Bank_failure_rand_forests_m2_ex2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Modeling of bank failures by FDIC \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "In this assignment you will be using:\n",
    "- Decision Trees\n",
    "- Random Forests\n",
    "- Boosted Trees\n",
    "\n",
    "All in the context of classification problem as applied to bank defaults data set.\n",
    "Let's get started!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## About iPython Notebooks ##\n",
    "\n",
    "iPython Notebooks are interactive coding environments embedded in a webpage. You will be using iPython notebooks in this class. You only need to write code between the ### START CODE HERE ### and ### END CODE HERE ### comments. After writing your code, you can run the cell by either pressing \"SHIFT\"+\"ENTER\" or by clicking on \"Run Cell\" (denoted by a play symbol) in the upper bar of the notebook. \n",
    "\n",
    "We will often specify \"(≈ X lines of code)\" in the comments to tell you about how much code you need to write. It is just a rough estimate, so don't feel bad if your code is longer or shorter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import time\n",
    "\n",
    "import os\n",
    "import functools\n",
    "import math\n",
    "import random\n",
    "import sys, getopt\n",
    "\n",
    "sys.path.append(\"..\")\n",
    "import grading\n",
    "\n",
    "try:\n",
    "    import matplotlib.pyplot as plt\n",
    "    %matplotlib inline\n",
    "except:\n",
    "    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### ONLY FOR GRADING. DO NOT EDIT ###\n",
    "submissions=dict()\n",
    "assignment_key=\"HnqO7_GcEeeQcBJXIhP2bA\" \n",
    "all_parts=[\"Pb9kd\", \"ZdjyW\", \"IfVpy\",\"Tifr3\",\"X8djk\"]\n",
    "### ONLY FOR GRADING. DO NOT EDIT ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# COURSERA_TOKEN = # the key provided to the Student under his/her email on submission page\n",
    "# COURSERA_EMAIL = # the email\n",
    "COURSERA_TOKEN=\"xJXNdGTU5VBJ16zB\"\n",
    "COURSERA_EMAIL=\"819563421@qq.com\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# common cell - share this across notebooks\n",
    "state_cols = ['log_TA','NI_to_TA', 'Equity_to_TA', 'NPL_to_TL', 'REO_to_TA', \n",
    "              'ALLL_to_TL', 'core_deposits_to_TA', 'brokered_deposits_to_TA', \n",
    "              'liquid_assets_to_TA', 'loss_provision_to_TL', \n",
    "              'ROA', \n",
    "              'NIM', 'assets_growth']\n",
    "\n",
    "# Macro Economic Variables (MEVs)\n",
    "all_MEVs = np.array(['term_spread',\n",
    "                    'stock_mkt_growth',\n",
    "                    'real_gdp_growth',\n",
    "                    'unemployment_rate_change',\n",
    "                    'treasury_yield_3m',\n",
    "                    'bbb_spread',\n",
    "                    'bbb_spread_change'])\n",
    "\n",
    "MEV_cols = all_MEVs.tolist()\n",
    "\n",
    "next_state_cols = ['log_TA_plus_1Q','NI_to_TA_plus_1Q', 'Equity_to_TA_plus_1Q', 'NPL_to_TL_plus_1Q', 'REO_to_TA_plus_1Q', \n",
    "                   'ALLL_to_TL_plus_1Q', 'core_deposits_to_TA_plus_1Q', 'brokered_deposits_to_TA_plus_1Q', \n",
    "                   'liquid_assets_to_TA_plus_1Q', 'loss_provision_to_TL_plus_1Q', \n",
    "                   'ROA_plus_1Q', \n",
    "                   'NIM_plus_1Q', \n",
    "                   'assets_growth_plus_1Q', \n",
    "                   'FDIC_assessment_base_plus_1Q_n']\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Opening /home/jovyan/work/readonly/df_train_FDIC_defaults_1Y.h5 in read-only mode\n",
      "Opening /home/jovyan/work/readonly/df_test_FDIC_defaults_1Y.h5 in read-only mode\n",
      "Opening /home/jovyan/work/readonly/data_adj_FDIC_small.h5 in read-only mode\n",
      "Opening /home/jovyan/work/readonly/df_FDIC_learn.h5 in read-only mode\n",
      "Number of unique dates on which defaulted within 1-st year 39\n",
      "Number of unique banks defaulted within 1-st year 472\n"
     ]
    }
   ],
   "source": [
    "df_train = pd.read_hdf('/home/jovyan/work/readonly/df_train_FDIC_defaults_1Y.h5', key='df')\n",
    "df_test = pd.read_hdf('/home/jovyan/work/readonly/df_test_FDIC_defaults_1Y.h5', key='df')\n",
    "df_data = pd.read_hdf('/home/jovyan/work/readonly/data_adj_FDIC_small.h5', key='df')\n",
    "df_closure_learn = pd.read_hdf('/home/jovyan/work/readonly/df_FDIC_learn.h5', key='df')\n",
    "\n",
    "df_all_defaulters_in_1Y = df_closure_learn[df_closure_learn.defaulter == 1].reset_index().groupby('IDRSSD').apply(lambda g: g.iloc[len(g)-3])\n",
    "selected_dates = df_all_defaulters_in_1Y.date.unique()\n",
    "defaulted_banks = df_all_defaulters_in_1Y['IDRSSD'].unique()\n",
    "print('Number of unique dates on which defaulted within 1-st year %d'  % len(selected_dates))\n",
    "print('Number of unique banks defaulted within 1-st year %d' % len(defaulted_banks))\n",
    "\n",
    "# failure dates\n",
    "# df_data[df_data['Failure / Assistance '].notnull()].date.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# function for a flexible way to make the train and test sets\n",
    "def make_train_and_test(df_in, perc_train=66.0, split_by_IDRSSD=True):\n",
    "    \n",
    "    reset_index_flag=False\n",
    "    df = df_in.copy()\n",
    "    if 'IDRSSD' in df.index.names:\n",
    "        reset_index_flag = True\n",
    "        df.reset_index(inplace=True)\n",
    "    \n",
    "    len_df = len(df)\n",
    "    len_df_train = int(np.floor(0.01*perc_train*len_df))\n",
    "    \n",
    "    if split_by_IDRSSD == True:\n",
    "        # split by names\n",
    "        unique_IDRSSD = df.IDRSSD.unique()\n",
    "        num_unique_IDRSSD = len(unique_IDRSSD)\n",
    "        num_IDRSSD_train = int(np.floor(0.01*perc_train*num_unique_IDRSSD))\n",
    "        \n",
    "        # re-shuffle the list of IDRSSD\n",
    "        np.random.shuffle(unique_IDRSSD)\n",
    "        IDRSSD_train = unique_IDRSSD[0:num_IDRSSD_train]\n",
    "        IDRSSD_test = unique_IDRSSD[num_IDRSSD_train:]\n",
    "        \n",
    "        df_train = df[np.in1d(df.IDRSSD, IDRSSD_train)].copy()\n",
    "        df_test = df[np.in1d(df.IDRSSD, IDRSSD_test)].copy()\n",
    "        \n",
    "    elif split_by_IDRSSD == False:\n",
    "        # split by rows\n",
    "    \n",
    "        idx = np.arange(len_df)\n",
    "        np.random.shuffle(idx)\n",
    "        df_train = df.ix[idx[0:len_df_train]]\n",
    "        df_test = df.ix[idx[len_df_train:]]\n",
    "\n",
    "    return df_train, df_test   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize binary classification data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f9bb7439128>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAELCAYAAAA2mZrgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9bb74a0860>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_test.plot(x=state_cols[0], y='defaulter', kind='scatter')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmoAAAJdCAYAAACCrVtqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9ba8249898>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot 4 scatter plots together\n",
    "\n",
    "# log_TA / NI_to_TA\n",
    "# log_TA / NPL_to_TL\n",
    "# log_TA / Equity_to_TA\n",
    "# log_TA / ROA\n",
    "\n",
    "first_indx = [0, 0, 0, 0]\n",
    "second_indx = [1, 3, 2, 10]\n",
    "\n",
    "X_train = df_train[state_cols].values\n",
    "y_train = df_train.defaulter.values # .reshape(-1,1)\n",
    "\n",
    "num_plots = 4\n",
    "if num_plots % 2 == 0:\n",
    "    f, axs = plt.subplots(num_plots // 2, 2)\n",
    "else:\n",
    "    f, axs = plt.subplots(num_plots// 2 + 1, 2)\n",
    "    \n",
    "f.subplots_adjust(hspace=.3)\n",
    "f.set_figheight(10.0)\n",
    "f.set_figwidth(10.0)\n",
    "    \n",
    "for i in range(num_plots):\n",
    "    if i % 2 == 0:\n",
    "        first_idx = i // 2\n",
    "        second_idx = 0\n",
    "    else:\n",
    "        first_idx = i // 2\n",
    "        second_idx = 1\n",
    "        \n",
    "    axs[first_idx,second_idx].plot(X_train[y_train == 1.0, first_indx[i]], \n",
    "                                   X_train[y_train == 1.0, second_indx[i]], 'r^', label=\"Failed\")\n",
    "    axs[first_idx,second_idx].plot(X_train[y_train == 0.0, first_indx[i]], \n",
    "                                   X_train[y_train == 0.0, second_indx[i]], 'go', label=\"Non-failed\") \n",
    "    \n",
    "    axs[first_idx, second_idx].legend()\n",
    "    axs[first_idx, second_idx].set_xlabel('%s' % state_cols[first_indx[i]])\n",
    "    axs[first_idx, second_idx].set_ylabel('%s' % state_cols[second_indx[i]])\n",
    "    axs[first_idx, second_idx].set_title('Failed banks vs non-failed banks')\n",
    "    axs[first_idx, second_idx].grid(True) \n",
    "    \n",
    "if num_plots % 2 != 0:\n",
    "    f.delaxes(axs[i // 2, 1])\n",
    "    \n",
    "# plt.savefig('Failed_vs_nonfailed_rr_plot.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['log_TA',\n",
       " 'NI_to_TA',\n",
       " 'Equity_to_TA',\n",
       " 'NPL_to_TL',\n",
       " 'REO_to_TA',\n",
       " 'ALLL_to_TL',\n",
       " 'core_deposits_to_TA',\n",
       " 'brokered_deposits_to_TA',\n",
       " 'liquid_assets_to_TA',\n",
       " 'loss_provision_to_TL',\n",
       " 'ROA',\n",
       " 'NIM',\n",
       " 'assets_growth']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "state_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['log_TA', 'NI_to_TA', 'Equity_to_TA', 'NPL_to_TL', 'REO_to_TA', 'ALLL_to_TL', 'core_deposits_to_TA', 'brokered_deposits_to_TA', 'liquid_assets_to_TA', 'loss_provision_to_TL', 'NIM', 'assets_growth']\n",
      "Len state_cols =  12\n"
     ]
    }
   ],
   "source": [
    "# Column 'ROA' is redundant, it is the same as NI_to_TA, so remove it\n",
    "state_cols = [c for c in state_cols if c != 'ROA']\n",
    "\n",
    "print(state_cols)\n",
    "print('Len state_cols = ', len(state_cols)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# make the train and test datasets\n",
    "choice = 0  # selection of predictors. Add tangible equity and assessment base as predictors\n",
    "predict_within_1Y = False # True\n",
    "\n",
    "if choice == -1: # only state cols\n",
    "    cols = state_cols \n",
    "elif choice == 0:  # original variables\n",
    "    cols = state_cols + MEV_cols \n",
    "\n",
    "trX = df_train[cols].values\n",
    "teX = df_test[cols].values\n",
    "num_features = len(cols)    \n",
    "\n",
    "if predict_within_1Y == True:\n",
    "    trY = df_train[['default_within_1Y','no_default_within_1Y']].values\n",
    "    teY = df_test[['default_within_1Y','no_default_within_1Y']].values\n",
    "else:\n",
    "    trY = df_train[['defaulter','non_defaulter']].values\n",
    "    teY = df_test[['defaulter','non_defaulter']].values\n",
    "\n",
    "num_classes = 2\n",
    "num_components = len(cols)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>term_spread</th>\n",
       "      <th>stock_mkt_growth</th>\n",
       "      <th>real_gdp_growth</th>\n",
       "      <th>unemployment_rate_change</th>\n",
       "      <th>treasury_yield_3m</th>\n",
       "      <th>bbb_spread</th>\n",
       "      <th>bbb_spread_change</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>term_spread</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.002993</td>\n",
       "      <td>-0.145941</td>\n",
       "      <td>0.299972</td>\n",
       "      <td>-0.633991</td>\n",
       "      <td>0.392349</td>\n",
       "      <td>-0.465767</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>stock_mkt_growth</th>\n",
       "      <td>0.002993</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.148941</td>\n",
       "      <td>0.461947</td>\n",
       "      <td>-0.081915</td>\n",
       "      <td>0.417379</td>\n",
       "      <td>-0.762702</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>real_gdp_growth</th>\n",
       "      <td>-0.145941</td>\n",
       "      <td>-0.148941</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.825802</td>\n",
       "      <td>0.041596</td>\n",
       "      <td>-0.820518</td>\n",
       "      <td>0.385007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unemployment_rate_change</th>\n",
       "      <td>0.299972</td>\n",
       "      <td>0.461947</td>\n",
       "      <td>-0.825802</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.034355</td>\n",
       "      <td>0.881223</td>\n",
       "      <td>-0.657093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>treasury_yield_3m</th>\n",
       "      <td>-0.633991</td>\n",
       "      <td>-0.081915</td>\n",
       "      <td>0.041596</td>\n",
       "      <td>0.034355</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.272072</td>\n",
       "      <td>0.290414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bbb_spread</th>\n",
       "      <td>0.392349</td>\n",
       "      <td>0.417379</td>\n",
       "      <td>-0.820518</td>\n",
       "      <td>0.881223</td>\n",
       "      <td>-0.272072</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.716249</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bbb_spread_change</th>\n",
       "      <td>-0.465767</td>\n",
       "      <td>-0.762702</td>\n",
       "      <td>0.385007</td>\n",
       "      <td>-0.657093</td>\n",
       "      <td>0.290414</td>\n",
       "      <td>-0.716249</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          term_spread  stock_mkt_growth  real_gdp_growth  \\\n",
       "term_spread                  1.000000          0.002993        -0.145941   \n",
       "stock_mkt_growth             0.002993          1.000000        -0.148941   \n",
       "real_gdp_growth             -0.145941         -0.148941         1.000000   \n",
       "unemployment_rate_change     0.299972          0.461947        -0.825802   \n",
       "treasury_yield_3m           -0.633991         -0.081915         0.041596   \n",
       "bbb_spread                   0.392349          0.417379        -0.820518   \n",
       "bbb_spread_change           -0.465767         -0.762702         0.385007   \n",
       "\n",
       "                          unemployment_rate_change  treasury_yield_3m  \\\n",
       "term_spread                               0.299972          -0.633991   \n",
       "stock_mkt_growth                          0.461947          -0.081915   \n",
       "real_gdp_growth                          -0.825802           0.041596   \n",
       "unemployment_rate_change                  1.000000           0.034355   \n",
       "treasury_yield_3m                         0.034355           1.000000   \n",
       "bbb_spread                                0.881223          -0.272072   \n",
       "bbb_spread_change                        -0.657093           0.290414   \n",
       "\n",
       "                          bbb_spread  bbb_spread_change  \n",
       "term_spread                 0.392349          -0.465767  \n",
       "stock_mkt_growth            0.417379          -0.762702  \n",
       "real_gdp_growth            -0.820518           0.385007  \n",
       "unemployment_rate_change    0.881223          -0.657093  \n",
       "treasury_yield_3m          -0.272072           0.290414  \n",
       "bbb_spread                  1.000000          -0.716249  \n",
       "bbb_spread_change          -0.716249           1.000000  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# look at correlations\n",
    "df_train[MEV_cols].corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "161.0\n",
      "161.0\n",
      "num_components: 19\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['log_TA',\n",
       " 'NI_to_TA',\n",
       " 'Equity_to_TA',\n",
       " 'NPL_to_TL',\n",
       " 'REO_to_TA',\n",
       " 'ALLL_to_TL',\n",
       " 'core_deposits_to_TA',\n",
       " 'brokered_deposits_to_TA',\n",
       " 'liquid_assets_to_TA',\n",
       " 'loss_provision_to_TL',\n",
       " 'NIM',\n",
       " 'assets_growth']"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(teY[:, 0].sum())\n",
    "print(df_test.defaulter.sum())\n",
    "print('num_components: %d' % num_components)\n",
    "state_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn import neighbors, linear_model\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "col_start = 0 # 19\n",
    "comp_to_keep = num_components"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Do Logistic Regression with a smaller number of predictor, based on analysis of P-values \n",
    "# for the logistic regression with a full set of variables\n",
    "\n",
    "# the original set of predictors\n",
    "# cols_to_use = state_cols + MEV_cols  \n",
    "\n",
    "# a smaller set is based on the analysis of P-values for the logistic regression\n",
    "cols_to_use = ['log_TA', 'NI_to_TA', 'Equity_to_TA', 'NPL_to_TL',\n",
    "               'core_deposits_to_TA',\n",
    "               'brokered_deposits_to_TA',\n",
    "               'liquid_assets_to_TA',\n",
    "               'FDIC_assessment_base_n'\n",
    "              ] + ['term_spread','stock_mkt_growth']\n",
    "              \n",
    "cols_to_use = state_cols + MEV_cols"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Make corrections for downsampling\n",
    "compute the ratio \n",
    "$$r = N_{Y=0}^{\\text{sub_sampled}}/N_{Y=0}^{\\text{total}}$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "48.7881355932\n",
      "47.1636542885\n",
      "1.03444349954\n"
     ]
    }
   ],
   "source": [
    "# it would be more accurate to adjust for a different average number of observations for failed and non-failed banks\n",
    "# but they turn out to be almost equal, so this can be neglected:\n",
    "\n",
    "df_data['num_records'] = df_data.groupby('IDRSSD').size()\n",
    "\n",
    "# average number of observations for failed banks\n",
    "mean_num_obs_failed_banks = df_data.loc[df_data.defaulter==1,'num_records'].mean()\n",
    "mean_num_obs_nonfailed_banks = df_data.loc[df_data.defaulter==0,'num_records'].mean()\n",
    "\n",
    "ratio_av_num_steps = float(mean_num_obs_failed_banks)/mean_num_obs_nonfailed_banks\n",
    "\n",
    "print(mean_num_obs_failed_banks)\n",
    "print(mean_num_obs_nonfailed_banks)\n",
    "print(ratio_av_num_steps)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Receiver Operating Characteristic (ROC)\n",
    "As described in scikit-learn documentation\n",
    "http://scikit-learn.org/stable/auto_examples/model_selection/plot_roc.html :\n",
    "\n",
    "Receiver Operating Characteristic (ROC) - metric to evaluate classifier output quality.\n",
    "ROC curves typically feature true positive rate on the Y axis, and false positive rate on the X axis. This means that the top left corner of the plot is the “ideal” point - a false positive rate of zero, and a true positive rate of one. This is not very realistic, but it does mean that a larger area under the curve (AUC) is usually better.\n",
    "\n",
    "The “steepness” of ROC curves is also important, since it is ideal to maximize the true positive rate while minimizing the false positive rate."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Decision Trees\n",
    "\n",
    "###  Decision Tree Classifier Part 1\n",
    "\n",
    "In this exercise use **DecisionTreeClassifier** from module sklearn.tree to classify banks into defaulters and non-defaulters. Use the following parameters to instantiate DecisionTreeClassifier\n",
    "\n",
    " - The maximum depth of the tree. Value = 5\n",
    " - The minimum number of samples required to split an internal node. Value = 50 (be sure to pass an integer)\n",
    " - Class weight: 'balanced'. The “balanced” mode uses the values of y to automatically adjust weights inversely proportional to class frequencies in the input data as n_samples / (n_classes * np.bincount(y))\n",
    " - Threshold for early stopping in tree growth. A node will split if its impurity is above the threshold, otherwise it is a leaf. Value = 0.1\n",
    " - The minimum weighted fraction of the sum total of weights (of all the input samples) required to be at a leaf node. Samples have equal weight when sample_weight is not provided. Value = 0.01"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC_auc =  0.966952868104\n",
      "KS_test =  0.876762879065\n",
      "AUC score: 0.966953 \n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9b8eec6978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn import neighbors, linear_model\n",
    "from sklearn import metrics \n",
    "\n",
    "random_state = 1873111803\n",
    "def plot_roc(model, parameters, y_true):\n",
    "    \"\"\"\n",
    "    Arguments:\n",
    "    model - trained model such as DecisionTreeClassifier, etc.\n",
    "    parameters - array-like or sparse matrix of shape  [n_samples, n_features]. The input samples. \n",
    "    y_true - True binary labels in range {0, 1} or {-1, 1}. If labels are not binary, pos_label should be explicitly given.\n",
    "    \"\"\"\n",
    "    if model is None:\n",
    "        return 0., 0., np.array([])\n",
    "    \n",
    "    predicted = model.predict_proba(parameters)[:,1]\n",
    "    threshold = 0.5\n",
    "    predicted_binary = (predicted > threshold).astype(int)\n",
    "    \n",
    "    fpr, tpr, threshold = metrics.roc_curve(y_true, predicted, pos_label=1)\n",
    "\n",
    "    roc_auc = metrics.auc(fpr, tpr) \n",
    "    ks = np.max(tpr - fpr) # Kolmogorov-Smirnov test\n",
    "\n",
    "    print('ROC_auc = ', roc_auc)\n",
    "    print('KS_test = ', ks)\n",
    "    print('AUC score: %f ' % metrics.roc_auc_score(y_true, predicted)) \n",
    "\n",
    "    try:\n",
    "        plt.title('%s ROC curve ' % model.__class__.__name__)\n",
    "        plt.plot(fpr, tpr, 'b', label='AUC = %0.2f' % roc_auc)\n",
    "        plt.legend(loc='lower right')\n",
    "        plt.plot([0,1], [0,1], 'r--')\n",
    "        plt.xlabel('False positive rate')\n",
    "        plt.ylabel('True positive rate')\n",
    "\n",
    "        # plt.savefig('ROC_curve.png')\n",
    "        plt.show()\n",
    "    except: pass\n",
    "    return roc_auc, ks, threshold\n",
    "\n",
    "roc_auc = 0. \n",
    "ks = 0.\n",
    "single_tree = None\n",
    "### START CODE HERE ### (≈ 5-8 lines of code)\n",
    "# Instantiate Decision Tree Classifier and fit the model\n",
    "# Reference Decision Tree Classifier model using single_tree variable\n",
    "# Use cols_to_use attributes from df_train data set. Labels are in column defaulter\n",
    "dtc = DecisionTreeClassifier(max_depth = 5, min_samples_split = 50, class_weight = 'balanced' , \n",
    "                             min_impurity_split = 0.1, min_weight_fraction_leaf=0.001)\n",
    "single_tree = dtc.fit(df_train[cols_to_use].values,df_train['defaulter'].values)\n",
    "\n",
    "### END CODE HERE ###\n",
    "\n",
    "if single_tree is not None:\n",
    "    # threshold : Decreasing thresholds on the decision function used to compute fpr and tpr. thresholds[0] \n",
    "    # represents no instances being predicted and is arbitrarily set to max(y_score) + 1.\n",
    "    single_tree.fit(df_train[cols_to_use].values, df_train.defaulter)\n",
    "    \n",
    "roc_auc, ks, threshold = plot_roc(single_tree, df_test[cols_to_use], df_test.defaulter)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Submission successful, please check on the coursera grader page for the status\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.96695286810376313, 0.87676287906466932]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### GRADED PART (DO NOT EDIT) ###\n",
    "part_1 = np.array([roc_auc, ks]).squeeze()\n",
    "try:\n",
    "    part1 = \" \".join(map(repr, part_1))\n",
    "except TypeError:\n",
    "    part1 = repr(part_1)\n",
    "submissions[all_parts[0]]=part1\n",
    "grading.submit(COURSERA_EMAIL, COURSERA_TOKEN, assignment_key,all_parts[:1],all_parts,submissions)\n",
    "[roc_auc, ks]\n",
    "### GRADED PART (DO NOT EDIT) ###"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  Decision Tree Classifier Part 2\n",
    "Now using DecisionTreeClassifier and the same predictors of bank failure we will predict bank failure within the first year."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC_auc =  0.976360245722\n",
      "KS_test =  0.887743162206\n",
      "AUC score: 0.976360 \n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XeYFFXWwOHfARQMGIAxwBAlg4iIKGYlCKiga8KMC7LI\n4qcYMSOrKCZMIBJURAGRVUFFwVVYXRARExIkiOQcBREmcL4/bs3QDBNqQnV1OO/z9DPd1dXVp3pm\n6vS9t+pcUVWMMcYYgFJhB2CMMSZ2WFIwxhiTzZKCMcaYbJYUjDHGZLOkYIwxJpslBWOMMdksKZgD\niMinInKTj/V2ikitaMQULSLSWkSWBbj94SLyQMTjXiKywfssj/R+Vgvq/Y0piCWFOCUiy0TkLxHZ\nISLbRGSGiPQQkWL/TlW1vaqO9LHe4aq6tDjvJSLzvAPhThHJFJHdEY8fKHgLRXrP00XkMxHZLiJb\nRORbEbkxiPfKSVW7qWp/L45ywLPA+d5nud37uaIk3ktEunmf6U5vX38UkfY51iknIgNEZIX397RI\nRO4SEcmxXnsR+dr7e9sgItNE5KKSiNPEFksK8e0SVS0PVAeeAu4DRoQbUuGoaiPvQHg48DXQK+tx\n1sEzkoiUKc77ichZwH+AL4BaQEWgF9ChONstouOAsqo6r7gbyudz+dr7bI8GhgPjRKS89xoB/g2c\nC7QDygNdgJ7AcxHb7gy8C7wOVAGOBx4DOhY37vwU93dtikhV7RaHN2AZ0DrHshbAXqCx97gs7pvo\nCmA9MAQ4JGL9TsBPwB/Ab0A7b/k0oJt3vzbwX2A7sAl4N+L1CtT27h8JvAVsBJYDDwGlvOe6AP/z\nYtkK/A60z2Wfst83Ylk34CvgJWAL0Ddi+a/e9j4Fqka8piHuwL/FW+fyiOdmAi/m87m2BpZFPH4I\nWArsAOYBHSOeq+vFlvXZjPaWl/Li3eA9Nwdo6D33NtAXaAD86X2GO4EpQBnvcQ1v3XLA88BK7/c3\nGCgXGSfwALAOeCOXfekGTIt4fIS3/ZO9xxcCfwGVc7zuDCATqOnty2qgdyH+NssAD3t/U38As4HK\n3t+S5lj3f0CXPH7XA7zX149Y/zgv5ore447Az8A2b1uNw/7fjPebtRQSiKrOAlYBZ3uLBuAOXE1x\n/5BVgEcARKQF7iB+D3AUcA7uIJPTv3AHrKOBVODlPN7+ZVxiqIX75nkjcHPE86cBC4FKwNPAiJxd\nFPk4A1gApAADROQKL+5O3rJvgdHefpUHPvf27RjgOmCoiNTznmsBjPf5vgCLgDO9fXsCGC0ix3rP\nPQF8wr7PZpC3vD1wOlDHe64z7iCXTVUXACd59w9X1ba5vPezuANzE29bNYAHI55PBQ4HquG+3efJ\n+9Z9M5CGSzIAbYAZqromR2wzcInmAlyCrUzhPrN7gCtwrY+jcAf73T5fG/m77gt8CFwT8fzVwBeq\nullETgWGeduviGvJTBCRgwsRq8nBkkLiWQNU8A64t+C+4W1R1R1Af9wBCqAr8Lqqfq6qe1V1tar+\nmsv20nHdU5VVdbeq/i/nCiJSGvfPer+q7lDVZbjuhxsiVluuqsNUNRMYieuCODbntvKwQlVfVdVM\nVf0L+AfQX1UXqmoG8DjQQkSq4L45LlLVt1Q1Q1W/xx1YrgAqAAKs9fm+qOo4VV3rfUajcYmzecRn\nUwM43vtspkcsPwKo721jvqqu8/ueAN7YUDfgDlXdqqp/AE+y7/cHkIFrOaV5n0tuzhKRbbhv108C\n16rqJu+5SuT9Waz1nq8Y8divbsADqrrY+9x+UtUtBb7Kyfm7Hs3+SeFabxlAd2Cwqn7nrf+6t/zU\nQsRqcrCkkHiq4L6VpgCHAt97A9HbgM+85QBVcc37gtyLO5DO8gaF/57LOpWAg3HdRlmWe7FkyT4o\nquou7+7hPt4f9n2zzVIdGBSxX5tw3Wap3nNnZj3nPX81LgltwXWfHO/zfRGRLiLyc8S26uP2F+Au\n4CBgtoj8knXGlqpOwXXVvQqsF5EhWf34hXAcrvsv8r0/xrV+sqxX1bQCtvM/VT0KlxAnAWdFPLeJ\nvD+L473nN0c89svv31Zucv6u/wMcJSKniMgJQCNggvdcdeC+HL/r49n/784UkiWFBOI1p6vg+lY3\n4b4dNlLVo7zbkeoGHcH9851Q0DZVdZ2q3qKqlXHf0AeLSO0cq21iX4siSzVcX3RJyFnKdyXQNWK/\njlLVQ1T1W++5L3I8d7iq9vJaS7OAy/28qXe67avArbg+7KNwYxQC4LUguqnq8cA/cd1UNb3nXlDV\nZkBjXBfMnYXc5/W4rp56OX5/R+bzueTJ2/dbga4i0sRb/B/gDBGpnGO/z8AlpanAfFzr09dn5snr\nb+tPb/uHRiw7LmeoOeLOAN7DtRauBSao6p8R7/NYjt/1oao6rhCxmhwsKSQAETlCRC4GxgJvq+ov\nqroX1986UESO8darIiIXei8bAdwsIq1EpJT3XP1ctn2liKR6D7fi/mkzI9fxuoTGAU+ISHkRqY47\nCL4dwO6C+xb+oIg08GI8yhtnAJgINBKRa0XkIO/WQkTqec/fA3QTkTtFpIL3+pNFZPQB7+JaMoob\nPBcR6YbXJeS97iqvywrcQKcCmd77tfD68f/EHdwzKQTvMx0OvCAiKeKkikhuYw9+t7kR1+/+sLdo\nMm5g930RaSgiZUSkJTAKeEVVl3p/R3cBfUXkJu9vrZSInC0iQ/J4q+HA4yJyghd3U++zXufdrheR\n0iLSnf2/SORlNK61F9l1BDAU+KeInOq9z+EicomIHFaoD8bsx5JCfPtIRHbgvjE9iDtTJXJw9z5g\nCTBTRP7AfTOsB9mD0jcDA3FnyPyX3P9BTwW+FZGduAPu7ar6ey7r3YY7AC7FtVRG4w5AJU5V38Pt\n63vefs3BnUmDqm737l+P6wdfh+tLL+s9/zXuzJ0LgWUisgXXGpiUy/vMwZ0JM8vbVn3coHaW04Dv\nRORP4H3gn+quMTgKl3S34cYg1uI+58K6C9cNNwv3O5qCG3AujoFARxFppKoKXIo7FXgK7gyrt3BJ\n946sF6jqWNwB+RZcq2Ed0I993Tg5PYMbx/kCd/bQUNxZU+pt4wFc67I2+3+eeZmBGz9J8eLMiutb\nXOvnVdwXlkW437spBnG/J2OMMcZaCsYYYyJYUjDGGJPNkoIxxphslhSMMcZki7uCU5UqVdIaNWqE\nHYYxxsSV77//fpOqphS0XtwlhRo1ajB79uywwzDGmLgiIssLXsu6j4wxxkSwpGCMMSabJQVjjDHZ\nLCkYY4zJZknBGGNMtsCSgoi8Lm6C77l5PC8i8pKILBGROSLSLKhYjDHG+BNkS+FN3HR8eWmPq/hY\nBzeD0qsBxmKMMcaHwK5TUNWvRKRGPqt0At7yyunO9GriH6+qhZn2zxhjAqUKe/dCRoa7ZWbuux95\nK8zywm6j1F9/UvaPjZx1fQ1ODXiy0TAvXqvC/lPvrfKWHZAUvMk4ugNUq1YtKsEZk8wiD4RBHuxi\neRuRy8N0Pl8yjFvYzpHMqjebU08Ndig4zKQguSzLdXIHVR2Km6iD5s2b2wQQJhBZB8J4PIAFsY1Y\nUabM/rfSpQ9clt/yMmWgXLnib6cwy0tk2zu3ccgj91DmzeFo7drI8IE0Ozf4c4PCTAqrcBN8Z0nF\nzepkoiyyaRzrB6qgtxErSuLAknUgDOygFYUDa6lSILl9fUx0mZlw5hmwcCHcey/Sty8cckhU3jrM\npDAR6CUiY3HTGm638YTCWb8e/vMfmDIFfv216AfSWBF5YCjqQaVs2dg4mBVnG0l7IDSweTNUqOD+\nKJ54AqpWhebNoxpCYElBRMYA5wGVRGQV8ChwEICqDsHNidsBN4fwLvafW9jkYvdumD7dJYEpU+Cn\nn9zyihXh5JMPPCDG07fE0qXtQGiSmCq88w7cfjs89RTccgtcdlkooQSWFFT1mgKeV+CfQb1/IlCF\n+fP3JYH//hf++gsOOgjOPBP694e2bV1CKGWXIRoTn1auhB49YNIkOP10988dojC7j0wuNmzY1yX0\n+eewxhtlqV/ffXlo2xbOPRcOPzzcOI0xJWDMGPjHP1zf7gsvQK9ertkcIksKIduzZ/8uoR9/dMsr\nVIDWrV0SaNMG7ExcYxLQ0UfDaafB0KFQs2bY0QCWFKJOFRYs2L9LaNcu17d+5plubKlNG2jWLPQv\nDMaYkpaRAQMHQloaPPggtGsHF14YUwNqlhSiYNOmfV1CU6bA6tVueb160LXrvi6h8uXDjdMYE6Cf\nf3b/8N9/D1dd5b4hisRUQgBLCoHYswe++WZfEvjhB/f7P/ro/buEqlcPO1JjTOD27IHHH3dnFVWo\nAO+9B5dfHnPJIIslhRKg6q4xyUoC06bBn3+6LqGWLaFfP5cITjnFuoSMSTqLF8OAAXDttfD88+4c\n8hhmSaEYNm6Et9+G11+HuV6B8Dp1oEsXlwTOOw+OOCLMCI0xodi5EyZMgOuug8aN3dWltWqFHZUv\nlhQKKSMDJk92ieCjjyA9HVq0gFdegQ4dYuYEAmNMWD7/HLp3h+XL3RkjDRrETUIASwq+LVoEb7wB\nI0fC2rWQkgK33QZ//zs0ahR2dMaY0G3dCnff7b4x1q3rTi1s0CDsqArNkkI+du6E8ePd7/jrr91V\nwx06uERw0UVw8MFhR2iMiQmZme6c8kWL4P774ZFHXEXCOGRJIQdVd+bQ66/Du++6xFC3rjtx4IYb\noHLlsCM0xsSMTZv2FbDr399dZdosvmcWtqTgWbsWRo1yyWDhQjjsMLj6atcqOOOMmD17zBgTBlV3\nwLjjDveNsXt3uPTSsKMqEUmdFNLT4ZNPXCKYNGlfC/Dee+HKK+1iMmNMLpYvd/WKJk923xjPOSfs\niEpU0iaF556Dp592BeiOO86ND918s7vK2BhjcvX223Drra6l8PLL0LNnwpUoTtqk0L8/pKbCiBGu\n/EiZpP0kjDG+paS47oTXXkvYkgRJeyhMS4NWreDii8OOxBgTs9LTXbdCejo8/LArXte2bUIPMiZW\nu6cQ0tPdZDXGGJOrH390Za3vv9/NdqXqlidwQoAkTwp2nYEx5gC7d8MDD8Cpp7pZrv79bzcZToIn\ngyxJmRT27nU3aykYYw6wZAk8+yzceKOb/ORvfws7oqhKyjGF9HT305KCMQZwV6l+8IG7QrVxY3ex\nUpIWMkvKloIlBWNMtsmTXQGzm25yLQNI2oQASZoU0tLcT0sKxiSxzZtdImjXDg491BU4i8MCdiXN\nuo+MMcknq3zBkiVuruSHHorbAnYlzZKCMSZ5bNzoZj4rXdrNhla9OjRtGnZUMSUpu48sKRiTZFTd\nhCh168KwYW5Zp06WEHJhScEYk9iWLXNXIv/973DiiXD++WFHFNMsKRhjEteoUe4U02++gcGDYdo0\n11owebIxBWNM4jr2WFfaesgQNwGOKVBSJwUrc2FMgklPdzXxMzPdlJht27qb8c26j4wxieGHH1y9\nooceclckZxWwM4ViScEYE9/++gv69IEWLWD9eleu4p13kqaAXUkLNCmISDsRWSgiS0SkTy7PVxOR\nqSLyo4jMEZEOQcaTxa5oNiaBLF0Kzz8PXbq4EtcJMldyWAJLCiJSGhgEtAcaAteISMMcqz0EjFPV\nk4HOwOCg4olkLQVj4twff8Cbb7r7jRrB4sUwfDgcfXSoYSWCIFsKLYAlqrpUVdOAsUCnHOsocIR3\n/0hgTYDxZLOkYEwcmzTJnWbateu+AnYJOjVmGIJMClWAlRGPV3nLIvUFrheRVcAk4LbcNiQi3UVk\ntojM3rhxY7EDs6RgTBzatMmVtr7oIihfHqZPtwJ2AQgyKeQ2ypPzdIBrgDdVNRXoAIwSkQNiUtWh\nqtpcVZunpKQUOzBLCsbEmawCdmPHulNNf/gBTj897KgSUpDXKawCqkY8TuXA7qGuQDsAVf1GRMoB\nlYANAcZlScGYeLF+PaSkuAJ2zz7ruomaNAk7qoQWZEvhO6COiNQUkYNxA8kTc6yzAmgFICINgHJA\n8fuHCmBJwZgYpwojRkC9ejB0qFt2ySWWEKIgsKSgqhlAL2AysAB3ltE8EeknIh291e4CbhGRn4Ex\nQBfV4K84sSuajYlhS5dC69bQrZurYtq6ddgRJZVAy1yo6iTcAHLkskci7s8HzgwyhtxYS8GYGDVy\nJPTs6bqLhgyBW26BUkl5jW1okrr2kSUFY2JM5cpwwQXw6quQmhp2NEkpKZOCXdFsTIxIS4OnnoK9\ne6FvX2jTxt1MaJKyXWYtBWNiwHffwSmnwKOPunEEK2AXE5I6KZQuHW4cxiSlXbvg7rvddQZbt8LE\nifDWW1bALkYkbVI46CD7GzQmFL//Di+/7AaR581zp5qamJGUYwpZScEYEyXbt8P778PNN7sCdkuW\nQNWqBb/ORF1StxSMMVHwyScuEXTrBr/+6pZZQohZlhSMMcHYuBGuuw4uvtiVtP7mG6hfP+yoTAGs\n+8gYU/IyM+Gss9z4wWOPuZnRrIRAXEjapGB/n8YEYN06OOYYd2rfc89BjRpu7gMTN6z7yBhTfHv3\nwmuvQd267ie4biNLCHGnwKQgIoeIyP0iMsR7XFtE2gcfWnAsKRhTgpYsgVatoEcPOPVUuPDCsCMy\nxeCnpfA6bsKcs7zHa4D+gUUUBWlplhSMKRFvvAEnnugmvRk2DP7zH6hVK+yoTDH4SQp1VLU/kA6g\nqrvIfVa1uGEtBWNKSLVqrmUwf7475dSuCI17fgaa07wZ0RRARGoCaYFGFTBLCsYU0Z498OSTbgyh\nXz/XbdSqVdhRmRLkp6XwL+AzIFVERgJTgQcCjSpglhSMKYJvv3UF7B57DFassAJ2CarAloKqfioi\ns4EzcN1G96hqoHMoBy09HcqWDTsKY+LEn3/Cww/DCy9AlSrw8cdw0UVhR2UC4ufsoymqulFVJ6jq\nh6q6QUSmRCO4oFhLwZhCWL4cBg92ZxfNm2cJIcHl2VIQkYOBcsCxIlKefYPLRwDVohBbYCwpGFOA\nbdtg/Hg3eNywoTvt1GZCSwr5dR/9E7gTOAaYx76k8AcwJOC4AmVXNBuTjwkT4NZbYcMGV6qifn1L\nCEkkz+4jVR2oqlWB+1S1mqpW9W6NVPWFKMZY4qylYEwuNmyAzp3h0kshJQVmzrQCdknIz0DzCyJS\nH2iI607KWj46yMCCZEnBmBwyM+HMM91ZRY8/Dvfea/8kSarApCAiDwFtgfrAZOBC4H9A3CYFu6LZ\nGM+aNXDcca6A3YsvugJ2DRuGHZUJkZ/rFK4GzgfWquoNwEnEeXVVaymYpLd3L7z6quseGuINEXbo\nYAnB+EoKf6lqJpDhnYW0Dojr4iaWFExSW7QIzj8fevaE006D9nFd39KUMD/f+H8UkaNwhfFm484+\n+iHQqAJmScEkrREjoFcvKFcOXn8dunSxekVmP/kmBRERoK+qbgMGichk4AhVtaRgTDyqUcO1DAYN\nguOPDzsaE4PyTQqqqiLyMXCK93hJVKIKmCUFkzT27IF//cvdf/xxK2BnCuRnTGGWiDQLPJIoUXVn\n31lSMAlvxgxo2hSeeALWrrUCdsYXP0nhLFxiWCgiP4jIjyISt91H6enupyUFk7B27oTbb3dXI+/a\nBZ995sYSbOzA+OBnoPnSom5cRNoBLwKlgeGq+lQu61wF9MXN1/Czql5b1PfzIyspWJkLk7BWrHDz\nJP/zn9C/P5QvH3ZEJo74uaL5t6JsWERKA4OANsAq4DsRmaiq8yPWqQPcD5ypqltF5JiivFdhWEvB\nJKStW+G996B7d3etwdKlULly2FGZOOSn+6ioWgBLVHWpqqYBY4FOOda5BRikqlsBojFPQ5o3Z5wl\nBZMwPvjAJYKePWHhQrfMEoIpoiCTQhVgZcTjVd6ySHWBuiIyXURmet1NBxCR7iIyW0Rmb9y4sVhB\nWUvBJIx16+DKK+Fvf3OlKmbNgnr1wo7KxDlfSUFEUkXkfO9+WRE5zM/LclmW8/SHMkAd4DzgGmC4\nd6Hc/i9SHaqqzVW1eUpKip+Q82RJwSSEzEw4+2z46CM3bjBrFjRLmJMETYj8FMT7O9ALOBI4AagO\nDAZaF/DSVUDViMepwJpc1pmpqunA7yKyEJckvvMVfRFYUjBxbdUq1zVUujS89BLUrGnlrU2J8tNS\n+D/gdFx5C1R1EW7inYJ8B9QRkZreLG6dgYk51vkQV2wPEamE605a6i/0orGkYOLS3r3w8ssuAbz6\nqlvWvr0lBFPi/CSF3d5AMZB9VlGBJzyragauhTEZWACMU9V5ItJPRDp6q00GNovIfGAqcI+qbi7s\nThSGJQUTd379Fc45B/7v/9y1BxdfHHZEJoH5uU5huojcC5TzxhX+CXzsZ+OqOgmYlGPZIxH3FTfl\n552+Iy4mSwomrgwf7grYHXoojBwJN9xgF6GZQPlpKdwL7AB+BW4HvgAeDDKoIFlSMHHlhBPgkktg\nwQK48UZLCCZwfloKHXBXI78adDDRYFc0m5i2ezf06+fu9+/v5j04//xwYzJJxU9L4SpgiYi8ISIX\nemMKcctaCiZmTZ/uCtg9+SRs3GgF7EwoCkwK3hScdYGPgL8DS0VkSNCBBcWSgok5O3bAbbe56w72\n7IHJk2HYMOsqMqHwdfGaqu4BJgBv4k41vSrAmAJlZS5MzFm1yg0o33Yb/PILtG0bdkQmiRWYFESk\ntYgMB34DrgfeAo4LOrCgWEvBxITNm/ddb9CggStg9+KLcPjh4cZlkp6fgeYeuGJ2t6nqXwHHEzhL\nCiZUqvDvf7uy1lu2wAUXuHpFNjWmiRF+xhSuUNXxiZAQwJKCCdHatXD55a6IXdWqMHu2FbAzMSfP\nloKI/FdVzxWRrexfyE5w151VCDy6AFhSMKHIKmC3ejU8/TT07g1l/DTUjYmu/P4qs06OrhSNQKLF\nkoKJqpUroUoVV8Bu0CBXwK5u3bCjMiZPeXYfqepe7+4IVc2MvAEjohNeybOkYKIiM9NVMY0sYHfh\nhZYQTMzz035tEvnAu3jt1GDCCZ4lBRO4BQuga1f45htXyfSSS8KOyBjf8mwpiMh93nhCExHZ4t22\nAhvJUeQunliZCxOooUPdVcmLFsGoUfDJJ1CtWthRGeNbfmcfPQ2kAAO9nylAJVWtoKr3RCO4IFhL\nwQSqTh247DKYPx+uv96uSjZxJ7/uo9qqulhERgGNshaK90euqnMCji0QWVc024kfpkT89Rf07esO\n/k89ZQXsTNzL79DYB+gKDMrlOQXOCSSigKWnu4RgX+BMsX31FXTrBosXQ48e7sI0+8MycS7PpKCq\nXb2fZ0cvnOClp1vXkSmmP/6APn3cWUW1asEXX7grk41JAH5qH/1NRMp79/uIyDgROSn40IJhScEU\n25o18OabcOedMGeOJQSTUPxUSe2rqjtE5AzgEuBd4LVgwwqOJQVTJJs2weDB7n79+vD77/Dcc3DY\nYeHGZUwJ85MUMr2fFwODVfXfQNngQgqWJQVTKKrw7rvQsCHccYc71RTg2GPDjcuYgPhJCmtFZBDQ\nGZgkIgf7fF1MsqRgfFuzBi69FDp3hurV4fvv7Ypkk/D8nJh5FW6e5pdVdauIVMadmRSXLCkYXzIz\n4ZxzXAG7Z5+F22+385hNUijwr1xVd4rIfOA8ETkP+FpVPw08soCkp9vVzCYfy5dDaqorYDd4sDu7\nqHbtsKMyJmr8nH3UCxgHVPNu40SkZ9CBBcVaCiZXmZnw/PNuFrSsAnZt21pCMEnHT3u4O9BCVXcC\niEh/YAYwOMjAgpKWZknB5DB3ritgN2sWXHyxG0cwJkn5GTAWID3icbq3LC5ZS8HsZ8gQaNbMzZE8\nejRMnOi6j4xJUn5aCqOAmSLyb1wyuBQYGWhUAbKkYIB9JSkaNHDTY77wAqSkhB2VMaHzM9D8tIhM\nBbLKXfRQ1e+CDSs4lhSS3K5d8MgjbiB5wAA491x3M8YA/q832OPd/vJ+xi1LCkls2jRo0sRdibxz\np2stGGP24+fsoweBMcDxQCowWkTuDzqwoFhSSELbt8M//rGvpPWXX7r5kq2iqTEH8DOmcD1wiqru\nAhCRJ4DvgSeDDCwolhSS0Nq18PbbcPfd8NhjcOihYUdkTMzy0320nP2TRxlgqZ+Ni0g7EVkoIktE\nJM+roEXkChFREWnuZ7vFYUkhSWzcCC+/7O7Xrw/LlsEzz1hCMKYAfpLCLmCeiAwXkWHAL8A2EXle\nRJ7P60UiUho3QU97oCFwjYg0zGW98sD/Ad8WZQcKy5JCglN1p5Y2aAB33bWvgJ2dWWSML366jz7x\nbllm+tx2C2CJqi4FEJGxQCdgfo71/oWbD/pun9stFitzkcBWroRbb4VPPoHTToMRI6yAnTGF5OeU\n1BFF3HYVYGXE41XAaZEriMjJQFVV/VhE8kwKItIdd2U11apVK2I4jl3RnKAyMuC882DdOhg4EG67\nzZ12aowplCDLPuZ2akf2OYAiUgoYCHQpaEOqOhQYCtC8efNinUdo3UcJZtkyqFrVVTB97TVXwK5W\nrbCjMiZuBTkvwiqgasTjVGBNxOPyQGNgmogsA04HJgY92GxJIUFkZLiS1g0a7JsRrXVrSwjGFJPv\nloKIlFXVwly49h1QR0RqAqtxk/Rcm/Wkqm4HKkVsfxpwt6rOLsR7FJolhQQwZ44rYDd7NnTqBJdf\nHnZExiQMPxevtRCRX4DF3uOTROTlgl6nqhlAL2AysAAYp6rzRKSfiHQsZtxFZkkhzg0eDKec4uY9\nePdd+OADqFw57KiMSRh+Wgov4eZn/hBAVX8WkfP9bFxVJwGTcix7JI91z/OzzeJQdb0OlhTiUFYB\nu8aN3fSYAwdCpUoFv84YUyh+kkIpVV0u+5cEyAwonkBlZLiflhTiyJ9/wkMPuYHkZ55xU2Sec07Y\nURmTsPwMNK8UkRaAikhpEbkDWBRwXIFI92aFsKQQJ774Ak480ZW13rPHCtgZEwV+ksKtwJ24qTjX\n484SujXIoIJiSSFObNsG3bq5s4nKlIGvvoKXXrICdsZEgZ+L1zbgzhyKe5YU4sT69TB2LNx3Hzz6\nKBxySNgfGwqhAAAV+klEQVQRGZM0CkwKXr2jA9rtqto9kIgClJUUrMxFDMpKBLffDvXquYvSbCDZ\nmKjzM9D8n4j75YDL2L98RdxIS3M/raUQQ1ThnXdcMti5Ezp0gDp1LCEYExI/3UfvRj4WkVHA54FF\nFCDrPooxK1ZAjx7w6afQsqUrYFenTthRGZPUilL7qCZQvaQDiQZLCjEkq4Ddhg1uELlnTytgZ0wM\n8DOmsJV9YwqlgC1AnhPmxDJLCjFg6VKoXt2dVTRsGJxwAtSoEXZUxhhPvqekirti7SQgxbsdraq1\nVHVcNIIraZYUQpSRAQMGQMOGbn5kgFatLCEYE2PyTQqqqsAHqprp3eL66iFLCiH56Sc36U2fPm4g\n+corw47IGJMHPxevzRKRZoFHEgWWFELwyitw6qmwejWMHw/vvw/HHx92VMaYPOQ5piAiZbxKp2cB\nt4jIb8CfuMlzVFXjLlFYUoiirAJ2TZrAddfB889DhQphR2WMKUB+A82zgGbApVGKJXCWFKJg5054\n8EH3IT/7rBWwMybO5JcUBEBVf4tSLIGzK5oDNmUKdO/urj+47bZ9rQVjTNzILymkiMideT2pqs8H\nEE+g7IrmgGzdCnfeCW++6UpUfPUVnHVW2FEZY4ogv6RQGjgcr8WQCKz7KCAbNrhB5Pvvh0cegXLl\nwo7IGFNE+SWFtaraL2qRRIElhRK0bh2MGQO9e+8rYFexYthRGWOKKb9TUhOmhZDFkkIJUIWRI91F\naPffD4sXu+WWEIxJCPklhVZRiyJKLCkU07Jl0K4ddOniksJPP1kBO2MSTJ7dR6q6JZqBRIMlhWLI\nyIDzz4dNm1yZih49oJSfax+NMfGkKFVS45YlhSJYsgRq1nQF7F5/HWrVcgXtjDEJKam+6llSKIT0\ndOjfHxo12lfA7vzzLSEYk+CspWAO9MMP0LWrGzO48kq4+uqwIzLGRIm1FMz+XnoJWrRwp5y+/z6M\nGwfHHht2VMaYKEmqpJCW5ib3svHRXGRVRT/5ZLjxRpg/Hy67LNyYjDFRl3TdR9ZKyGHHDne9Qdmy\n8NxzcPbZ7maMSUpJ9Z3ZkkIOn30GjRvD4MGupRDfcygZY0qAJYVktHkz3HQTtG8Phx0G06e7+Q6s\noqkxSc+SQjLavBk++AAefhh+/BFatgw7ImNMjAg0KYhIOxFZKCJLRKRPLs/fKSLzRWSOiHwhIoGe\nBJ/USWHtWjfpjSrUrQvLl0O/fm4swRhjPIElBREpDQwC2gMNgWtEpGGO1X4EmqtqE2A88HRQ8UCS\nJgVVdyVygwauZbBkiVt+9NHhxmWMiUlBthRaAEtUdamqpgFjgU6RK6jqVFXd5T2cCaQGGE/yJYXf\nf4e2bd2FaCedBD//bAXsjDH5CvKU1CrAyojHq4DT8lm/K/Bpbk+ISHegO0C1atWKHFBSJYWMDLjg\nAjd+8OqrbppMu0DDGFOAIJNCbqey5HrOo4hcDzQHzs3teVUdCgwFaN68eZHPm0yKpLB4sStaV6YM\nvPEGnHACVK0adlTGmDgR5FfHVUDk0SgVWJNzJRFpDTwIdFTVPQHGQ3o6HHxwkO8QovR0ePxxd93B\nK6+4ZeedZwnBGFMoQbYUvgPqiEhNYDXQGbg2cgURORl4DWinqhsCjAVwZS4SsqUwe7YbN5gzBzp3\nhmuuCTsiY0ycCqyloKoZQC9gMrAAGKeq80Skn4h09FZ7BjgceE9EfhKRiUHFAwnaffTii3DaaW7y\nmwkT3LzJxxwTdlTGmDgVaO0jVZ0ETMqx7JGI+62DfP+c0tOhXLlovmOAVN0VyM2bu1bC00/DUUeF\nHZUxJs5ZQbx488cfcN99LrsNHAhnnuluxhhTApLqHMW4TwqTJrmZ0IYOdWcXWQE7Y0wJs6QQDzZt\nguuvh4sugiOPhBkz4JlnrICdMabEWVKIB1u3wkcfwaOPuqkyT8vvGkBjjCk6G1OIVatXwzvvwD33\nuNIUy5fbQLIxJnDWUog1qjBsGDRsCH37wm+/ueWWEIwxUWBJIZb89hu0auXqFDVr5i5Gq1077KiM\nMUkkqbqP0tJiuMxFRoZLCFu2wGuvQbduVsDOGBN1SZUUYrKlsHChK1pXpgyMHOnupwZaQdwYY/KU\nVF9FYyoppKXBY4/BiSfCoEFu2bnnWkIwxoQqaVoKqjGUFGbNcqUp5s6Fa6+F664LOyJjjAGSqKWQ\nmel+hp4UXngBWrbcd+3BO+9ApUohB2WMMU7SJIX0dPcztKSQVZKiRQu45RaYNw8uvjikYIwxJndJ\n030UWlLYvh3uvRcOOcS1Es44w92MMSYGWUshSB995C5CGz4cypa1AnbGmJhnSSEIGze6AeSOHaFi\nRZg5EwYMsAJ2xpiYZ0khCNu3uzLXjz3mpso89dQovKkxxhRf0owppKW5n4Fd0bxyJbz9NvTp40pT\nLF/uylwbY0wcsZZCce3dC0OGuMlvHn98XwE7SwjGmDhkSaE4Fi+GCy6AW291p5r+8osVsDPGxLWk\n6T4q8aSQkQFt2sC2bTBiBNx8sw0kG2PiniWFwlqwwE16U6YMjBrlCthVrlzs+IxJROnp6axatYrd\nu3eHHUrSKFeuHKmpqRxUxIOdJQW/9uyB/v3d7Zln4I474OyzSyw+YxLRqlWrKF++PDVq1ECsJR04\nVWXz5s2sWrWKmjVrFmkbNqbgx8yZbtKbfv3gmmvghhtKNDZjEtXu3bupWLGiJYQoEREqVqxYrJaZ\nJYWCPPecK0uxY4e79uCtt9wFacYYXywhRFdxP29LCnnZu9f9bNkSevRwZa7btw8kNmOMiRWWFHLa\nts3NdXD77e7xGWfA4MFwxBGBxmeMCc4HH3yAiPDrr79mL5s2bRoX56hU3KVLF8aPHw+4QfI+ffpQ\np04dGjduTIsWLfj000+LHcuTTz5J7dq1qVevHpMnT851nS+//JJmzZrRuHFjbrrpJjIyMgDYvn07\nl1xyCSeddBKNGjXijTfeKHY8OVlSiPThh66A3ciRUL68FbAzJkGMGTOGs846i7Fjx/p+zcMPP8za\ntWuZO3cuc+fO5aOPPmLHjh3FimP+/PmMHTuWefPm8dlnn9GzZ08ysyZ78ezdu5ebbrqJsWPHMnfu\nXKpXr87IkSMBGDRoEA0bNuTnn39m2rRp3HXXXaRllWsoIUlz9lG+ZS42bIBeveC996BpU/j4Yzew\nbIwpMXfcAT/9VLLbbNrUVaTPz86dO5k+fTpTp06lY8eO9O3bt8Dt7tq1i2HDhvH7779TtmxZAI49\n9liuuuqqYsU7YcIEOnfuTNmyZalZsya1a9dm1qxZtGzZMnudzZs3U7ZsWerWrQtAmzZtePLJJ+na\ntSsiwo4dO1BVdu7cSYUKFShTpmQP49ZSAPjjD/j8c3jiCTdVpiUEYxLGhx9+SLt27ahbty4VKlTg\nhx9+KPA1S5YsoVq1ahzho9u4d+/eNG3a9IDbU089dcC6q1evpmrVqtmPU1NTWb169X7rVKpUifT0\ndGbPng3A+PHjWblyJQC9evViwYIFVK5cmRNPPJEXX3yRUqVK9jCeNC2FA5LCihXu4rMHHnClKVas\ncF1GxphAFPSNPihjxozhjjvuAKBz586MGTOGZs2a5XmWTmHP3hk4cKDvdTWXLumc7ycijB07lt69\ne7Nnzx7atm2b3RqYPHkyTZs25csvv+S3336jTZs2nH322b6Sl1+BJgURaQe8CJQGhqvqUzmeLwu8\nBZwCbAauVtVlQcSSnRRK74XBQ+C++9wZRldf7ZKCJQRjEs7mzZv58ssvmTt3LiJCZmYmIsLTTz9N\nxYoV2bp1637rb9myhUqVKlG7dm1WrFjBjh07KF/AsaF3795MnTr1gOWdO3emT58++y1LTU3N/tYP\n7uK+yrlURGjZsiVff/01AFOmTGHRokUAvPHGG/Tp0wcRoXbt2tSsWZNff/2VFi1a+PtA/FDVQG64\nRPAbUAs4GPgZaJhjnZ7AEO9+Z+DdgrZ7yimnaFG88IJqXX7VtJZnq4Jqmzaqv/9epG0ZY/yZP39+\nqO8/ZMgQ7d69+37LzjnnHP3qq6909+7dWqNGjewYly1bptWqVdNt27apquo999yjXbp00T179qiq\n6po1a3TUqFHFimfu3LnapEkT3b17ty5dulRr1qypGRkZB6y3fv16VVXdvXu3XnDBBfrFF1+oqmqP\nHj300UcfVVXVdevWaeXKlXXjxo0HvD63zx2YrT6O3UGOKbQAlqjqUlVNA8YCnXKs0wkY6d0fD7SS\ngK50ydidwWQupMyCX+CNN2DyZKhRI4i3MsbEiDFjxnDZZZftt+zyyy9n9OjRlC1blrfffpubb76Z\npk2bcsUVVzB8+HCO9MreP/7446SkpNCwYUMaN27MpZdeSkpKSrHiadSoEVdddRUNGzakXbt2DBo0\niNKlSwPQoUMH1qxZA8AzzzxDgwYNaNKkCZdccgkXXHAB4M6ImjFjBieeeCKtWrViwIABVKpUqVgx\n5SQa0GmXInIF0E5Vu3mPbwBOU9VeEevM9dZZ5T3+zVtnU45tdQe6A1SrVu2U5cuXFzqeCRNg9gv/\n4+E3T+Dg6scXdbeMMYWwYMECGjRoEHYYSSe3z11EvlfV5gW9Nsgxhdy+8efMQH7WQVWHAkMBmjdv\nXqQs1qkTdOp0VlFeaowxSSPI7qNVQNWIx6nAmrzWEZEywJHAlgBjMsYYk48gk8J3QB0RqSkiB+MG\nkifmWGcicJN3/wrgSw2qP8sYEwr7l46u4n7egSUFVc0AegGTgQXAOFWdJyL9RKSjt9oIoKKILAHu\nBPrkvjVjTDwqV64cmzdvtsQQJerNp1CuXLkibyOwgeagNG/eXLOu9DPGxDabeS368pp5LRYGmo0x\nSe6ggw4q8gxgJhxJU/vIGGNMwSwpGGOMyWZJwRhjTLa4G2gWkY1A4S9pdioBmwpcK7HYPicH2+fk\nUJx9rq6qBdbpiLukUBwiMtvP6HsisX1ODrbPySEa+2zdR8YYY7JZUjDGGJMt2ZLC0LADCIHtc3Kw\nfU4Oge9zUo0pGGOMyV+ytRSMMcbkw5KCMcaYbAmZFESknYgsFJElInJA5VURKSsi73rPfysiNaIf\nZcnysc93ish8EZkjIl+ISPUw4ixJBe1zxHpXiIiKSNyfvuhnn0XkKu93PU9ERkc7xpLm42+7mohM\nFZEfvb/vDmHEWVJE5HUR2eDNTJnb8yIiL3mfxxwRaVaiAfiZyDmebkBp4DegFnAw8DPQMMc6PYEh\n3v3OwLthxx2FfT4fONS7f2sy7LO3XnngK2Am0DzsuKPwe64D/Agc7T0+Juy4o7DPQ4FbvfsNgWVh\nx13MfT4HaAbMzeP5DsCnuJkrTwe+Lcn3T8SWQgtgiaouVdU0YCzQKcc6nYCR3v3xQCsRyW1q0HhR\n4D6r6lRV3eU9nImbCS+e+fk9A/wLeBpIhNrNfvb5FmCQqm4FUNUNUY6xpPnZZwWO8O4fyYEzPMYV\nVf2K/Geg7AS8pc5M4CgRKbGJ5xMxKVQBVkY8XuUty3UddZMBbQcqRiW6YPjZ50hdcd804lmB+ywi\nJwNVVfXjaAYWID+/57pAXRGZLiIzRaRd1KILhp997gtcLyKrgEnAbdEJLTSF/X8vlEScTyG3b/w5\nz7v1s0488b0/InI90Bw4N9CIgpfvPotIKWAg0CVaAUWBn99zGVwX0nm41uDXItJYVbcFHFtQ/Ozz\nNcCbqvqciLQERnn7vDf48EIR6PErEVsKq4CqEY9TObA5mb2OiJTBNTnza67FOj/7jIi0Bh4EOqrq\nnijFFpSC9rk80BiYJiLLcH2vE+N8sNnv3/YEVU1X1d+BhbgkEa/87HNXYByAqn4DlMMVjktUvv7f\niyoRk8J3QB0RqSkiB+MGkifmWGcicJN3/wrgS/VGcOJUgfvsdaW8hksI8d7PDAXss6puV9VKqlpD\nVWvgxlE6qmo8z+Xq52/7Q9xJBYhIJVx30tKoRlmy/OzzCqAVgIg0wCWFjVGNMromAjd6ZyGdDmxX\n1bUltfGE6z5S1QwR6QVMxp258LqqzhORfsBsVZ0IjMA1MZfgWgidw4u4+Hzu8zPA4cB73pj6ClXt\nGFrQxeRznxOKz32eDLQVkflAJnCPqm4OL+ri8bnPdwHDRKQ3rhulSzx/yRORMbjuv0reOMmjwEEA\nqjoEN27SAVgC7AJuLtH3j+PPzhhjTAlLxO4jY4wxRWRJwRhjTDZLCsYYY7JZUjDGGJPNkoIxxphs\nlhRMzBKRTBH5KeJWI591a+RVVTLaRKS5iLzk3T9PRM6IeK6HiNwYxViaxnvVUBNdCXedgkkof6lq\n07CDKCzvArmsi+TOA3YCM7znhpT0+4lIGa+GV26a4sqaTCrp9zWJyVoKJq54LYKvReQH73ZGLus0\nEpFZXutijojU8ZZfH7H8NREpnctrl4nIAG+9WSJS21teXdw8FFnzUVTzll8pInNF5GcR+cpbdp6I\nfOy1bHoAvb33PFtE+orI3SLSQERm5divOd79U0TkvyLyvYhMzq0Cpoi8KSLPi8hUYICItBCRGeLm\nFJghIvW8K4D7AVd773+1iBwmrl7/d966uVWWNcks7NrhdrNbXjfcFbk/ebcPvGWHAuW8+3VwV7UC\n1MCrPw+8DFzn3T8YOARoAHwEHOQtHwzcmMt7LgMe9O7fCHzs3f8IuMm7/3fgQ+/+L0AV7/5R3s/z\nIl7XF7g7YvvZj739quXdvw94CHfl6gwgxVt+Ne4q3pxxvgl8DJT2Hh8BlPHutwb+7d3vArwS8br+\nwPVZ8QKLgMPC/l3bLXZu1n1kYllu3UcHAa+ISFNc0qiby+u+AR4UkVTgfVVdLCKtgFOA77wyH4cA\nedWAGhPxc6B3vyXwN+/+KNwcDQDTgTdFZBzwfmF2DlfE7SrgKdzB/2qgHq6Q3+denKWBvOravKeq\nmd79I4GRXqtI8coi5KIt0FFE7vYelwOqAQsKGbtJUJYUTLzpDawHTsJ1fx4weY6qjhaRb4GLgMki\n0g1Xbnikqt7v4z00j/sHrKOqPUTkNO+9fvKSlV/v4mpRve82pYtF5ERgnqq29PH6PyPu/wuYqqqX\ned1W0/J4jQCXq+rCQsRpkoiNKZh4cySwVl2t/Btw36T3IyK1gKWq+hKuomQT4AvgChE5xlunguQ9\nT/XVET+/8e7PYF/hxOuA/3nbOUFVv1XVR4BN7F/SGGAHroz3AVT1N1xr52FcggBX6jpF3LwAiMhB\nItIojzgjHQms9u53yef9JwO3idcMEVc915hslhRMvBkM3CQiM3FdR3/mss7VwFwR+Qmoj5u6cD6u\nz36KN6D7OZDXFIZlvZbG7biWCcD/ATd7r73Bew7gGRH5xTsd9ivcHMKRPgIuyxpozuW93gWuZ998\nAGm4cu4DRORn3LjDAYPpuXgaeFJEprN/opwKNMwaaMa1KA4C5ngx/8vHtk0SsSqpxkQQNyFPc1Xd\nFHYsxoTBWgrGGGOyWUvBGGNMNmspGGOMyWZJwRhjTDZLCsYYY7JZUjDGGJPNkoIxxphs/w9bj6He\n6tymZAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9b8ef07d30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "if single_tree is not None:\n",
    "    single_tree.fit(df_train[cols_to_use].values, df_train.default_within_1Y)\n",
    "roc_auc, ks, threshold = plot_roc(single_tree, df_test[cols_to_use], df_test.default_within_1Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Submission successful, please check on the coursera grader page for the status\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.9763602457218078, 0.88774316220564575]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### GRADED PART (DO NOT EDIT) ###\n",
    "part_2 = np.array([roc_auc, ks]).squeeze()\n",
    "try:\n",
    "    part2 = \" \".join(map(repr, part_2))\n",
    "except TypeError:\n",
    "    part2 = repr(part_2)\n",
    "submissions[all_parts[1]]=part2\n",
    "grading.submit(COURSERA_EMAIL, COURSERA_TOKEN, assignment_key,all_parts[:2],all_parts,submissions)\n",
    "[roc_auc, ks]\n",
    "### GRADED PART (DO NOT EDIT) ###"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Random Forests\n",
    "\n",
    "### Random Forests Part 3\n",
    "As defined by http://scikit-learn.org/stable/modules/ensemble.html:\n",
    "\n",
    "In random forests (see RandomForestClassifier and RandomForestRegressor classes), each tree in the ensemble is built from a sample drawn with replacement (i.e., a bootstrap sample) from the training set. In addition, when splitting a node during the construction of the tree, the split that is chosen is no longer the best split among all features. Instead, the split that is picked is the best split among a random subset of the features. As a result of this randomness, the bias of the forest usually slightly increases (with respect to the bias of a single non-random tree) but, due to averaging, its variance also decreases, usually more than compensating for the increase in bias, hence yielding an overall better model.\n",
    "\n",
    "Use the following parameters to instantiate **RandomForestClassifier**:\n",
    "\n",
    "- The number of features to consider when looking for the best split. Value = 0.5\n",
    "- The number of trees in the forest. Value = 1000\n",
    "- The maximum depth of the tree. Value = 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC_auc =  0.993697478992\n",
      "KS_test =  0.956850566313\n",
      "AUC score: 0.993697 \n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9b8e879240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "clf  = None\n",
    " ### START CODE HERE ### (≈ 2 lines of code) \n",
    "# instantiate Random Forest Classifier\n",
    "# reference Random Forest Classifier model using clf variable\n",
    "# ... please set random_state=42 in a random forest classifier of sklearn\n",
    "clf = RandomForestClassifier(max_depth = 10, max_features = 0.5, n_estimators = 1000, random_state = 42)\n",
    "### END CODE HERE ###\n",
    "\n",
    "# use all columns for this examples, including marcro economic variables\n",
    "cols_to_use = state_cols[:-1] + MEV_cols  \n",
    "if clf is not None:\n",
    "    clf.fit(df_train[cols_to_use].values, df_train.defaulter)\n",
    "roc_auc, ks, threshold = plot_roc(clf, df_test[cols_to_use], df_test.defaulter)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Submission successful, please check on the coursera grader page for the status\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.99369747899159666, 0.95685056631348198]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### GRADED PART (DO NOT EDIT) ###\n",
    "part_3 = np.array([roc_auc, ks]).squeeze()\n",
    "try:\n",
    "    part3 = \" \".join(map(repr, part_3))\n",
    "except TypeError:\n",
    "    part3 = repr(part_3)\n",
    "submissions[all_parts[2]]=part3\n",
    "grading.submit(COURSERA_EMAIL, COURSERA_TOKEN, assignment_key,all_parts[:3],all_parts,submissions)\n",
    "[roc_auc, ks]\n",
    "### GRADED PART (DO NOT EDIT) ###"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature importance from random forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nnJOCmZnl/j/cLc48AsJsyQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9b8e741eb8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "importance = []\n",
    "if clf is not None:\n",
    "    importance = clf.feature_importances_\n",
    "    importance = pd.DataFrame(importance, index=df_train[cols_to_use].columns, \n",
    "                              columns=[\"Importance\"])\n",
    "\n",
    "    importance[\"Std\"] = np.std([tree.feature_importances_\n",
    "                                for tree in clf.estimators_], axis=0)\n",
    "\n",
    "    x = range(importance.shape[0])\n",
    "    y = importance.iloc[:, 0] \n",
    "    yerr = importance.iloc[:, 1] \n",
    "\n",
    "    plt.bar(x, y, yerr=yerr, align=\"center\")\n",
    "    plt.xlabel('features')\n",
    "    plt.ylabel('Feature importance')\n",
    "    plt.title('Importance of different features')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Importance</th>\n",
       "      <th>Std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>log_TA</th>\n",
       "      <td>0.426811</td>\n",
       "      <td>0.246425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NI_to_TA</th>\n",
       "      <td>0.188403</td>\n",
       "      <td>0.223947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Equity_to_TA</th>\n",
       "      <td>0.047592</td>\n",
       "      <td>0.066365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPL_to_TL</th>\n",
       "      <td>0.167138</td>\n",
       "      <td>0.204541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>REO_to_TA</th>\n",
       "      <td>0.042497</td>\n",
       "      <td>0.096580</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ALLL_to_TL</th>\n",
       "      <td>0.023428</td>\n",
       "      <td>0.041117</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>core_deposits_to_TA</th>\n",
       "      <td>0.012578</td>\n",
       "      <td>0.013076</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>brokered_deposits_to_TA</th>\n",
       "      <td>0.002719</td>\n",
       "      <td>0.005775</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>liquid_assets_to_TA</th>\n",
       "      <td>0.014635</td>\n",
       "      <td>0.013077</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>loss_provision_to_TL</th>\n",
       "      <td>0.012047</td>\n",
       "      <td>0.017647</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NIM</th>\n",
       "      <td>0.016491</td>\n",
       "      <td>0.016764</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>term_spread</th>\n",
       "      <td>0.006083</td>\n",
       "      <td>0.008352</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>stock_mkt_growth</th>\n",
       "      <td>0.003786</td>\n",
       "      <td>0.007342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>real_gdp_growth</th>\n",
       "      <td>0.008559</td>\n",
       "      <td>0.015601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unemployment_rate_change</th>\n",
       "      <td>0.008981</td>\n",
       "      <td>0.018968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>treasury_yield_3m</th>\n",
       "      <td>0.004526</td>\n",
       "      <td>0.008093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bbb_spread</th>\n",
       "      <td>0.010086</td>\n",
       "      <td>0.017194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bbb_spread_change</th>\n",
       "      <td>0.003639</td>\n",
       "      <td>0.005959</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          Importance       Std\n",
       "log_TA                      0.426811  0.246425\n",
       "NI_to_TA                    0.188403  0.223947\n",
       "Equity_to_TA                0.047592  0.066365\n",
       "NPL_to_TL                   0.167138  0.204541\n",
       "REO_to_TA                   0.042497  0.096580\n",
       "ALLL_to_TL                  0.023428  0.041117\n",
       "core_deposits_to_TA         0.012578  0.013076\n",
       "brokered_deposits_to_TA     0.002719  0.005775\n",
       "liquid_assets_to_TA         0.014635  0.013077\n",
       "loss_provision_to_TL        0.012047  0.017647\n",
       "NIM                         0.016491  0.016764\n",
       "term_spread                 0.006083  0.008352\n",
       "stock_mkt_growth            0.003786  0.007342\n",
       "real_gdp_growth             0.008559  0.015601\n",
       "unemployment_rate_change    0.008981  0.018968\n",
       "treasury_yield_3m           0.004526  0.008093\n",
       "bbb_spread                  0.010086  0.017194\n",
       "bbb_spread_change           0.003639  0.005959"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "importance"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize decision trees as binary tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6AAAAFnCAYAAAC4tySUAAAABmJLR0QA/wD/AP+gvaeTAAAgAElE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\nI75ERERERETEdv8DxMxhTXzqsMIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the tree\n",
    "from sklearn.externals.six import StringIO\n",
    "from sklearn import tree\n",
    "import pydot\n",
    "\n",
    "if clf is not None:\n",
    "    try:\n",
    "        dot_data = StringIO()\n",
    "        tree_estimator = clf.estimators_[0]\n",
    "        tree.export_graphviz(single_tree.tree_, out_file=dot_data,\n",
    "                            feature_names=cols_to_use,\n",
    "                            proportion=True,\n",
    "                            max_depth=4)\n",
    "\n",
    "        graph = pydot.graph_from_dot_data(dot_data.getvalue())\n",
    "        # help(graph[0])\n",
    "\n",
    "        from IPython.display import Image, display\n",
    "\n",
    "        def viewPydot(pdot):\n",
    "            plt = Image(pdot.create_png())\n",
    "            display(plt)\n",
    "\n",
    "        viewPydot(graph[0])\n",
    "        # do not write graph to disk, instead write it to the screen\n",
    "        # dependencies on OS X: brew install graphviz\n",
    "        # dependencies on Ubuntu: sudo apt-get install graphviz\n",
    "        # graph[0].write_png('single_tree_graph' + '.png')\n",
    "    except: pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Boosted tree\n",
    "\n",
    "### Part 4\n",
    "As per http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingClassifier.html:\n",
    "\n",
    "GradientBoostingClassifier builds an additive model in a forward stage-wise fashion; it allows for the optimization of arbitrary differentiable loss functions. In each stage n_classes_ regression trees are fit on the negative gradient of the binomial or multinomial deviance loss function. Binary classification is a special case where only a single regression tree is induced.\n",
    "\n",
    "Use the following parameters to instantiate **GradientBoostingClassifier**:\n",
    "\n",
    "- The minimum number of samples required to split an internal node. Value = 5\n",
    "- The number of boosting stages to perform. Gradient boosting is fairly robust to over-fitting so a large number usually results in better performance. Value = 500\n",
    "- Learning rate. Learning rate shrinks the contribution of each tree by learning_rate. There is a trade-off between learning rate and number of boosting stages to perform. Value = 0.05\n",
    "- The maximum depth of the tree. Value = 4\n",
    "- The number of features to consider when looking for the best split. Value = 1.0 (be sure to use float). If float, then max features is a percentage and int(max_features * n_features) features are considered at each split.\n",
    "- random state (be sure to provide and int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC_auc =  0.99104859335\n",
      "KS_test =  0.919583485568\n",
      "AUC score: 0.991049 \n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XeYFFXWx/HvERAMmABdlbyAJBERUUxgBiOuCTMuyqKv\neQ0YF1FRcRXDgoAYEAUMa0DFRdewRkRMSBBBBBlAgREURJBw3j9uzdAME5qZ6anp7t/nefqZ6urq\n7lPdPX363lt1rrk7IiIiAFvEHYCIiFQeSgoiIpJPSUFERPIpKYiISD4lBRERyaekICIi+ZQUsoiZ\nzTGzI6LlG8xseNwxpVJF7qOZ9TWzp1L4+FPNrHO0bGb2uJktNbOJZnawmc1I1XNLdlFSqCTMrLuZ\nfWJmv5nZomj5YjOzVDyfu/d39wvK+jhm1tDM3MyqJqzrYWbrzGxFdJltZheV9blKiKOzmeUkriuv\nfUx4jjPNbFK0TwvN7HUzO6i8Hr847t7K3d+Nrh4EHAnUdfcO7v6+u+9RXs9lZk+Y2R/Rfv5sZm+a\nWfMC29Q1s6fNLDf6zE40s+MKbGNmdpmZTYm2yTGz58xsz/KKVcqfkkIlYGZ/Bx4A7gH+BOwC9AYO\nBLYs4j5VKizA0vnY3bd1922BU4ABZrZ33EGVlpldBdwP9Ce8P/WBwcCJMYTTAJjj7r+V9YESk3kB\nA6L3bndgPvBown12Aj4A/gBaAbWBgcAoMzsl4TEeAC4HLgN2ApoBLwHHljXu4qTB/0bl5u66xHgB\ntgd+A04uYbsngIeBcdH2RxD+ub4AfgXmAX0L3OccYC6QC9wIzAGOiG7rCzyVsO3+wEfAMuAroHPC\nbe8CtwEfAsuBN4Da0W0/AA6siC4dgR7ABwVimQicmXD9BGBq9HzvAi0SbmsRrVsWbXNCwm3HANOi\nOOYDVwPbAL8D6xPi2C1xH4GGUZznRTEvAW5MeNytgBHAUmA6cC2Qk/AerQBOLeb9Kfh6Pgf8CPwC\nvAe0Km4fovW1gVej/f4ZeB/YIrptTvSe9wRWAeuimG4FOufFGm27G/BvYDHwPXBZgTifB54ifG4u\nKOKzdnuBeH9LuH4bMCUvtoT11xE+bwY0jWLssBn/CzsBjwMLovfhpWh9YZ8nB5oU8b9xU/TaV0nY\n/iRgcrS8BdAH+I7wv/EssFPc3wWV5RJ7ANl+AboAa4GqJWz3RPQFc2D0oa4RfRnsGV1vA/wEdIu2\nbxl9aRwCVAfui55nk6RA+DWYG/3zb0HomsgF6kS3vxv9AzUjfHm+C9wV3dYw+getmhDrRv/EwL6E\nL7pm0fVm0T/vkUA1whfwLEKrqFq0fEN0/TDCl+ce0X0XAgdHyzsC7aLlziR8MRayj3lxPhLtw17A\naqJkBNwF/C96zLrAZDYkhRLfIzZNCn8Fakav/f3Alwm3FbUPdwJDotegGnAwYNFtcxLeu4Kvb/6+\nR+/fZ8At0evXGJgNHJ0Q5xqgW7TtVkV81m6PlrcBRgJfJdw+Abi1kPs1il7jPQgt3bmb+b/wGvBM\n9JpUAzoVtr/RuoJJoeD/xnfAkQnbPwf0iZaviPahbvT+DAVGx/1dUFku6j6KX21gibuvzVthZh+Z\n2TIz+93MDknY9mV3/9Dd17v7Knd/192/jq5PBkYDnaJtTwFedff33H01cDPhl3RhzgbGufu46LHe\nBCYRkkSex939W3f/nfDLqm0J+7V/tA8rCK2EkcDM6LbTgdfc/U13XwP8k/BFfQChxbItIen84e5v\nE349nxHddw3Q0sy2c/el7v55CXEUdKu7/+7uXxFaRHtF608D+kePmQM8mHCfWhR4j0ri7o+5+/Lo\nte8L7GVm25ewD2uAXYEG7r7Gw1jB5hYn25eQzPtFr99sQiLsnrDNx+7+UvRe/17E41xtZssICfkg\nQqszT21CYitoYcLttYrYplBmtivQFegdvSZr3P1/yd6fAv8bhP+FM6LHrkn4LI+Otv0boZWYk/D+\nnFJMV1pWUVKIXy5QO/ED6e4HuPsO0W2J79G8xDua2X5m9o6ZLTazXwi/zmpHN++WuL2H/ufcImJo\nAJwafYkvi74MDiJ8QeX5MWF5JeGLuzgT3H0HD/3SfyL0PfdPiG1uQmzro1h3z4s7WpdnbnQbwMmE\nf/C5ZvY/M+tYQhwFFbUfG71eBZY3eY+KY2ZVzOwuM/vOzH4l/MqHDe9NUftwD6GV9EY0ON8n2Z1K\n0ADYrcB7eQNhHKSwfSvKP6PPYENC11ziQPYSNv5s5Nk14fbcIrYpSj3gZ3dfuhn3SVRwn0YBfzGz\n6sBfgM/dPe8z1wB4MeH1mU7o6toFUVKoBD4mdGMkM2BZ8FfjKGAsUM/dtyd0PeQdrbSQ8I8GgJlt\nTfj1Vph5wMjoSzzvso2731WKmDbdwP0nQh/38dGqBYR/zLzYLIp1fnRbPTNL/GzWj27D3T919xOB\nnQmDls8mG0cJFhK6E/LUS1j+mNCP3y3JxzqT8H4eQRiPaBitNyh6H6KWxd/dvTHhtbrKzA7fzP2Y\nB3xf4L2s6e6Jrb6kXyt3/4EwWPyAmW0Vrf4vcHKB9whCa2se8C3wFlDXzNpvRtw7mdkOhdz2G7B1\n3hUz+1NhoRaIexrhx0RXwvsxqsBzdS3wGtVw9/lJxprRlBRi5u7LCIOFg83sFDPb1sy2MLO2hP7c\n4tQk/LpaZWYdCB/+PM8Dx5nZQWa2JdCPot/vp4Djzezo6FdujegQz7pFbJ9oMaFbqnFRG5hZLcJA\n39Ro1bPAsWZ2uJlVA/5OSIwfAZ8QvgSuNbNqFo7NPx4YY2ZbmtlZZrZ91O30K+EXHoTxlFoJXTSb\n61ngejPb0cx2By7Ju8HdfyH00Q8ys25mtnUUW1czG1DIY9WM9ieX8GWW10KiuH0ws+PMrEmUJPPW\nr9vk0Ys3EfjVzK4zs62i97O1me27mY+TL+pOXAD0ilYNBLYDHjWzP0WflzMIBzNc48FMwtFZo6PP\n0pbRdt0LawG5+0LgdcL/wY7R65vXdfoV0MrM2ppZDUJ3TzJGEY58OoQwppBnCHCHmTUAMLM6ZhbH\nUWSVkpJCJeDuA4CrCAOuiwhfcEMJR3N8VMxdLwb6mdlywpdW3q9m3H0q8H+Ef4yFhKM5cgp7EHef\nR/hlewPhS34ecA1JfD7cfSVwB/Bh1BzfP7qpo0XnKRCa54uBS6P7zCCMYzxE6Go4Hjg+6gP/g3Bk\nUtfotsHAue7+TfS45wBzom6Z3tHjEN0+GpgdxbFbSbEX0I/w+nxP+CX8POGLPW8/7yO8RzclvEaX\nEH7pF/Qk4VfqfMJRRhMK3F7oPhCO2Pkv4QCBj4HBvuHchKS4+zrC69k22pclwHBCi6Us7iEk6uru\nnkvoXqxB2L9cwmtzjrs/k3Cfy4B/AYMIBxp8R/hx8EoRz3EOYVzlG8L/wRXRPn1LeH/+SxiX+iDJ\nmEcTBuHfdvclCesfILSw34j+dyYA+yX5mBkv78gGEUlg4WS77u7eqcSNRTKIWgoihKNfzOzAqOtu\nD0KX1otxxyVS0XQIlkiwJaHLrhGhq2MMoetKJKuo+0hERPKp+0hERPKlXfdR7dq1vWHDhnGHISKS\nVj777LMl7l6npO3SLik0bNiQSZMmxR2GiEhaMbO5JW+l7iMREUmgpCAiIvmUFEREJJ+SgoiI5FNS\nEBGRfClLCmb2mIUJ6KcUcbuZ2YNmNsvMJptZu1TFIiIiyUllS+EJwjSGRelKqArZlFCS9+EUxiIi\nIklI2XkK7v6emTUsZpMTgSej6QYnmNkOZrZrVFddJFbDhsGoUSVvJ1IRaqz7jR3WLOZP+zfk/vtT\n+1xxnry2OxtPoZcTrdskKZhZL6IJPurXr18hwUnFqmxfwv+LZgfupMLZErO9l77N1d9eyG9Vt+fh\n/SaR6qHgOJOCFbKu0Op87j4MGAbQvn17VfDLQKNGwZdfQtu2cUcSdOoEZ54JvXqVvK1ISixbBtdc\nA8OHQ5MmMHwgAzul/tigOJNCDhvPg1uXMOWfFKOy/aIuL3kJ4d13445EpBJYtw4OOABmzIBrr4W+\nfWGrrUq8W3mIMymMBS4xszGEqfB+ybTxhFR8gWdqt0bbtuGXuUhWy82FnXaCKlXgjjugXj1o375C\nQ0hZUjCzvPlRa5tZDvAPoBqAuw8BxgHHALOAlcD5qYqlPG3OF30qvsDVrSGSgdzh6afh8svhrrvg\nwgvhpJNiCSWVRx+dUcLtTphYPq1sTt+3vsBFpETz5kHv3jBuHOy/Pxx4YKzhpF3p7MpAfd8iUi5G\nj4a//S2MIdx/P1xySeg6ipHKXCRp2DDo3Dm0EkREysWOO8J++8GUKaHrKOaEAGopJGXYsJDMYUOX\nkIjIZlu7FgYOhD/+gBtvhC5d4OijwQo7Qj8eSgpJyBtYHjpU4wMiUkpffQU9e8Jnn8Fpp4XBZbNK\nlRBA3UfFSuwy6tRJCUFESmH1arj55nBo6bx58NxzMGZMpUsGeZQUipF4pJG6jESkVGbOhLvvDl8i\n06bBKadU2oQA6j4qkY40EpHNtmIFvPwynHUWtG4N33wDjRvHHVVSlBQSFDwxrTLV4hGRNPHmm6Gv\nee5caNcOWrRIm4QAWZgUijsjueAZyOo2EpGkLV0KV18Njz0GzZqFL5QWLeKOarNlTVLISwbFlZ7Q\nGcgiUirr1oUzkb/9Fq6/Hm65BWrUiDuqUsmapJA3aKwvfhEpN0uWbChg178/1K8fuozSWFYdfZQ3\naKyEICJl4g5PPhm6iYYPD+u6dUv7hABZlhRERMps7lzo2hXOOy+MGRxySNwRlSslBRGRZD31VDjE\n9IMP4KGH4P33oXnzuKMqV1kzpiAiUmZ16oQB5aFDoUGDuKNJCSUFEZGirFkD994b/t58cyhed9RR\nlfqM5LJS95GISGG++CKUtb7++lCewj2sz+CEAEoKIiIbW7UKbrgB9t0XFiyAf/87TIaT4ckgj5KC\niEiiWbPgn/+Ec8+F6dPhL3+JO6IKpTEFEZEVK+DFF+Gcc8LRRTNmQKNGcUcVC7UURCS7jR8PrVqF\n8w6mTw/rsjQhgJKCiGSr3NyQCLp0ga23DuccpGEBu/Km7iMRyT55BexmzQpzJd90U9oWsCtvSgoi\nkj0WL4ZatUIBu7vvDiegadKUjaj7SEQynzs8/ngoYPfII2HdiScqIRRCSUFEMtucOeFM5L/+Ffbc\nEw49NO6IKjUlBRHJXCNHhkNMP/4YBg8OtfObNYs7qkpNYwoikrl22SWUth4yJEyAIyVSUhCRzLFm\nDQwYEI4uuuWWULzuqKPijiqtqPtIRDLD55+HekU33RTOSM4rYCebRUlBRNLb779Dnz7QoQP89FMo\nV/H001lTwK68pTQpmFkXM5thZrPMrE8ht9c3s3fM7Aszm2xmx6QyHhHJQLNnw333QY8eocR1t25x\nR5TWUpYUzKwKMAjoCrQEzjCzlgU2uwl41t33BroDg1MVj4hkkF9/hSeeCMutWsHMmTB8OOy4Y6xh\nZYJUthQ6ALPcfba7/wGMAU4ssI0D20XL2wMLUhiPiGSCcePCYaY9e24oYJehU2PGIZVJYXdgXsL1\nnGhdor7A2WaWA4wDLi3sgcysl5lNMrNJixcvTkWsIlLZLVkSSlsfeyzUrAkffqgCdimQyqRQ2ChP\nwcMBzgCecPe6wDHASDPbJCZ3H+bu7d29fZ06dVIQqohUankF7MaMCYeafv457L9/3FFlpFSep5AD\n1Eu4XpdNu4d6Al0A3P1jM6sB1AYWpTAuEUkXP/0EdeqEAnb//GfoJmrTJu6oMloqWwqfAk3NrJGZ\nbUkYSB5bYJsfgMMBzKwFUANQ/5BItnOHRx+FPfaAYcPCuuOPV0KoAClLCu6+FrgEGA9MJxxlNNXM\n+pnZCdFmfwcuNLOvgNFAD3edcSKS1WbPhiOOgAsuCFVMjzgi7oiySkrLXLj7OMIAcuK6WxKWpwEH\npjIGEUkjI0bAxReH7qIhQ+DCC2ELnWNbkVT7SEQqj912g8MOg4cfhrp1444mKykpiEh8/vgD7roL\n1q+Hvn3hyCPDRWKjdpmIxOPTT2GffeAf/wjjCBpOrBSUFESkYq1cCVdfHc4zWLoUxo6FJ59UAbtK\nQklBRCrW99/DQw+FQeSpU8OhplJpaExBRFLvl1/ghRfg/PNDAbtZs6BevZLvJxVOLQURSa3XXguJ\n4IIL4JtvwjolhEpLSUFEUmPxYjjrLDjuuFDS+uOPoXnzuKOSEqj7SETK37p1cNBBYfzg1lvDzGhb\nbhl3VJIEJQURKT8//gg77xzOSL73XmjYMMx9IGlD3UciUnbr18PQodCsWfgLodtICSHtlJgUzGwr\nM7vezIZE15uYWdfUhyYiaWHWLDj8cOjdG/bdF44+Ou6IpAySaSk8Rpgw56Do+gKgf8oiEpH08fjj\nsOeeYdKbRx6B//4XGjeOOyopg2SSQlN37w+sAXD3lRQ+q5qIZJv69UPLYNq0cMipzkpOe8kMNP8R\nzYjmAGbWCPgjpVGJSOW0ejXceWcYQ+jXL3QbHX543FFJOUqmpXAb8B+grpmNAN4BbkhpVCJS+Xzy\nSShgd+ut8MMPKmCXoUpsKbj762Y2CTiA0G10jbtrDmWRbPHbb3DzzXD//bD77vDqq3DssXFHJSmS\nzNFHb7j7Ynd/2d1fcvdFZvZGRQQnIpXA3LkweHA4umjqVCWEDFdkS8HMtgRqALuYWU02DC5vB9Sv\ngNhEJC7LlsHzz4fB45Ytw2GnmgktKxTXffR/wFXAzsBUNiSFX4EhKY5LROLy8stw0UWwaFEoVdG8\nuRJCFimy+8jdB7p7PeA6d6/v7vWiSyt3v78CYxSRirBoEXTvDt26QZ06MGGCCthloWQGmu83s+ZA\nS0J3Ut76UakMTEQq0Lp1cOCB4aii22+Ha6+FatXijkpiUGJSMLObgKOA5sB44GjgA0BJQSTdLVgA\nf/pTKGD3wAOhgF3LlnFHJTFK5jyF04FDgYXufg6wF6quKpLe1q+Hhx8O3UNDoiHCY45RQpCkksLv\n7r4OWBsdhfQjoOImIunq22/h0EPh4othv/2gq+pbygbJ/OL/wsx2IBTGm0Q4+ujzlEYlIqnx6KNw\nySVQowY89hj06KF6RbKRYpOCmRnQ192XAYPMbDywnbsrKYiko4YNQ8tg0CDYdde4o5FKqNik4O5u\nZq8C+0TXZ1VIVCJSPlavhttuC8u3364CdlKiZMYUJppZu5RHIiLl66OPoG1buOMOWLhQBewkKckk\nhYMIiWGGmX1uZl+YmbqPRCqrFSvg8svD2cgrV8J//hPGEjR2IElIZqC5W2kf3My6AA8AVYDh7n5X\nIducBvQlzNfwlbufWdrnExHCCWhDh8L//R/07w81a8YdkaSRZM5o/q40D2xmVYBBwJFADvCpmY11\n92kJ2zQFrgcOdPelZrZzaZ5LJOstXQrPPQe9eoVzDWbPht12izsqSUPJdB+VVgdglrvPdvc/gDHA\niQW2uRAY5O5LATRPg0gpvPhiSAQXXwwzZoR1SghSSqlMCrsD8xKu50TrEjUDmpnZh2Y2Iepu2oSZ\n9TKzSWY2afHixSkKVyTN/PgjnHoq/OUvoVTFxImwxx5xRyVpLqmkYGZ1zezQaLm6mW2TzN0KWVfw\n8IeqQFOgM3AGMDw6UW7jO7kPc/f27t6+Tp06yYQsktnWrYODD4ZXXgnjBhMnQjsdJChll0xBvL8C\nlwDbA38GGgCDgSNKuGsOUC/hel1gQSHbTHD3NcD3ZjaDkCQ+TSp6kWyTkxO6hqpUgQcfhEaNVN5a\nylUyLYXLgP0J5S1w928JE++U5FOgqZk1imZx6w6MLbDNS4Rie5hZbUJ30uzkQhfJIuvXw0MPhQTw\n8MNhXdeuSghS7pJJCquigWIg/6iiEg94dve1hBbGeGA68Ky7TzWzfmZ2QrTZeCDXzKYB7wDXuHvu\n5u6ESEb75hs45BC47LJw7sFxx8UdkWSwZM5T+NDMrgVqROMK/we8msyDu/s4YFyBdbckLDthys+r\nko5YJJsMHx4K2G29NYwYAeeco5PQJKWSaSlcCywHvgEuB94CbkxlUCIS+fOf4fjjYfp0OPdcJQRJ\nuWRaCscQzkZ+ONXBiGS9VaugX7+w3L9/mPfg0EPjjUmySjIthdOAWWb2uJkdHY0piEh5+/DDUMDu\nzjth8WIVsJNYlJgUoik4mwGvAH8FZpvZkFQHJpI1li+HSy8N5x2sXg3jx8Mjj6irSGKR1Mlr7r4a\neBl4gnCo6WkpjEkku+TkhAHlSy+Fr7+Go46KOyLJYiUmBTM7wsyGA98BZwNPAn9KdWAiGS03d8P5\nBi1ahAJ2DzwA224bb1yS9ZIZaO5NKGZ3qbv/nuJ4RDKbO/z736Gs9c8/w2GHhXpFmhpTKolkxhRO\ncffnlRBEymjhQjj55FDErl49mDRJBeyk0imypWBm/3P3Tma2lI0L2RnhvLOdUh6dSKbIK2A3fz4M\nGABXXglVk2moi1Ss4j6VeQdH166IQEQy0rx5sPvuoYDdoEGhgF2zZnFHJVKkIruP3H19tPiou69L\nvACPVkx4Imlq3bpQxTSxgN3RRyshSKWXTPu1TeKV6OS1fVMTjkgGmD4devaEjz8OlUyPPz7uiESS\nVmRLwcyui8YT2pjZz9FlKbCYAkXuRCQybFg4K/nbb2HkSHjtNahfP+6oRJJW3NFHA4A6wMDobx2g\ntrvv5O7XVERwImmnaVM46SSYNg3OPltnJUvaKa77qIm7zzSzkUCrvJUWfcjdfXKKYxOp/H7/Hfr2\nDV/+d92lAnaS9opLCn2AnsCgQm5z4JCURCSSLt57Dy64AGbOhN69w4lpahlImisyKbh7z+jvwRUX\njkga+PVX6NMnHFXUuDG89VY4M1kkAyRT++gvZlYzWu5jZs+a2V6pD02kklqwAJ54Aq66CiZPVkKQ\njJJMldS+7r7czA4AjgeeAYamNiyRSmbJEhg8OCw3bw7ffw/33gvbbBNvXCLlLJmksC76exww2N3/\nDVRPXUgilYg7PPMMtGwJV1wRDjUF2GWXeOMSSZFkksJCMxsEdAfGmdmWSd5PJL0tWADdukH37tCg\nAXz2mc5IloyXzBnNpxHmaX7I3Zea2W6EI5NEMte6dXDIIaGA3T//CZdfrgJ2khVK/JS7+wozmwZ0\nNrPOwPvu/nrKIxOJw9y5ULduKGA3eHA4uqhJk7ijEqkwyRx9dAnwLFA/ujxrZhenOjCRCrVuHdx3\nX5gFLa+A3VFHKSFI1kmmPdwL6ODuKwDMrD/wETA4lYGJVJgpU0IBu4kT4bjjwjiCSJZKZsDYgDUJ\n19dE60TS35Ah0K5dmCN51CgYOzZ0H4lkqWRaCiOBCWb2b0Iy6AaMSGlUIqmWV5KiRYswPeb990Od\nOnFHJRK7ZAaaB5jZO0BeuYve7v5pasMSSZGVK+GWW8JA8t13Q6dO4SIiQPLnG6yOLr9Hf0XSz7vv\nQps24UzkFStCa0FENpLM0Uc3AqOBXYG6wCgzuz7VgYmUm19+gb/9bUNJ67ffDvMlq6KpyCaSGVM4\nG9jH3VcCmNkdwGfAnakMTKTcLFwITz0FV18Nt94KW28dd0QilVYy3Udz2Th5VAVmJ/PgZtbFzGaY\n2SwzK/IsaDM7xczczNon87giJVq8GB56KCw3bw5z5sA99yghiJQgmaSwEphqZsPN7BHga2CZmd1n\nZvcVdSczq0KYoKcr0BI4w8xaFrJdTeAy4JPS7IDIRtzDoaUtWsDf/76hgJ2OLBJJSjLdR69FlzwT\nknzsDsAsd58NYGZjgBOBaQW2u40wH/TVST6uSOHmzYOLLoLXXoP99oNHH1UBO5HNlMwhqY+W8rF3\nB+YlXM8B9kvcwMz2Buq5+6tmVmRSMLNehDOrqV+/finDkYy2di107gw//ggDB8Kll4bDTkVks6Sy\n7GNhh3bkHwNoZlsAA4EeJT2Quw8DhgG0b99exxHKBnPmQL16oYLp0KGhgF3jxnFHJZK2UjkvQg5Q\nL+F6XWBBwvWaQGvgXTObA+wPjNVgsyRl7dpQ0rpFiw0zoh1xhBKCSBkl3VIws+ruvjknrn0KNDWz\nRsB8wiQ9Z+bd6O6/ALUTHv9d4Gp3n7QZzyHZaPLkUMBu0iQ48UQ4+eS4IxLJGMmcvNbBzL4GZkbX\n9zKzh0q6n7uvBS4BxgPTgWfdfaqZ9TOzE8oYt2SrwYNhn33CvAfPPAMvvgi77RZ3VCIZI5mWwoOE\n+ZlfAnD3r8zs0GQe3N3HAeMKrLuliG07J/OYkqXyCti1bh2mxxw4EGrXLvl+IrJZkkkKW7j7XNu4\nJMC6FMUjsrHffoObbgoDyffcE6bIPOSQuKMSyVjJDDTPM7MOgJtZFTO7Avg2xXGJwFtvwZ57hrLW\nq1ergJ1IBUgmKVwEXEWYivMnwlFCF6UyKMlyy5bBBReEo4mqVoX33oMHH1QBO5EKkMzJa4sIRw6J\nVIyffoIxY+C66+Af/4Cttoo7IpGsUWJSiOodbdJud/deKYlIslNeIrj8cthjj3BSmgaSRSpcMgPN\n/01YrgGcxMblK0RKzx2efjokgxUr4JhjoGlTJQSRmCTTffRM4nUzGwm8mbKIJHv88AP07g2vvw4d\nO4YCdk2bxh2VSFYrTe2jRkCD8g5EskxeAbtFi8Ig8sUXq4CdSCWQzJjCUjaMKWwB/AwUOWGOSLFm\nz4YGDcJRRY88An/+MzRsGHdUIhIp9pBUC2es7QXUiS47untjd3+2IoKTDLJ2Ldx9N7RsGeZHBjj8\ncCUEkUqm2KTg7g686O7roovOHpLN9+WXYdKbPn3CQPKpp8YdkYgUIZmT1yaaWbuURyKZ6V//gn33\nhfnz4fnn4YUXYNdd445KRIpQ5JiCmVWNKp0eBFxoZt8BvxEmz3F3V6KQouUVsGvTBs46C+67D3ba\nKe6oRKQExQ00TwTaAd0qKBbJBCtWwI03QrVqYRIcFbATSSvFJQUDcPfvKigWSXdvvAG9eoXzDy69\ndENrQUTSRnFJoY6ZXVXUje5+XwrikXS0dClcdRU88UQoUfHee3DQQXFHJSKlUFxSqAJsS9RiECnS\nokVhEPkh5AteAAATcElEQVT66+GWW6BGjbgjEpFSKi4pLHT3fhUWiaSXH3+E0aPhyis3FLCrVSvu\nqESkjIo7JFUtBNmUO4wYEU5Cu/56mDkzrFdCEMkIxSWFwyssCkkPc+ZAly7Qo0dICl9+qQJ2Ihmm\nyO4jd/+5IgORSm7tWjj0UFiyJJSp6N0btkjm3EcRSSelqZIq2WTWLGjUKBSwe+wxaNw4FLQTkYyk\nn3pSuDVroH9/aNVqQwG7Qw9VQhDJcGopyKY+/xx69gxjBqeeCqefHndEIlJB1FKQjT34IHToEA45\nfeEFePZZ2GWXuKMSkQqipCBBXlX0vfeGc8+FadPgpJPijUlEKpy6j7Ld8uXhfIPq1eHee+Hgg8NF\nRLKSWgrZ7D//gdatYfDg0FLQHEoiWU9JIRvl5sJ550HXrrDNNvDhh2G+A1U0Fcl6SgrZKDcXXnwR\nbr4ZvvgCOnaMOyIRqSRSmhTMrIuZzTCzWWbWp5DbrzKzaWY22czeMjMdBJ8qCxeGSW/coVkzmDsX\n+vULYwkiIpGUJQUzqwIMAroCLYEzzKxlgc2+ANq7exvgeWBAquLJWu7hTOQWLULLYNassH7HHeON\nS0QqpVS2FDoAs9x9trv/AYwBTkzcwN3fcfeV0dUJQN0UxpN9vv8ejjoqnIi2117w1VcqYCcixUrl\nIam7A/MSrucA+xWzfU/g9cJuMLNeQC+A+vXrl1d8mW3tWjjssDB+8PDDYZpMFbATkRKkMikUdihL\nocc8mtnZQHugU2G3u/swYBhA+/btddxkcWbODEXrqlaFxx+HP/8Z6tWLOyoRSROp/OmYAyR+G9UF\nFhTcyMyOAG4ETnD31SmMJ7OtWQO33x7OO/jXv8K6zp2VEERks6SypfAp0NTMGgHzge7AmYkbmNne\nwFCgi7svSmEsmW3SpDBuMHkydO8OZ5wRd0QikqZS1lJw97XAJcB4YDrwrLtPNbN+ZnZCtNk9wLbA\nc2b2pZmNTVU8GeuBB2C//cLkNy+/HOZN3nnnuKMSkTSV0tpH7j4OGFdg3S0Jy0ek8vkzmns4A7l9\n+9BKGDAAdtgh7qhEJM2pIF66+fVXuO46qFEDBg6EAw8MFxGRcqBjFNPJuHFhJrRhw8LRRSpgJyLl\nTEkhHSxZAmefDcceC9tvDx99BPfcowJ2IlLulBTSwdKl8Mor8I9/hKky9yvuHEARkdLTmEJlNX8+\nPP00XHNNKE0xd64GkkUk5dRSqGzc4ZFHoGVL6NsXvvsurFdCEJEKoKRQmXz3HRx+eKhT1K5dOBmt\nSZO4oxKRLKLuo8pi7dqQEH7+GYYOhQsuUAE7EalwSgpxmzEjFK2rWhVGjAjLdVVBXETioZ+icfnj\nD7j1VthzTxg0KKzr1EkJQURipZZCHCZODKUppkyBM8+Es86KOyIREUAthYp3//3QseOGcw+efhpq\n1447KhERQEmh4uSVpOjQAS68EKZOheOOizcmEZEC1H2Uar/8AtdeC1ttFVoJBxwQLiIilZBaCqn0\nyivhJLThw6F6dRWwE5FKT0khFRYvDgPIJ5wAtWrBhAlw990qYCcilZ6SQir88ksoc33rrWGqzH33\njTsiEZGkaEyhvMybB089BX36hNIUc+eGMtciImlELYWyWr8ehgwJk9/cfvuGAnZKCCKShpQUymLm\nTDjsMLjoonCo6ddfq4CdiKQ1dR+V1tq1cOSRsGwZPPoonH++BpJFJO0pKWyu6dPDpDdVq8LIkaGA\n3W67xR2VSKW0Zs0acnJyWLVqVdyhZI0aNWpQt25dqlWrVqr7Kykka/Vq6N8/XO65B664Ag4+OO6o\nRCq1nJwcatasScOGDTG1pFPO3cnNzSUnJ4dGjRqV6jE0ppCMCRPCpDf9+sEZZ8A558QdkUhaWLVq\nFbVq1VJCqCBmRq1atcrUMlNSKMm994ayFMuXh3MPnnwynJAmIklRQqhYZX29lRSKsn59+NuxI/Tu\nHcpcd+0ab0wiIimmpFDQsmVhroPLLw/XDzgABg+G7baLNy4RKbUXX3wRM+Obb77JX/fuu+9yXIFK\nxT169OD5558HwiB5nz59aNq0Ka1bt6ZDhw68/vrrZY7lzjvvpEmTJuyxxx6MHz++0G3efvtt2rVr\nR+vWrTnvvPNYu3YtAEuXLuWkk06iTZs2dOjQgSlTppQ5noKUFBK99FIoYDdiBNSsqQJ2Ihli9OjR\nHHTQQYwZMybp+9x8880sXLiQKVOmMGXKFF555RWWL19epjimTZvGmDFjmDp1Kv/5z3+4+OKLWbdu\n3UbbrF+/nvPOO48xY8YwZcoUGjRowIgRIwDo378/bdu2ZfLkyTz55JNcnvfjtRzp6COARYvgkkvg\nueegbVt49dUwsCwi5eaKK+DLL8v3Mdu2DRXpi7NixQo+/PBD3nnnHU444QT69u1b4uOuXLmSRx55\nhO+//57q1asDsMsuu3DaaaeVKd6XX36Z7t27U716dRo1akSTJk2YOHEiHTt2zN8mNzeX6tWr06xZ\nMwCOPPJI7rzzTnr27Mm0adO4/vrrAWjevDlz5szhp59+YpdddilTXInUUgD49Vd48024444wVaYS\ngkjGeOmll+jSpQvNmjVjp5124vPPPy/xPrNmzaJ+/fpsl0S38ZVXXknbtm03udx1112bbDt//nzq\n1auXf71u3brMnz9/o21q167NmjVrmDRpEgDPP/888+bNA2CvvfbihRdeAGDixInMnTuXnJycEmPc\nHNnbUvjhh3Dy2Q03hNIUP/wQuoxEJCVK+kWfKqNHj+aKK64AoHv37owePZp27doVeZTO5h69M3Dg\nwKS39UK6pAs+n5kxZswYrrzySlavXs1RRx1F1arhq7pPnz5cfvnltG3blj333JO99947/7byktKk\nYGZdgAeAKsBwd7+rwO3VgSeBfYBc4HR3n5PKmPIL2F13XVg+/fSQFJQQRDJObm4ub7/9NlOmTMHM\nWLduHWbGgAEDqFWrFkuXLt1o+59//pnatWvTpEkTfvjhB5YvX07NEr4brrzySt55551N1nfv3p0+\nffpstK5u3br5v/ohnNy3WyEVETp27Mj7778PwBtvvMG3334LwHbbbcfjjz8OhATTqFGjUp+kViR3\nT8mFkAi+AxoDWwJfAS0LbHMxMCRa7g48U9Lj7rPPPl4anTq5n73vN+4HH+wO7kce6f7996V6LBFJ\nzrRp02J9/iFDhnivXr02WnfIIYf4e++956tWrfKGDRvmxzhnzhyvX7++L1u2zN3dr7nmGu/Ro4ev\nXr3a3d0XLFjgI0eOLFM8U6ZM8TZt2viqVat89uzZ3qhRI1+7du0m2/3000/u7r5q1So/7LDD/K23\n3nJ396VLl+bHM2zYMD/nnHMKfZ7CXndgkifx3Z3KMYUOwCx3n+3ufwBjgBMLbHMiMCJafh443FJ0\npku7Nmt54JujQyXTxx+H8eOhYcNUPJWIVBKjR4/mpJNO2mjdySefzKhRo6hevTpPPfUU559/Pm3b\ntuWUU05h+PDhbB+Vvb/99tupU6cOLVu2pHXr1nTr1o06deqUKZ5WrVpx2mmn0bJlS7p06cKgQYOo\nUqUKAMcccwwLFiwA4J577qFFixa0adOG448/nsMOOwyA6dOn06pVK5o3b87rr7/OAw88UKZ4CmOe\nosMuzewUoIu7XxBdPwfYz90vSdhmSrRNTnT9u2ibJQUeqxfQC6B+/fr7zJ07t3RBffBBKGC3666l\nu7+IbJbp06fTokWLuMPIOoW97mb2mbu3L+m+qWwpFPaLv2AGSmYb3H2Yu7d39/ZlytQHHaSEICJS\njFQmhRygXsL1usCCorYxs6rA9sDPKYxJRESKkcqk8CnQ1MwamdmWhIHksQW2GQucFy2fArztqerP\nEpFY6F+6YpX19U5ZUnD3tcAlwHhgOvCsu081s35mdkK02aNALTObBVwF9Cn80UQkHdWoUYPc3Fwl\nhgri0XwKNWrUKPVjpGygOVXat2/veWf6iUjlppnXKl5RM68lO9CcvWc0i0jKVatWrfxPrpKUUu0j\nERHJp6QgIiL5lBRERCRf2g00m9lioJSnNFMbWFLiVplF+5wdtM/ZoSz73MDdSzz7N+2SQlmY2aRk\nRt8zifY5O2ifs0NF7LO6j0REJJ+SgoiI5Mu2pDAs7gBioH3ODtrn7JDyfc6qMQURESletrUURESk\nGEoKIiKSLyOTgpl1MbMZZjbLzDapvGpm1c3smej2T8ysYcVHWb6S2OerzGyamU02s7fMrEEccZan\nkvY5YbtTzMzNLO0PX0xmn83stOi9nmpmoyo6xvKWxGe7vpm9Y2ZfRJ/vY+KIs7yY2WNmtiiambKw\n283MHoxej8lm1q5cA0hmIud0ugBVgO+AxsCWwFdAywLbXAwMiZa7A8/EHXcF7POhwNbR8kXZsM/R\ndjWB94AJQPu4466A97kp8AWwY3R957jjroB9HgZcFC23BObEHXcZ9/kQoB0wpYjbjwFeJ8xcuT/w\nSXk+fya2FDoAs9x9trv/AYwBTiywzYnAiGj5eeBwMytsatB0UeI+u/s77r4yujqBMBNeOkvmfQa4\nDRgAZELt5mT2+UJgkLsvBXD3RRUcY3lLZp8d2C5a3p5NZ3hMK+7+HsXPQHki8KQHE4AdzKzc5hnO\nxKSwOzAv4XpOtK7QbTxMBvQLUKtCokuNZPY5UU/CL410VuI+m9neQD13f7UiA0uhZN7nZkAzM/vQ\nzCaYWZcKiy41ktnnvsDZZpYDjAMurZjQYrO5/++bJRPnUyjsF3/B426T2SadJL0/ZnY20B7olNKI\nUq/YfTazLYCBQI+KCqgCJPM+VyV0IXUmtAbfN7PW7r4sxbGlSjL7fAbwhLvfa2YdgZHRPq9PfXix\nSOn3Vya2FHKAegnX67JpczJ/GzOrSmhyFtdcq+yS2WfM7AjgRuAEd19dQbGlSkn7XBNoDbxrZnMI\nfa9j03ywOdnP9svuvsbdvwdmEJJEukpmn3sCzwK4+8dADULhuEyV1P97aWViUvgUaGpmjcxsS8JA\n8tgC24wFzouWTwHe9mgEJ02VuM9RV8pQQkJI935mKGGf3f0Xd6/t7g3dvSFhHOUEd0/nuVyT+Wy/\nRDioADOrTehOml2hUZavZPb5B+BwADNrQUgKiys0yoo1Fjg3Ogppf+AXd19YXg+ecd1H7r7WzC4B\nxhOOXHjM3aeaWT9gkruPBR4lNDFnEVoI3eOLuOyS3Od7gG2B56Ix9R/c/YTYgi6jJPc5oyS5z+OB\no8xsGrAOuMbdc+OLumyS3Oe/A4+Y2ZWEbpQe6fwjz8xGE7r/akfjJP8AqgG4+xDCuMkxwCxgJXB+\nuT5/Gr92IiJSzjKx+0hEREpJSUFERPIpKYiISD4lBRERyaekICIi+ZQUpNIys3Vm9mXCpWEx2zYs\nqqpkRTOz9mb2YLTc2cwOSLitt5mdW4GxtE33qqFSsTLuPAXJKL+7e9u4g9hc0QlyeSfJdQZWAB9F\ntw0p7+czs6pRDa/CtCWUNRlX3s8rmUktBUkrUYvgfTP7PLocUMg2rcxsYtS6mGxmTaP1ZyesH2pm\nVQq57xwzuzvabqKZNYnWN7AwD0XefBT1o/WnmtkUM/vKzN6L1nU2s1ejlk1v4MroOQ82s75mdrWZ\ntTCziQX2a3K0vI+Z/c/MPjOz8YVVwDSzJ8zsPjN7B7jbzDqY2UcW5hT4yMz2iM4A7gecHj3/6Wa2\njYV6/Z9G2xZWWVayWdy1w3XRpagL4YzcL6PLi9G6rYEa0XJTwlmtAA2J6s8DDwFnRctbAlsBLYBX\ngGrR+sHAuYU85xzgxmj5XODVaPkV4Lxo+a/AS9Hy18Du0fIO0d/OCffrC1yd8Pj516P9ahwtXwfc\nRDhz9SOgTrT+dMJZvAXjfAJ4FagSXd8OqBotHwH8O1ruAfwr4X79gbPz4gW+BbaJ+73WpfJc1H0k\nlVlh3UfVgH+ZWVtC0mhWyP0+Bm40s7rAC+4+08wOB/YBPo3KfGwFFFUDanTC34HRckfgL9HySMIc\nDQAfAk+Y2bPAC5uzc4QibqcBdxG+/E8H9iAU8nszirMKUFRdm+fcfV20vD0wImoVOVFZhEIcBZxg\nZldH12sA9YHpmxm7ZCglBUk3VwI/AXsRuj83mTzH3UeZ2SfAscB4M7uAUG54hLtfn8RzeBHLm2zj\n7r3NbL/oub6MklWyniHUonohPJTPNLM9ganu3jGJ+/+WsHwb8I67nxR1W71bxH0MONndZ2xGnJJF\nNKYg6WZ7YKGHWvnnEH5Jb8TMGgOz3f1BQkXJNsBbwClmtnO0zU5W9DzVpyf8/Tha/ogNhRPPAj6I\nHufP7v6Ju98CLGHjksYAywllvDfh7t8RWjs3ExIEhFLXdSzMC4CZVTOzVkXEmWh7YH603KOY5x8P\nXGpRM8RC9VyRfEoKkm4GA+eZ2QRC19FvhWxzOjDFzL4EmhOmLpxG6LN/IxrQfRMoagrD6lFL43JC\nywTgMuD86L7nRLcB3GNmX0eHw75HmEM40SvASXkDzYU81zPA2WyYD+APQjn3u83sK8K4wyaD6YUY\nANxpZh+ycaJ8B2iZN9BMaFFUAyZHMd+WxGNLFlGVVJEEFibkae/uS+KORSQOaimIiEg+tRRERCSf\nWgoiIpJPSUFERPIpKYiISD4lBRERyaekICIi+f4fJmjPtJ7v4gAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9b8e828668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gbt = None\n",
    "roc_auc = 0.\n",
    "ks = 0.\n",
    "### START CODE HERE ### (≈ 7 lines of code)\n",
    "# ... please set random_state=42 in a random forest classifier of sklearn\n",
    "gbt = GradientBoostingClassifier(random_state =42,min_samples_split =5, \n",
    "                                 max_depth =4, max_features = 1.0 ,learning_rate = 0.05, n_estimators = 500)\n",
    "### END CODE HERE ###\n",
    "if gbt is not None:\n",
    "    gbt.fit(df_train[cols_to_use].values,df_train.defaulter)\n",
    "    roc_auc, ks, threshold = plot_roc(gbt, df_test[cols_to_use], df_test.defaulter)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Submission successful, please check on the coursera grader page for the status\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.99104859335038364, 0.91958348556814029]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### GRADED PART (DO NOT EDIT) ###\n",
    "part_4 = np.array([roc_auc, ks]).squeeze()\n",
    "try:\n",
    "    part4 = \" \".join(map(repr, part_4))\n",
    "except TypeError:\n",
    "    part4 = repr(part_4)\n",
    "submissions[all_parts[3]]=part4\n",
    "grading.submit(COURSERA_EMAIL, COURSERA_TOKEN, assignment_key,all_parts[:4],all_parts,submissions)\n",
    "[roc_auc, ks]\n",
    "### GRADED PART (DO NOT EDIT) ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC_auc =  0.997111306128\n",
      "KS_test =  0.941385110429\n",
      "AUC score: 0.997111 \n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKYiISAElBRERKaCkICIiBf4fkECNB2w5BRoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9b8e85cfd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "roc_auc = 0.\n",
    "ks = 0.\n",
    "if gbt is not None:\n",
    "    gbt.fit(df_train[cols_to_use].values, df_train.default_within_1Y)\n",
    "    roc_auc, ks, threshold = plot_roc(gbt, df_test[cols_to_use], df_test.default_within_1Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Submission successful, please check on the coursera grader page for the status\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.99711130612841892, 0.94138511042855055]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### GRADED PART (DO NOT EDIT) ###\n",
    "part_5 = np.array([roc_auc, ks]).squeeze()\n",
    "try:\n",
    "    part5 = \" \".join(map(repr, part_5))\n",
    "except TypeError:\n",
    "    part5 = repr(part_5)\n",
    "submissions[all_parts[4]]=part5\n",
    "grading.submit(COURSERA_EMAIL, COURSERA_TOKEN, assignment_key,all_parts[:5],all_parts,submissions)\n",
    "[roc_auc, ks]\n",
    "### GRADED PART (DO NOT EDIT) ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "coursera": {
   "course_slug": "machine-learning-in-finance"
  },
  "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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}