryu5JfJo4IKIuJC0P/YGzpW0Y16my4/520q/3y8C9wILgKUR8a/A1cBdkgZ3Yrdsf6+qM45q+Vwch71TS0k9fZoNBp6rUCxmVsIV5i33KPCkpNGSekXEYlLr7qmSTu6sICLiMeA+4HhJ20bEX4BLgQOA4/Iym3ul8jfAr0hjoD4aEesi4ifA7sAHOi76d07S30j6uKSdIuJ+UnemXzY/n08SrqTgbk2S3iNpeN7mzcBfJf2iZJE/kCojuxUch/fHRuRK1R+AuryPlpF6ZewDnJeX6S5X9Kv++91KDTAQ+FzzZxQR3wamA9/M3cwL4e9V1cZRLZ+L47CO8ACwj6QhSnlxTiANszOzCnOX7HdA0kdIPzgvRcQcSQ8AJwMvSFoSEYslfQX4lKTrijoBl/Rx0on+6oj4L9LYvqOBFyXNi4i/SPoOcL6kaRGxenPWHymhzvWkLkHnSXofsAbYBXi+Q9/MOyDpGOA7wOPAq5KeiIizJE3P3c/G5grRDsBeBcbxaVK3/JclPQnMAU4Hvifptog4MiLekBSkpB4/kzo+6Yr3x0Zj2p3UQllD6oI8Ghgq6XcRsUzSF4B5kn4VEb8qKo5qUu3fbwBJI4HDgd+TLgj+M3CfpLci4gd5sZ8Ak3KreREx+HtVnXFUy+fiOKxDRESjpLOA20lD/KZFxMIKh2WdRNIMYCSwk6SlwDcjoqGyUVkzdZ/GlI4h6TBSwplZwM7AgIg4WtLXgPcCN5Ku6J5GOtE7ISLWFRDHSGAaqUvi35JahcYCxwL7k7otXk+anuBzwDERsfYdbqsX8GHgi8Bq4LKIeHgL38IWUepaOx34SUTMlnQwqTu8ImKipO8De5IqAH8LHFvED4+k3sANpDGw90o6Afgg8AJwOfAjUgVkCXAYcGRE/F8BcXh/tB/TkaQTyXtIQyi+nLd9InAJsDgiXpR0CTA7In5dZDzVphq/37B+rPn1pM9oH2AA6X/8UVL38UuBm0gXP/4JODQilndwDP5eVWcc1fK5OA4zs27ALcyb75PA1IiYCiDpekl3RcTHJJ1GqrSeC2wPfL6IynI2EvhxRHwnx/FD4Drgs6STlaOA20if8RnvtLIMkF87N1+ljk4cK7gxNaSWseZM4L8jtSB+SdJ5EfFlSe8nJTZ6LndRLyqOAbzdxe1/gZdI/yefjIiTJB0K9AS+FxGPFxiH90cr2jBR1D+S9s2BeZHTgHX5au4JwA+LjKcaVen3G1J33u9GxCWSdiVdCPwSKVnbwcC3gbpcfmpHV5Yzf6+qN45q+Vwch5lZVxcRvm3GDTgD+Eqrsv8Gbs73e5ImnN+14DiOJrW8DCgpuwq4u+TxXsAOld5nHfy++wN98v1RwB+Bj+fHPYCPAT8Gti04jlqgR77/D6RpeQ7Mj3sBp5AuaHh/dOL+aCOuvsA1wE683aPmW6RWyu1ICb+OJ1W+3tfZ8fm20c/uNNKJf/P/VT/gE/n/eZf82dYWcYzz96pq46iWz8Vx+Oabb751o5uTfpVB0u6S+uaH9wKn56vozU4BVkk6KCLeiohFEfHXAuJ4t6TtlKbu+D0podcRzc9HxATg+TzGjIh4KiJe7eg4KkXSP5Ba0WdLGksaq/WvwNmSDo2Ixoi4i3QyPazAOD5F6g5/s6QPAQ+Tkiedlv8H1kbEdGAvSfsVGIf3x6a1lSjqW8D/ABcCz0bEjcDXo+Cu4bZpkvaSNAwgIqaRxi3/l6TeEfEmKZFUDXBARKyKlKisQ49x/l5VbRzV8rk4DjOzbsZdsjchJ9I4B1gu6V5gBjCelMjkuIi4IyLWSVpH+mEqKo6jgH8jJ/QA/guYBMyQJOBXEfEa8FfS/ItdiqQhpArOiaSx4ocBfwPcTTqZ+76kqaST6UHAUwXFMZTUxfcLpLFgXwVuAZorW9+S9GPSd2s7Ure4IuLw/mg/ppFsOlHUdcCXIyeKaq5MW+XkC33/Trro9yLpM7oS+DwwTdLnI+I5Sc+R/tdmFRCDv1fVGUe1fC6Ow8ysG3KFeSMk7UY6WTiZ1CXtA8BU0hzHxwBXSbqO1FVwOClzaBFx7ETqMjqJNA7zIFJylUmkxB7fBT6aK86HkU4yu5ptSXOvPgA8IOlR0jjtj5D2xT+Rugr2BU6MiGcLimMXYEm+cn+XpMfydtcBN5PmUTyZlFzl1Ih4qaA4vD/aoJaJog7NsUwnJbWan5MW3UQaA3uApIFRzNhX2wxKc8WeTPpffVDSl0l5GgYBDaR5fX8r6U5SnoZDCgrF36vqjKNaPhfHYWbWHVW6T3g130hZsG8sebwjKeP0raTxwe8HTiJd6R1aYBy9ST+CO5WUHUbqGvf3wK6kTLH/Auxb6f1W4H64GZhY8vhAUpbwT+TH6oQYepG68x5dUjaG1NrVPKavJ1Dr/dG5+yNv60TgnHx/1xzLbaTK8x6kyvM1pNbn93dGTL6V9bn1Be4Cjm/1WV4CHJ4fH0vK3fDegmPx96rK4qiWz8Vx+ObbhjfSufEj+fZX4NmSx72AT5GS0r2v5DV7A6vyMotIY+17ljx/IDCP1Kvyofw7/v783LdabeMRUi6S5vtvkPKUPEIFcqf41jVvnlaqDdLbc0ZKmgm8GBFfyI93IlWSayPikoLj6Bm5y6ik/yJVmI8qef6zpCk9zo0tyIJdrXLX2ncBvSPiutw9fgTwQKRxpyhlJj+GNE1GIftA0odJJ/RExK8lfYmUjfTnEXFvXuZcUpe4wjKje3+UFdtppDlhPxRpTst+pJbKE0gXlF4H1gIDowuN799a5V4xNZGGtRxLSlzUEBEP5ef+BTgoIj5dYAwj8feqGuMYSXV8Lo7DrEySvgW8ERHfLSm7iTTrwZxIOUSQtDfwi4gYJqkWuIN07L9e0i7A/cBnI+KevPwI0jnwLW1to1UM80jnxQ8W8iatW3LSr1YkfRz4F0ln5S/xBKCXpHqAiHiZlInyI5IK69KuNN/zRZImS9orIj4PrJU0vWSx+0knMl2ua72kUaTx4nsCX5b0PdIVw78AB0o6Jy+6ClhRYByH5TgOA+olXUy6EgpwjKTj8v1XgZVAIVPyeH9sNKaKJ4qyzaeUqGgacE2upN1Pms/3aEkHRHIxsJ2kdxcUg79X1RlHtXwujsNsC0jahjQkajzpwvUG8kW33/H2dHVnAdObK8t5mfkRcUvB4Zq1yy3MJfJJ20xSq8ZJpG4g9wNPksYLN5IS0HwWOJV0FbfDxz9KOojU5e2fSNOovE5KsnITaeqoXqTprUaTWtOOzBX5LiG3LF0EPB8Rl0rqQ0py9hfgWuA9pAsZA0hXLU+MiIcLiKOWdEJ/d0Q05HGWvwbm5PhOJJ1YDiB19z0hIh4pIA7vj/ZjWp8oipRw6DrSPOSfJ/34fj4i1uYT/9ci4t+LjMfKI+kDwI2khGx7kY51XweWAx8FBpMSGDUC5wMf6eiLHP5eVW0c1fK5OA6zzdS69VfSScCoiBgv6R7grNyDaG/ebmHuA8wGJkXEHyXdTKow/3wj2/hH0hzwAMsiYlTJ8/NwC7N1sC7XMrmF3gtcGhHTJP2MVGn+e9JV9C+Txj5eR8oWemIRleVsJ+BHEXGjpFmkuRQPBz4dEZ+S9ENSpf5A4B+7UmUZUsZiSQ8BIyXtEhEvSBpPOjk4IyLOBmZJ2gd4NSJe6cjtS6lLfu4m+jtgB0l9ImKl0nRid5LG2nxV0tWk6b2ejgKmEgPvj43EVS2Jomzz7Qr8X0TMApD0FHAmKVfDD0mJDSeQWstOKqJHgL9X1RVHs0p/Lo7DrEONA76f79+QHz+UH79b0iPAPsBPI+KPba1A0v2kRHe/iohJufjS9rpkmxXBXbJbeg74lKT3RcQy0g/SUuDgiHg5Io7h7fGRfygwjuXA8ZKGR8QK0lX+24Fh+QfzdOBLwCHtHWC2RpL2kNRbac7re0lXzP9GUt/ctfZUUjbwTwFExOMFnRy8q+T+n4CPk1rBiIg38uMRkg6O1L33d0WcPHp/bFITaSqbd+dYvk/6Id4H2DsiJpJaZn4LjI6IxzohJivPI8Drkg6SVJMrzlcA3wH+X0TMBD4NnNLRrWX+XlVnHNXyuTgOs44haUfSheprJD1Jaug5PveaAPhzRPwtqZfEByUdncsXAn/XvJ6IOIjUA2lgZ8Vu1lq3rzBLeo+kvSVtGxG/JHXJ/rTSuOEVpBblD+UrukTEGxHxegFxDJV0gKTtI2I+KTvs6ZKGRcRqUrbAXchjQHKLwMqOjqNSJB1J6pIzldQtcC1pzNaXSePFd8snCXPyc0XG8TNJV0u6gHRi/zPgx5Lel09WXidldSzs++P9sdGYJKk2IlaR9s9HJTX/uP6ENA52AkBE/DQiZkbEo50Rm7UvV44/li8EvkAa6nI8sGf+PGeRWpePy62ea6KDExf5e1XVcVTL5+I4zDrGsaQs1XtFxN4RsQdpKMGI0oUi4nlgMnBeLroCOFXSh0oW69cZAZu1p1t3yZZ0FPCfpOx8fyfpc+TWKOBzkn4WEQsl/YI0brjIOK4mtSL/naSvkRKL9QX+WdKPIuJ+SfcCO+YWmcKTKXWGfKVxMGlqrrOAxcAppLHjB5PmlD4pL/ssqWvtNQXF8m7SCcrnSa2Xo0hjyY8iTYlwKXCfpCbS/8iFBcUxiNQyWun98R7S/jiNNO9qRfZHq5jGkrLAIuka0n4ZSkoUpYj4PXCxpDmS3h0Rfy46Jts0SWOAy4G5wK6SlkTEPysNL/kSqZL2G9L/Ve+Ijk2u4eNMdcZRZZ9LtRx3dwMuJg1RqFgcZh1gHBseN/6X9H97UavyW4BvSfpIRPxG0vGkxLeDSPlJXgYuKFn+bKXx0c2OiYgnOzR6sxLdtsIsaQfSD9Jn85fzX0g/UheTKtAjgJsk3U6ae/kjBcXRn9RqfEJEzJN0CvAPpBble4E3cxy/AI4DRnaVyjKsH5/1HOm9Pk6awutiSW8B95CmzXqINJb8AxTbtfZlYG5E3JVP5OaTkg79HPgk6SLG+4DhpERrHV4Zy93vXiZVHh6j8vtjdv6/rMj+KKWUKOoi3k4U9UNSN63fkBJFnS6pOVHUrsCyIuOx8iglkzoFuCDSdDjbArdLuioiJkj6OvBFSVNIyaQ+29ExVNlx5hXSWLyKHWeyih/v8oWRZ/LF4Iod76rluCtpd9K4/bsrGYfZOxV52qh8f2Qbz19e8nBYSXmQ/qebH99Hyt/T3ja+1dZz7W3XbEt16yzZSlM03R4RP8mPv0CqlJ4dEYskfYw0vuuhIisDkq4FlkTEt/PjsaSJ3m+MiNmShpNam5/pSlfQcgvm9sATpJb+30eaRqb5+fNI41HPiIg1BcYxAtgbuJ50onJbRFyYn6sBvgmsjojv5LL183R3cBxjSVlov0u6KvvHiKgveb6z9scH83ZeJmWgvi4ivpef67T90UZch5Pee3ML8yd4O1HU/bRMFHVxOFNs1ZD0VeC5iLiupOwe4LcR8S+StiedPP0lIpZ28Lar5TizHymh419JQ31uirczyXbmcaZajndHkcYu/oC0P/4YJVnsO/FzqZbj7uHAv5Ky/55LOif4t86Ow8zM2hAR3eoGbJf/1pASeH0dGFTy/GTg90D/guOoKbk/CvgP4KCSstNI48a2rfQ+K+j9N7dg3EU6YTqaNKbxvJJl9iZNo6WiPgNgG1KCiUdJXX13IyW+mVSy3OHAlQXvj4+Rpg47PD/eE3gK+OfO2h95G0fnz+W/8+cyFlgA/FNn7o92YtuFNCfsQc3fH9K0awtJifkAegO9Ojs239r8vPYtuX9S/j/as6RsJ+BmYGiBMVT8OJO3MSbHMZOUwf2Q/H97VskyhX6vqux4dxhpvHTz8W5v4Gngq538uVTLcbd5fzwNfIN0geepzt4fvvnmm2++tX3rVl2yc0vtPEmHRsS9ku4gdcE+UdJPImJpRFwoqQ7oDxSSVEvSaOBkSUuAh0ldwD8BHJmv5t8XaWqrj5NOaDo8yVglKSVy+C4wLiIelnQVaYqsD5HGzNWSph8YQcqUuB0FdK+N1LX9jdzTYB3wGdKJymjSdESNEXEF6TPYV9KASInginAAcE1E3C5pT9KJ7fnAf0paTUrucjAF7g+ljJbNwxQWSPoxsIo0xvTa1GuTK+mc/dEc00FAH2BlpKmjniQlinpB0jMRMSuPxTxO0n3hlpeqIOmTpKEkMyPihIj4b0nvJX2vPhwRT0fEy5LWkrL/FhFDVRxnJI0ELiNNj/U7SbeSekGcTNpHNaSxxIV+r6rleJc/l+uAo/L+2Ik0G8UxwG25+/EvSJ9TYZ9LVg3H3Y+Tej6MJQ0X+CWp9X806XylJ53wf2pmZu3rVhVm0olZLXC5pHMjjd36OimRwDZKc4G+RRqvXEgX09yN9D9IJ0h9SOP1fp8ffwk4NndVW0Ial1nUXM+VdmG83WV2CnBtRDyXTy7PJ41TPRD4fKQpvorUSGpZaCB1hxtManU5XtKBOY7PFFw5bOTtxHI3kKY4+3OO4zDSHOEfotj90Ujq+v8+Sc+Qvgc7khLP3AeMB/6G1CpT9P6oeKIoe2eU8jKcRcro+yFJMyJiXER8PV90uVXSf5JamD9ASuhSlGo4zrwAfDFXDnclVdK+TmrpvYmUGGcY6ftW+PeKyh/vXiH9zu6WL9L9T45pISmB1QGkrsfDKf74Xw3H3Vrg5EgJRrcj7YdPRMTUPCzsfFIX7QMKjsPMzNrRrcYw5yu1k0gttpNJ3eQeJ508fJR0BXdn4JtRwPzG+Ur6lcAPI2KOpF1IFeWrIuLXkt5FOlH4LOlixhVFxFFpuWWnf0S8nu/vBtxKOkl4XtJewLN5mcIvGDS3UObeBeeQxrH9W0RcIKkXqVv8ywXHMIyUPfJh0rj6/5K0LymD7X0R8XOlKccKPVmSdCxpaoe3gFl5HxxGmot1JvAgMCAiXio4jlpSK8ttUZIoCvhTvJ0o6r2kStcepFbxIudGt82glLzoddJFwR8Cb0XEuPzcp0hJ2Q4Avh8RCwqKoaqOMzmmKaTf3W9L+kfg/aTfgGeAbYo+zuQYquF49wHSBa9epHG7DcAXSBdQLoyIZzrpeFcVx90cS01ENEk6ArgWGJN7RvSJiNWStouI14qOw8zMNtRt5mFuPnkijR37BXA2cCep2/XqiPhxREwATiyqkppPQqYCf8g/ji+QKuwH5udfjIjfRsSZwMSuWFkGiIh18fZc1gJeA17NJ7EnAV8DenbWSSyp2/F78wns6cC3gYMknR4RazvjJDZXGs4ljc8dksseIyWdG5gXK/xkKSJ+Sqoc/4aUjZWI+BXpQs6OEbG66Mpy3uY60kls8+PXI+JgYJik/4iUDGciUE8af+jKchWJiOcizVn/MvBFoJekGfnpx0gXY75QVGU5x1Btxxkioj5ycseIuBrYl1RBXd0Zx5msGo53fyCNL/9ORFwdEU0RcRUpCdjOebHOON5VxXE3b7cp//0laazymHze0pjLXVk2M6uQ7tQlOyLiNUk/J/0QPkTqCvUKaRzXX/NCbxax8eZMoxFxd/Pj/NSqvP3mrKErI+LOiFhbRBzVJiIaSePqnpH0HVI3uFMjYlUnxvBc7oL8deDMiLhV0ihSt/jONJuUofZbeXgApBaXf89xdkp3kIhYJulO4DN5jGkf0jROhVdKJe0bb0+T8iwwWdJdEfF0LjsauErS0IhYRKrYWxWLiFckfRH4D0mPko67Izs5hoofZ5p/A0oef5pUOXy2s2Lg/7d3x6pRBVEYx7+TEKwsBAsrH8D0FiJo4ZYGBMMSsDD4BlpqpZWiL6BgIWKhSbGPYGUhFgYVRcTYpREsLIIGj8WZkXW5uih778ze/H/dhoQcdrOT++2dOUf1rHfp/fsmP04fwFf2AAACk0lEQVTPx2Gl56Or9U6VrLsTXio+1L+Z/nYBAAXti8CctzqlhwuSHilC81CxHfuhxbD0VsKy9GsO6GK6c6b0+78oLlIOmdlAcQGz1lYNNUofHCwpzu8tKeZKvi9Qyj1JI3d/kR4/9Y7nXacLowdm9krSeUXH53Vveb7xHzxT3FW+Kmk31bHd5i+soVEU2pFety3FMZiBz3h01DQ1rDM5eJnZAUXX8MuShu6+02UdSfH1LkuvzbriTu9q2nnVmcrW3VzTppkNFcdNtkvVAQAIvT/DPB6WzeyaYrTHqqQNdx+lr3dxVmq8juuKWY8bKSQ8SXVdanOLYs3M7KKk5+7+unAdncwUnhdmdlCxTrTaqT01itpUjBk6oWjilc+83lDcWc6Noi4ozqF+bLMmzI7FnOXHkq6UPGpSwzqTemkMJH1w93el6ki1FF/vUmA+JWnH3d+WrKUGNbwmAIDf9TowT4TUW4rxEKcVH/b/yHd82/4H1VDHcUln3H3Pomv2fUkn3b3rLcDV4CIBNTSKQnty86LCNbDOAACAf9LbwDwRUm9LWpa04u7fJ7ZGl6jj7Pi5JDM7UmhbHlAli3EzdyV9c/c1M1uW9NXdP035UQAAAGBmetsleyyk3pF0TBFSOw3Lf6ljz8wWzWwhfQ9hGRjj7p8V3ZV3U6OokaTO3rcAAACA1OPALElmdlQxq3Ulh9Quw/K0Oko1WQHmgcd4my1Fg7xzXTeKAgAAAHq7JTvLZ9ZKheXa6gDmRS2NogAAALB/9T4wA5hfNTSKAgAAwP5FYAYAAAAAoEGvzzADAAAAAPC/CMwAAAAAADQgMAMAAAAA0IDADAAAAABAAwIzAAAAAAANCMwAAAAAADQgMAMAAAAA0OAnRKDZbXn/KXEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1080x360 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "attr = \"credit_card_balance_AMT_DRAWINGS_CURRENT_min\"\n",
    "display(pdf_eval01.loc[attr].to_frame().T)\n",
    "plot_distribution_over_target(pdf_train, attr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Correlation\n",
    "\n",
    "Tương quan càng cao càng tốt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>TARGET</th>\n",
       "      <th>baseline_extend_EXT_SOURCE_2</th>\n",
       "      <th>credit_card_balance_AMT_DRAWINGS_CURRENT_min</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>TARGET</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>-0.1602</td>\n",
       "      <td>0.0158</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>baseline_extend_EXT_SOURCE_2</th>\n",
       "      <td>-0.1602</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>-0.0081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>credit_card_balance_AMT_DRAWINGS_CURRENT_min</th>\n",
       "      <td>0.0158</td>\n",
       "      <td>-0.0081</td>\n",
       "      <td>1.0000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              TARGET  baseline_extend_EXT_SOURCE_2  credit_card_balance_AMT_DRAWINGS_CURRENT_min\n",
       "TARGET                                        1.0000 -0.1602                        0.0158                                      \n",
       "baseline_extend_EXT_SOURCE_2                 -0.1602  1.0000                       -0.0081                                      \n",
       "credit_card_balance_AMT_DRAWINGS_CURRENT_min  0.0158 -0.0081                        1.0000                                      "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pdf_train[[\"TARGET\", \"baseline_extend_EXT_SOURCE_2\", \"credit_card_balance_AMT_DRAWINGS_CURRENT_min\"]].corr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Coverage\n",
    "\n",
    "Null càng ít càng tốt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0022217305529435624"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pdf_train[\"baseline_extend_EXT_SOURCE_2\"].isna().mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.717600531002821"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pdf_train[\"credit_card_balance_AMT_DRAWINGS_CURRENT_min\"].isna().mean()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}