lending_club_v2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Predicting if a loan will be paid off or not\n",
    "** Can we build a machine learning model that can accurately predict if a borrower will pay off their loan on time or not? **\n",
    "\n",
    "The objective is to get a model that has:\n",
    "\n",
    "* **TPR > 70%** (TPR: true positive rate - proportion of positives that are correctly identified)\n",
    "* **FPR < 7%**  (FPR: false positive rate - proportion of positives that are incorrect)\n",
    "\n",
    "In other words, we want to predicts as many 1s as possible (> 70%) making the fewest mistakes as possible (< 0.7%) on those predictions.\n",
    "\n",
    "### INDEX\n",
    "\n",
    "1. INTRODUCTION\n",
    "2. PREPARE DATA\n",
    "3. MODELS\n",
    "3.1 FEATURE SELECTION/ OVERSAMPLING\n",
    "3.2 LG\n",
    "3.3 RF\n",
    "\n",
    "\n",
    "\n",
    "### INTRODUCTION\n",
    "\n",
    "Credit has played a key role in the economy for centuries and some form of credit has existed since the beginning of commerce. We'll be working with financial lending data from Lending Club. Lending Club is a marketplace for personal loans that matches borrowers who are seeking a loan with investors looking to lend money and make a return. You can read more about their marketplace here.\n",
    "\n",
    "Each borrower fills out a comprehensive application, providing their past financial history, the reason for the loan, and more. Lending Club evaluates each borrower's credit score using past historical data (and their own data science process!) and assigns an interest rate to the borrower. The interest rate is the percent in addition to the requested loan amount the borrower has to pay back. You can read more about the interest rate that Lending Club assigns here. Lending Club also tries to verify each piece of information the borrower provides but it can't always verify all of the information (usually for regulation reasons).\n",
    "A higher interest rate means that the borrower is riskier and more unlikely to pay back the loan while a lower interest rate means that the borrower has a good credit history is more likely to pay back the loan. The interest rates range from 5.32% all the way to 30.99% and each borrower is given a grade according to the interest rate they were assigned. If the borrower accepts the interest rate, then the loan is listed on the Lending Club marketplace.\n",
    "\n",
    "Investors are primarily interested in receiving a return on their investments. Approved loans are listed on the Lending Club website, where qualified investors can browse recently approved loans, the borrower's credit score, the purpose for the loan, and other information from the application. Once they're ready to back a loan, they select the amount of money they want to fund. Once a loan's requested amount is fully funded, the borrower receives the money they requested minus the origination fee that Lending Club charges.\n",
    "\n",
    "The borrower then makes monthly payments back to Lending Club either over 36 months or over 60 months. Lending Club redistributes these payments to the investors. This means that investors don't have to wait until the full amount is paid off to start to see money back. If a loan is fully paid off on time, the investors make a return which corresponds to the interest rate the borrower had to pay in addition the requested amount. Many loans aren't completely paid off on time, however, and some borrowers default on the loan.\n",
    "\n",
    "While Lending Club has to be extremely savvy and rigorous with their credit modelling, investors on Lending Club need to be equally as savvy about determining which loans are more likely to be paid off. While at first, you may wonder why investors would put money into anything but low interest loans. The incentive investors have to back higher interest loans is, well, the higher interest! If investors believe the borrower can pay back the loan, even if he or she has a weak financial history, then investors can make more money through the larger additional amount the borrower has to pay.\n",
    "\n",
    "Most investors use a portfolio strategy to invest small amounts in many loans, with healthy mixes of low, medium, and interest loans. In this project, we'll focus on the mindset of a conservative investor who only wants to invest in the loans that have a good chance of being paid off on time. To do that, we'll need to first understand the features in the dataset and then experiment with building machine learning models that reliably predict if a loan will be paid off or not.\n",
    "\n",
    "Data dictionary: https://docs.google.com/spreadsheets/d/191B2yJ4H1ZPXq0_ByhUgWMFZOYem5jFz0Y3by_7YBY4/edit#gid=2081333097"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### START\n",
    "\n",
    "First Let's load the data and do some quick checks:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.cross_validation import KFold\n",
    "from sklearn.cross_validation import cross_val_predict\n",
    "from sklearn import pipeline, grid_search\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.cross_validation import train_test_split\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import auc\n",
    "from sklearn.metrics import roc_curve\n",
    "\n",
    "from nltk.stem.snowball import SnowballStemmer\n",
    "from nltk.tokenize import word_tokenize\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.decomposition import TruncatedSVD\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.tokenize import RegexpTokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\lib\\site-packages\\IPython\\core\\interactiveshell.py:2723: DtypeWarning: Columns (0,47) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "loan_df = pd.read_csv(\".\\\\data\\\\LoanStats3a.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(42538, 111)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>member_id</th>\n",
       "      <th>loan_amnt</th>\n",
       "      <th>funded_amnt</th>\n",
       "      <th>funded_amnt_inv</th>\n",
       "      <th>term</th>\n",
       "      <th>int_rate</th>\n",
       "      <th>installment</th>\n",
       "      <th>grade</th>\n",
       "      <th>sub_grade</th>\n",
       "      <th>...</th>\n",
       "      <th>num_tl_90g_dpd_24m</th>\n",
       "      <th>num_tl_op_past_12m</th>\n",
       "      <th>pct_tl_nvr_dlq</th>\n",
       "      <th>percent_bc_gt_75</th>\n",
       "      <th>pub_rec_bankruptcies</th>\n",
       "      <th>tax_liens</th>\n",
       "      <th>tot_hi_cred_lim</th>\n",
       "      <th>total_bal_ex_mort</th>\n",
       "      <th>total_bc_limit</th>\n",
       "      <th>total_il_high_credit_limit</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1077501</td>\n",
       "      <td>1296599.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>4975.0</td>\n",
       "      <td>36 months</td>\n",
       "      <td>10.65%</td>\n",
       "      <td>162.87</td>\n",
       "      <td>B</td>\n",
       "      <td>B2</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1077430</td>\n",
       "      <td>1314167.0</td>\n",
       "      <td>2500.0</td>\n",
       "      <td>2500.0</td>\n",
       "      <td>2500.0</td>\n",
       "      <td>60 months</td>\n",
       "      <td>15.27%</td>\n",
       "      <td>59.83</td>\n",
       "      <td>C</td>\n",
       "      <td>C4</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1077175</td>\n",
       "      <td>1313524.0</td>\n",
       "      <td>2400.0</td>\n",
       "      <td>2400.0</td>\n",
       "      <td>2400.0</td>\n",
       "      <td>36 months</td>\n",
       "      <td>15.96%</td>\n",
       "      <td>84.33</td>\n",
       "      <td>C</td>\n",
       "      <td>C5</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1076863</td>\n",
       "      <td>1277178.0</td>\n",
       "      <td>10000.0</td>\n",
       "      <td>10000.0</td>\n",
       "      <td>10000.0</td>\n",
       "      <td>36 months</td>\n",
       "      <td>13.49%</td>\n",
       "      <td>339.31</td>\n",
       "      <td>C</td>\n",
       "      <td>C1</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1075358</td>\n",
       "      <td>1311748.0</td>\n",
       "      <td>3000.0</td>\n",
       "      <td>3000.0</td>\n",
       "      <td>3000.0</td>\n",
       "      <td>60 months</td>\n",
       "      <td>12.69%</td>\n",
       "      <td>67.79</td>\n",
       "      <td>B</td>\n",
       "      <td>B5</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 111 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        id  member_id  loan_amnt  funded_amnt  funded_amnt_inv        term  \\\n",
       "0  1077501  1296599.0     5000.0       5000.0           4975.0   36 months   \n",
       "1  1077430  1314167.0     2500.0       2500.0           2500.0   60 months   \n",
       "2  1077175  1313524.0     2400.0       2400.0           2400.0   36 months   \n",
       "3  1076863  1277178.0    10000.0      10000.0          10000.0   36 months   \n",
       "4  1075358  1311748.0     3000.0       3000.0           3000.0   60 months   \n",
       "\n",
       "  int_rate  installment grade sub_grade            ...              \\\n",
       "0   10.65%       162.87     B        B2            ...               \n",
       "1   15.27%        59.83     C        C4            ...               \n",
       "2   15.96%        84.33     C        C5            ...               \n",
       "3   13.49%       339.31     C        C1            ...               \n",
       "4   12.69%        67.79     B        B5            ...               \n",
       "\n",
       "  num_tl_90g_dpd_24m num_tl_op_past_12m pct_tl_nvr_dlq  percent_bc_gt_75  \\\n",
       "0                NaN                NaN            NaN               NaN   \n",
       "1                NaN                NaN            NaN               NaN   \n",
       "2                NaN                NaN            NaN               NaN   \n",
       "3                NaN                NaN            NaN               NaN   \n",
       "4                NaN                NaN            NaN               NaN   \n",
       "\n",
       "  pub_rec_bankruptcies tax_liens tot_hi_cred_lim total_bal_ex_mort  \\\n",
       "0                  0.0       0.0             NaN               NaN   \n",
       "1                  0.0       0.0             NaN               NaN   \n",
       "2                  0.0       0.0             NaN               NaN   \n",
       "3                  0.0       0.0             NaN               NaN   \n",
       "4                  0.0       0.0             NaN               NaN   \n",
       "\n",
       "  total_bc_limit total_il_high_credit_limit  \n",
       "0            NaN                        NaN  \n",
       "1            NaN                        NaN  \n",
       "2            NaN                        NaN  \n",
       "3            NaN                        NaN  \n",
       "4            NaN                        NaN  \n",
       "\n",
       "[5 rows x 111 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's start getting rid of useless columns, by useless columns I mean columns that:\n",
    "\n",
    "* leak information from the future (after the loan has already been funded)\n",
    "* don't affect a borrower's ability to pay back a loan (e.g. a randomly generated ID value by Lending Club)\n",
    "* formatted poorly and need to be cleaned up\n",
    "* require more data or a lot of processing to turn into a useful feature\n",
    "* contain redundant information"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "removing the desc column (which contains a long text explanation for each loan) and removing the url column (which contains a link to each loan on Lending Club which can only be accessed with an investor account)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "loan_df = loan_df.drop(['desc', 'url'],axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "removing duplicates if any"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(42538, 109)\n",
      "Removing duplicates if any ...\n",
      "(42538, 109)\n"
     ]
    }
   ],
   "source": [
    "print(loan_df.shape)\n",
    "print(\"Removing duplicates if any ...\")\n",
    "loan_df.drop_duplicates()\n",
    "print(loan_df.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Dropping columns that are reduntant or leak data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "loan_df.drop(['id','member_id','funded_amnt','funded_amnt_inv','grade','sub_grade',\n",
    "              'issue_d','zip_code','out_prncp','out_prncp_inv','total_pymnt','total_pymnt_inv',\n",
    "              'total_rec_prncp','total_rec_int','total_rec_late_fee','recoveries','collection_recovery_fee',\n",
    "              'last_pymnt_d','last_pymnt_amnt'], axis=1,inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(42538, 90)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dropping columns with more than 90% missing values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['acc_open_past_24mths',\n",
       " 'all_util',\n",
       " 'annual_inc_joint',\n",
       " 'avg_cur_bal',\n",
       " 'bc_open_to_buy',\n",
       " 'bc_util',\n",
       " 'dti_joint',\n",
       " 'il_util',\n",
       " 'inq_fi',\n",
       " 'inq_last_12m',\n",
       " 'max_bal_bc',\n",
       " 'mo_sin_old_il_acct',\n",
       " 'mo_sin_old_rev_tl_op',\n",
       " 'mo_sin_rcnt_rev_tl_op',\n",
       " 'mo_sin_rcnt_tl',\n",
       " 'mort_acc',\n",
       " 'mths_since_last_major_derog',\n",
       " 'mths_since_last_record',\n",
       " 'mths_since_rcnt_il',\n",
       " 'mths_since_recent_bc',\n",
       " 'mths_since_recent_bc_dlq',\n",
       " 'mths_since_recent_inq',\n",
       " 'mths_since_recent_revol_delinq',\n",
       " 'next_pymnt_d',\n",
       " 'num_accts_ever_120_pd',\n",
       " 'num_actv_bc_tl',\n",
       " 'num_actv_rev_tl',\n",
       " 'num_bc_sats',\n",
       " 'num_bc_tl',\n",
       " 'num_il_tl',\n",
       " 'num_op_rev_tl',\n",
       " 'num_rev_accts',\n",
       " 'num_rev_tl_bal_gt_0',\n",
       " 'num_sats',\n",
       " 'num_tl_120dpd_2m',\n",
       " 'num_tl_30dpd',\n",
       " 'num_tl_90g_dpd_24m',\n",
       " 'num_tl_op_past_12m',\n",
       " 'open_acc_6m',\n",
       " 'open_il_12m',\n",
       " 'open_il_24m',\n",
       " 'open_il_6m',\n",
       " 'open_rv_12m',\n",
       " 'open_rv_24m',\n",
       " 'pct_tl_nvr_dlq',\n",
       " 'percent_bc_gt_75',\n",
       " 'tot_coll_amt',\n",
       " 'tot_cur_bal',\n",
       " 'tot_hi_cred_lim',\n",
       " 'total_bal_ex_mort',\n",
       " 'total_bal_il',\n",
       " 'total_bc_limit',\n",
       " 'total_cu_tl',\n",
       " 'total_il_high_credit_limit',\n",
       " 'total_rev_hi_lim',\n",
       " 'verification_status_joint']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check for columns with 90% of missing values\n",
    "percent = 0.9\n",
    "over_ninety_percernt_ma = loan_df.isnull().sum()[loan_df.isnull().sum() /  loan_df.shape[0] > percent].index.tolist()\n",
    "loan_df.drop(over_ninety_percernt_ma, axis=1,inplace=True)\n",
    "over_ninety_percernt_ma.sort()\n",
    "over_ninety_percernt_ma"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(42538, 34)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['loan_amnt', 'term', 'int_rate', 'installment', 'emp_title',\n",
       "       'emp_length', 'home_ownership', 'annual_inc', 'verification_status',\n",
       "       'loan_status', 'pymnt_plan', 'purpose', 'title', 'addr_state', 'dti',\n",
       "       'delinq_2yrs', 'earliest_cr_line', 'inq_last_6mths',\n",
       "       'mths_since_last_delinq', 'open_acc', 'pub_rec', 'revol_bal',\n",
       "       'revol_util', 'total_acc', 'initial_list_status', 'last_credit_pull_d',\n",
       "       'collections_12_mths_ex_med', 'policy_code', 'application_type',\n",
       "       'acc_now_delinq', 'chargeoff_within_12_mths', 'delinq_amnt',\n",
       "       'pub_rec_bankruptcies', 'tax_liens'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_df.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's focus on the target column.\n",
    "\n",
    "Since we're interesting in being able to predict which of these 2 values a loan will fall under, we can treat the problem as a binary classification one. Let's remove all the loans that don't contain either Fully Paid and Charged Off as the loan's status and then transform the Fully Paid values to 1 for the positive case and the Charged Off values to 0 for the negative case"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Fully Paid                                             33586\n",
       "Charged Off                                             5653\n",
       "Does not meet the credit policy. Status:Fully Paid      1988\n",
       "Does not meet the credit policy. Status:Charged Off      761\n",
       "Current                                                  513\n",
       "In Grace Period                                           16\n",
       "Late (31-120 days)                                        12\n",
       "Late (16-30 days)                                          5\n",
       "Default                                                    1\n",
       "Name: loan_status, dtype: int64"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#target coluumn\n",
    "loan_df.loan_status.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    33586\n",
       "0     5653\n",
       "Name: loan_status, dtype: int64"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# keep \"Full Paid\" and \"Charged Off\"\n",
    "loan_df = loan_df[loan_df[\"loan_status\"].isin([\"Fully Paid\",\"Charged Off\"])]\n",
    "mapping_dict = {\n",
    "    \"loan_status\": {\n",
    "        \"Fully Paid\": 1,\n",
    "        \"Charged Off\": 0,\n",
    "    }\n",
    "}\n",
    "loan_df = loan_df.replace(mapping_dict)\n",
    "loan_df.loan_status.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "let's also get rid of columns that only has 1 unique value (that won't give our model any info)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['initial_list_status', 'collections_12_mths_ex_med', 'policy_code', 'application_type', 'acc_now_delinq', 'chargeoff_within_12_mths', 'delinq_amnt', 'tax_liens']\n"
     ]
    }
   ],
   "source": [
    "# let's check for columns with just one value\n",
    "cols_to_drop = []\n",
    "for col in loan_df.columns:\n",
    "    if len(loan_df[col].dropna().unique()) == 1:\n",
    "           cols_to_drop.append(col)\n",
    "\n",
    "print(cols_to_drop)\n",
    "loan_df.drop(cols_to_drop, axis=1, inplace=True)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "n    39238\n",
      "y        1\n",
      "Name: pymnt_plan, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# this one has only 1 value for 1 ....\n",
    "print(loan_df.pymnt_plan.value_counts())\n",
    "loan_df.drop('pymnt_plan', axis=1, inplace=True) # get rid of it"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we are going to work on the features that still have missing values, we can:\n",
    "\n",
    "* remove rows\n",
    "* set new value (other, min, max, mean, etc ..)\n",
    "* regress new value based on other columns\n",
    "* ...\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "loan_amnt                     0\n",
       "term                          0\n",
       "int_rate                      0\n",
       "installment                   0\n",
       "emp_title                  2426\n",
       "emp_length                    0\n",
       "home_ownership                0\n",
       "annual_inc                    0\n",
       "verification_status           0\n",
       "loan_status                   0\n",
       "purpose                       0\n",
       "title                        10\n",
       "addr_state                    0\n",
       "dti                           0\n",
       "delinq_2yrs                   0\n",
       "earliest_cr_line              0\n",
       "inq_last_6mths                0\n",
       "mths_since_last_delinq    25352\n",
       "open_acc                      0\n",
       "pub_rec                       0\n",
       "revol_bal                     0\n",
       "revol_util                   50\n",
       "total_acc                     0\n",
       "last_credit_pull_d            2\n",
       "pub_rec_bankruptcies        697\n",
       "dtype: int64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After checking the variables let's manage missing values, we'll:\n",
    "\n",
    "* remove column emp_title. There are too many missing values and it relates with 'title'\n",
    "* Fill missing values with the median to 'mths_since_last_delinq'\n",
    "* Remove rows with missing values on pandas data frame"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "loan_amnt                 0\n",
       "term                      0\n",
       "int_rate                  0\n",
       "installment               0\n",
       "emp_length                0\n",
       "home_ownership            0\n",
       "annual_inc                0\n",
       "verification_status       0\n",
       "loan_status               0\n",
       "purpose                   0\n",
       "title                     0\n",
       "addr_state                0\n",
       "dti                       0\n",
       "delinq_2yrs               0\n",
       "earliest_cr_line          0\n",
       "inq_last_6mths            0\n",
       "mths_since_last_delinq    0\n",
       "open_acc                  0\n",
       "pub_rec                   0\n",
       "revol_bal                 0\n",
       "revol_util                0\n",
       "total_acc                 0\n",
       "last_credit_pull_d        0\n",
       "pub_rec_bankruptcies      0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fix missing values in columns\n",
    "loan_df.drop('emp_title', axis=1, inplace=True)\n",
    "#loan_df.mths_since_last_delinq = loan_df.mths_since_last_delinq.fillna(loan_df.mths_since_last_delinq.median())\n",
    "#loan_df.pub_rec_bankruptcies = loan_df.pub_rec_bankruptcies.fillna(0)\n",
    "# remove rows with missing values\n",
    "loan_df.dropna(axis=0, inplace=True)\n",
    "# check missing values again (should be none)\n",
    "loan_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's focus now on the text columns. We need to convert them to numerical "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "float64    12\n",
      "object     11\n",
      "int64       1\n",
      "dtype: int64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "term                         36 months\n",
       "int_rate                        13.49%\n",
       "emp_length                   10+ years\n",
       "home_ownership                    RENT\n",
       "verification_status    Source Verified\n",
       "purpose                          other\n",
       "title                         personel\n",
       "addr_state                          CA\n",
       "earliest_cr_line              Feb-1996\n",
       "revol_util                         21%\n",
       "last_credit_pull_d            Apr-2016\n",
       "Name: 3, dtype: object"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(loan_df.dtypes.value_counts())\n",
    "# show an example of object columns\n",
    "loan_df.select_dtypes(include=['object']).iloc[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's first convert revol_util and int_rate to numerical values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "loan_df.int_rate = loan_df.int_rate.map(lambda x: float(x[:-1]))\n",
    "loan_df.revol_util = loan_df.revol_util.map(lambda x: float(x[:-1]))\n",
    "loan_df.term = loan_df.term.map(lambda x: int(x.split()[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       " 10    3162\n",
       " 2     1402\n",
       " 3     1359\n",
       " 0     1337\n",
       " 4     1169\n",
       " 5     1080\n",
       " 1     1021\n",
       " 6      744\n",
       " 7      630\n",
       " 8      528\n",
       " 9      407\n",
       "-1      340\n",
       "Name: emp_length, dtype: int64"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# term\n",
    "loan_df.emp_length.value_counts()\n",
    "mapping_dict = {'emp_length':{\n",
    "        '10+ years':'10',\n",
    "        '< 1 year':'0',\n",
    "        'n/a':'-1'\n",
    "    }\n",
    "}\n",
    "loan_df = loan_df.replace(mapping_dict)\n",
    "loan_df.emp_length = loan_df.emp_length.map(lambda x: int(x.split()[0]))\n",
    "loan_df.emp_length.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To work with the column title we are going to:\n",
    "\n",
    "* Set to lowercase\n",
    "* Tokenize\n",
    "* Remove stopwords \n",
    "* Snowball stemmer\n",
    "* tf-idf\n",
    "* SVD (n=5)\n",
    "* Add the 5 resulting columns to our data frame"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'leaf'"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import nltk\n",
    "lemma = nltk.wordnet.WordNetLemmatizer()\n",
    "lemma.lemmatize('leaves')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "tokenizer = RegexpTokenizer(r'\\w+')\n",
    "title_tokens = loan_df['title'].map(lambda x: tokenizer.tokenize(x.lower()))\n",
    "title_tokens_filter = title_tokens.map(lambda x: [word for word in x if word not in stopwords.words('english')])\n",
    "title_tokens_filter = title_tokens_filter.map(lambda x: ['empty'] if x == [] else x)\n",
    "#sbst = SnowballStemmer(language='english')\n",
    "title_tokens_filter_stem = title_tokens_filter.map(lambda x: lemma.lemmatize((' ').join(x)))\n",
    "\n",
    "#tfidf\n",
    "tfidf = TfidfVectorizer(ngram_range=(1, 3), stop_words='english')\n",
    "tfidf_values = tfidf.fit_transform(title_tokens_filter_stem)\n",
    "tsvd = TruncatedSVD(n_components=15, random_state = 1)\n",
    "tfidf_values_truncatedSVD = tsvd.fit_transform(tfidf_values)\n",
    "\n",
    "#drop original title column\n",
    "loan_df.drop('title',1, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "for earliest_cr_line, last_credit_pull_d we want to get month and year in numbers, I create float that will tell me year and month (year + month/13)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# convert to date\n",
    "earliest_cr_line = pd.to_datetime(loan_df.earliest_cr_line)\n",
    "last_credit_pull_d = pd.to_datetime(loan_df.last_credit_pull_d)\n",
    "# extract month and year\n",
    "earliest_cr_line_year = earliest_cr_line.map(lambda x: x.year)\n",
    "earliest_cr_line_month = earliest_cr_line.map(lambda x: x.month)\n",
    "last_credit_pull_d_year = last_credit_pull_d.map(lambda x: x.year)\n",
    "last_credit_pull_d_month = last_credit_pull_d.map(lambda x: x.month)\n",
    "# create float with year + month/13\n",
    "loan_df.earliest_cr_line = earliest_cr_line_year + earliest_cr_line_month/13\n",
    "loan_df.last_credit_pull_d = last_credit_pull_d_year + last_credit_pull_d_month/13\n",
    "loan_df['earliest_cr_line_month'] = earliest_cr_line_month\n",
    "loan_df['last_credit_pull_d_month'] = last_credit_pull_d_month"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "purpose, verification_status, home_ownership, and addr_state as dummies because they don't have many values (maybe we could skip state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "loan_df = pd.concat([loan_df, pd.get_dummies(loan_df.purpose)], axis=1)\n",
    "loan_df = pd.concat([loan_df, pd.get_dummies(loan_df.verification_status)], axis=1)\n",
    "loan_df = pd.concat([loan_df, pd.get_dummies(loan_df.home_ownership)], axis=1)\n",
    "loan_df = pd.concat([loan_df, pd.get_dummies(loan_df.addr_state)], axis=1)\n",
    "loan_df = pd.concat([loan_df, pd.get_dummies(loan_df.earliest_cr_line_month)], axis=1)\n",
    "loan_df = pd.concat([loan_df, pd.get_dummies(loan_df.last_credit_pull_d_month)], axis=1)\n",
    "loan_df = pd.concat([loan_df.reset_index(drop=True), pd.DataFrame(tfidf_values_truncatedSVD)], axis=1)\n",
    "\n",
    "loan_df.drop(['purpose','verification_status','home_ownership','addr_state',\n",
    "              'mths_since_last_delinq','last_credit_pull_d_month','earliest_cr_line_month'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's save the processed data frame to apply our model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# save processed data\n",
    "loan_df.to_csv(\".\\\\data_processed\\loans.csv\",index=False,encoding='utf8')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Now we have data ready to apply some machine learning models. Or objective is to achive TPR > 50% adn FPR < 7%.\n",
    "We'll try logistic regression and random forest. On both models we'll use the param weights to account for class imbalace."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## MODELS\n",
    "\n",
    "Remember that our objective is to get TPR > 70% and FPR < 7% for both LR and RF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#penalty = {0: 4,1: 1}\n",
    "\n",
    "lr = LogisticRegression(class_weight='balanced', random_state = 3, penalty='l2', C=1)\n",
    "rf = RandomForestClassifier(n_estimators=400, min_samples_leaf=400, class_weight='balanced',random_state=3)\n",
    "selection_param = 0.0003"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## FEATURE SELECTION (RF) / OVERSAMPLING"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(0.41310000000000002, 'last_credit_pull_d'),\n",
       " (0.32529999999999998, '9.1'),\n",
       " (0.037499999999999999, 'int_rate'),\n",
       " (0.035700000000000003, 'term'),\n",
       " (0.0161, '7.1'),\n",
       " (0.015900000000000001, 'annual_inc'),\n",
       " (0.015800000000000002, '3.1'),\n",
       " (0.014500000000000001, '4.1'),\n",
       " (0.013599999999999999, '8.1'),\n",
       " (0.010500000000000001, '6.1'),\n",
       " (0.0091000000000000004, '2.1'),\n",
       " (0.0088000000000000005, '5.1'),\n",
       " (0.0074999999999999997, '12.1'),\n",
       " (0.0071999999999999998, '11.1'),\n",
       " (0.0053, '1.1'),\n",
       " (0.0044000000000000003, '10.1'),\n",
       " (0.0038, 'revol_util'),\n",
       " (0.0038, '7.2'),\n",
       " (0.0035000000000000001, 'loan_amnt'),\n",
       " (0.0028999999999999998, 'earliest_cr_line'),\n",
       " (0.0028, '9.2'),\n",
       " (0.0027000000000000001, '3.2'),\n",
       " (0.0025999999999999999, 'revol_bal'),\n",
       " (0.0025000000000000001, '1.2'),\n",
       " (0.0023999999999999998, 'installment'),\n",
       " (0.0023, 'inq_last_6mths'),\n",
       " (0.0023, 'total_acc'),\n",
       " (0.002, 'dti'),\n",
       " (0.0018, 'open_acc'),\n",
       " (0.0016999999999999999, '8.2'),\n",
       " (0.0016000000000000001, 'small_business'),\n",
       " (0.0016000000000000001, '5.2'),\n",
       " (0.0015, 'emp_length'),\n",
       " (0.0015, '6.2'),\n",
       " (0.0015, '11.2'),\n",
       " (0.0012999999999999999, '0'),\n",
       " (0.0012999999999999999, '10.2'),\n",
       " (0.0012999999999999999, '12.2'),\n",
       " (0.0011999999999999999, '2.2'),\n",
       " (0.0011999999999999999, '4.2'),\n",
       " (0.0011999999999999999, '13'),\n",
       " (0.0011000000000000001, 'Verified'),\n",
       " (0.00089999999999999998, '14'),\n",
       " (0.00080000000000000004, 'pub_rec'),\n",
       " (0.00069999999999999999, 'Not Verified'),\n",
       " (0.00069999999999999999, 'MORTGAGE'),\n",
       " (0.00059999999999999995, 'credit_card'),\n",
       " (0.00040000000000000002, 'RENT')]"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_df = pd.read_csv(\".\\\\data_processed\\loans.csv\")\n",
    "#loan_df_0 = loan_df[loan_df[\"loan_status\"]==0]\n",
    "#loan_df = pd.concat([loan_df,loan_df_0],ignore_index=True)\n",
    "#loan_df = pd.concat([loan_df,loan_df_0],ignore_index=True)\n",
    "#loan_df = pd.concat([loan_df,loan_df_0],ignore_index=True)\n",
    "#loan_df = pd.concat([loan_df,loan_df_0],ignore_index=True)\n",
    "#loan_df = pd.concat([loan_df,loan_df_0],ignore_index=True)\n",
    "#loan_df = pd.concat([loan_df,loan_df_0],ignore_index=True)\n",
    "\n",
    "features = loan_df.drop(\"loan_status\", 1)\n",
    "target = loan_df.loan_status\n",
    "\n",
    "rfsel = RandomForestClassifier(n_estimators=500, min_samples_leaf=200, class_weight='balanced',random_state=2)\n",
    "rfsel.fit(features, target)\n",
    "\n",
    "features_lb = sorted(zip(map(lambda x: round(x, 4), rfsel.feature_importances_), features.columns), reverse=True, key=lambda pair: pair[0])\n",
    "columns = [features_lb[i][1] for i in range(0,10)] \n",
    "\n",
    "\n",
    "features = features[[i[1] for i in features_lb if i[0]>selection_param]]\n",
    "sd = StandardScaler()\n",
    "features = sd.fit_transform(features)\n",
    "\n",
    "[i for i in features_lb if i[0]>selection_param]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LOGISTIC REGRESSION -  TRAIN-TEST "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TPR:  0.688032272523532\n",
      "FPR:  0.07901234567901234\n",
      "CONFUSION MATRIX:\n",
      "TN:  373    FP:  32\n",
      "FN:  696    TP:  1535\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nxA6h3YtuAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb951390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(features, target, stratify=target,test_size=0.20)\n",
    "\n",
    "lr.fit(X_train, y_train)\n",
    "\n",
    "predictions = lr.predict(X_test)\n",
    "predictions = pd.Series(predictions)\n",
    "\n",
    "tn = len(predictions[(predictions == 0) & (y_test.values==0)])  \n",
    "tp = len(predictions[(predictions == 1) & (y_test.values==1)]) \n",
    "fn = len(predictions[(predictions == 0) & (y_test.values==1)])  \n",
    "fp = len(predictions[(predictions == 1) & (y_test.values==0)]) \n",
    "fpr = fp / (fp + tn)\n",
    "tpr = tp / (tp + fn)\n",
    "\n",
    "#print(lr.coef_)\n",
    "print('TPR: ',tpr)\n",
    "print('FPR: ',fpr)\n",
    "fprpoints, tprpoints, thresholds = roc_curve(y_test, predictions)\n",
    "print('CONFUSION MATRIX:')\n",
    "cm = confusion_matrix(y_test,predictions)\n",
    "print('TN: ',cm[0,0], '   FP: ',cm[0,1])\n",
    "print('FN: ',cm[1,0], '   TP: ',cm[1,1])\n",
    "plt.figure()\n",
    "lw = 2\n",
    "plt.plot(fprpoints, tprpoints, color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % auc(fprpoints,tprpoints))\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver operating characteristic example')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LOGISTIC REGRESSION - CV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "TPR:  0.7022859704168535\n",
      "FPR:  0.07905138339920949\n",
      "CONFUSION MATRIX:\n",
      "TN:  1864    FP:  160\n",
      "FN:  3321    TP:  7834\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAgFAJRAAVAIBQCUQAFQCAUD1ANTH3sDmSWfxAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb951cf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "kf = KFold(features.shape[0], random_state=1)\n",
    "\n",
    "predictions = cross_val_predict(lr, features, target,cv=kf)\n",
    "predictions = pd.Series(predictions)\n",
    "\n",
    "tn = len(predictions[(predictions == 0) & (target==0)])  \n",
    "tp = len(predictions[(predictions == 1) & (target==1)]) \n",
    "fn = len(predictions[(predictions == 0) & (target==1)])  \n",
    "fp = len(predictions[(predictions == 1) & (target==0)]) \n",
    "fpr = fp / (fp + tn)\n",
    "tpr = tp / (tp + fn)\n",
    "\n",
    "print()\n",
    "print('TPR: ',tpr)\n",
    "print('FPR: ',fpr)\n",
    "fprpoints, tprpoints, thresholds = roc_curve(target, predictions)\n",
    "print('CONFUSION MATRIX:')\n",
    "cm = confusion_matrix(target,predictions)\n",
    "print('TN: ',cm[0,0], '   FP: ',cm[0,1])\n",
    "print('FN: ',cm[1,0], '   TP: ',cm[1,1])\n",
    "plt.figure()\n",
    "lw = 2\n",
    "plt.plot(fprpoints, tprpoints, color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % auc(fprpoints,tprpoints))\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver operating characteristic example')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## RANDOM FOREST -  TRAIN-TEST "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TPR:  0.7140295831465711\n",
      "FPR:  0.06172839506172839\n",
      "CONFUSION MATRIX:\n",
      "TN:  380    FP:  25\n",
      "FN:  638    TP:  1593\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xd4VGX2wPHvSSDUhN4hgKAiXUTEVTSKKCj87Cuirqgr\nig3RXbuCrgVWWSwoArLusvaODVHRqIAiCgJSRIrUUEISElpIOb8/7k2YJJNkUmbuTHI+zzNPZu7c\ncuZmZs689z33vaKqGGOMMYVFeR2AMcaY8GQJwhhjjF+WIIwxxvhlCcIYY4xfliCMMcb4ZQnCGGOM\nX5YgIpiIjBCRz7yOw2si0k5E0kVEQrjN9iKSKyJV4jMkIr+KyGnlWK7KvgdF5HQR2eJ1HF4SOw+i\ncojIH0BzIBvYB8wFblbVA17GVRWJyEbgOlX9ysMY2gMbgJqqmutVHG4suUBnVd0Q5O20BzYCNbx+\nzaEgIqcD/1PVeK9j8UqV+PUTJhQ4T1XjgN7A8cC93oZUPiISXR237ZVKeM3l/pVXxm2Lu62QtdSM\ntyxBVC4BUNVdOC2I3vlPiMSIyFMisklEkkTkBRGp5fP8+SKyVET2isjvInK2Oz1ORF4Ske0iskVE\n/pF3KEVErhaR79z7L4jIkwWCEflARG5377cSkXdEZJeIrBeRW33mGycib4vI/0QkDbi6yAtz4pjl\nLr9RRO73ee5qEZkvIs+JSJqIrBKRMwstW9JrmC8i/xKRZGCciBwlIvNEJNnd3isiEufOPwuIBz5y\nDyv9rfDhHhH5WkQecdebLiKfiUhjn3j+IiJ/iMhuEXnAfT358RZ63bVFZJI7f6qIfOvzfxPgSvd/\nuktE7vNZ7kQRWegus83dNzV8ns8VkZtEZC2w1p32tIhsdt8Di0XkVJ/5o0TkPhFZ576mxSLSVkS+\nceNY7k6/1J1/qPt+SnX3Qw+fdW0UkbtEZBmwT0SiffeBG/tiN44kEXnKXfQb92+au62TfN+D7rLd\nRORzEdnjLntPMfu18Odhat5+dWP7wef/OVpEVohIjPv4LXeZVBFJFJGuPut9WUSeF5FPRSRDRL4T\nkRYiMllEUtz3Zq9C++IeEVnpxjwzbzt+Yi72M1RlqardKuGG0/Q+073fFlgO/Mvn+cnAB0ADoB4w\nG3jMfa4fkOazfCvgGPf++8ALQG2gKfADcL373NXAt+79AcAmn+01BA4ALXC+QH4C7geigQ7AOmCQ\nO+84IBMY5j6u5ef1zXJjqQu0B34DrvGJIwu4zV3/n93X0zDA15AF3ITzg6UW0AkYCNQAmgCJhfbl\nRuAMn8ftgRwgyn38NfC7u55a7uPH3ee6AhnAye76n3Rf+5nF/F+fB74CWrr7sT9Q091mLjANiAF6\nAoeAY93l+rj/V8FJaCuB23zWm4vzI6JB3v4GRrj/tyhgLJAExLjP/R1YhnMoCaAH0MhnXR191n08\nsBPo627/Knef1fTZf0uA1j7b9n3/LgSucO/XBfoV2s/isy3f92B9YDtwu7tP6gEnFrNfS/o8iPs/\nfwjoDKQAPX2WHenGVRP4F7DU57mXgV04P85igHk4hwKvcNf7D+CrQu+l5e6+aAjMBx5xnzsd2OwT\nU7Gfoap68zyAqnJz32jp7i0X+AKI83l+X6EP8cnABvf+i8AkP+tsjvOlU8tn2vC8N7jvh9N9/Adw\nqnv/r8CX7v2TgD8KrfseYKZ7fxyQWMJri8L5Ej3WZ9qoQnFsLbTMIvdDGchr+KO4bbvznA/8XGhf\nn+nz2F+CuM/n+dHAp+79B4FXfZ6rQzEJwv1SOAB09/Nc3jZbFXrNfy7mNYwB3vV5nAucXsrrTgF6\nuPfXAEOLmS8XOMrn8QvAw4XmWQMM8Nl/V/t5/+YliET3PdGkmNcc5TPNN0EM9/0/lfLaiv08+Gxr\nD7AKuKuE9TR0X3+s+/hlYJrP87cAK30edwdSCr3u630eDwF+d+/7JogSP0NV9Zbf5DWV4nxV/VpE\nBgCv4fxaTheRZji/eH6WI4U2URw5ltsO+MTP+trj/EpKcpcT97a5mO2/CVyO8ytoBPA/d3o80EZE\nUtzH4m7/W59lS6rWaIrza9t3u5uANj6PtxVaZhPOr7JAXkOBbYtIc+AZnFZRfZxfbCmUzQ6f+wfc\n9eDGlL89VT0oInuKWUdTnBZISZ2/O/1tR0SOxvl12xcnCdUAfi607FbfByLyN+BanBYkQKwbAzjv\nkUA7odsDf/E5BCI4/4PWxW27kOtwfmmvEZENOL+o/b0/C2sHrC9tpgA+D6jqJhH5GucL+wWfZaOA\nx4FLcPaNuremOC1DKPg/OejncX0K8t0Xee/bwgL5DFU51gdRufL6IL4D/gtMcqcn43x5dFPVxu6t\noao2cJ/fgnM4pLAtOL++m7jLNHKX61nM9l8HLhGReJxfPO/6rGeDz7YbqWoDVR3ms6yW8LqScQ4D\ntfeZ1p6CSaENBcXjHG4I5DUU3vbjOL8Ku6lqQ+BKCnaMlhRraZJwDgECICJ1cA5j+ZPsxu7vf1Oa\nqcBqoJP7Gu6naOdu/utw+xv+Dlzi7qNGOK3RvGWKe4/4swXncI3v/7u+qr7pb9uFqep6VR2hqs2A\nfwLvuPuptP0eaIylfR4QkfNwWhXzgKd8lh0BDMNp7TTEOdST96OjvNr53G+P874tLJDPUJVjCSJ4\nngYGiUgPddqjM4Cn3V9PiEgbcTuigZnANSJyhjhai8ixqroD+ByYLCKx7nNHSTH16qr6C06z/CXg\nM1VNd5/6EchwO/9qu52S3USkbyAvRJ2SxreAx0SkvjjljmM50kIBaC4it4pIDbejtAvOYZ0yvQZX\nLM4hiAwRaYPzxelrB3BUoWmBfkG8AwwTkf4iUhMYX9yM7v/t38C/3A7KKJ/lSttmLJCuqgdEpAvO\nYa6SxOIk4T1uB+5D7rQ8LwH/EJHOACLSQ0Qauc8V3h8zgBtFpJ87bz0ROVdE6pUSA+78V4hIXstl\nL05iyAV2u3+LSwIfAy1F5Db3NdTPi8FXaZ8Hd9szcFpTI4GhIjLEXTwW55Bgqvt6nqDsPxgK/99u\ndrffGLgPeMPPMhX6DEUqSxCVp8CbVFWTcVoRD7mT7sHp1PpBnEqhz4Fj3HkXA9fgJJW9OMeA82qv\n/4LT2bYK5zDL2zgdpsV5DaeD91WfWHKBoTgddxtxOvFmAHFleH234fzq24DTrH5FVV/2eX4RcDTO\nr8N/ABeramo5X8PDwAk4Hd0fcaQllGcC8KBblXJH3sv0eb6kX8ergFtxDsdtx/mVvgvnS8efvwEr\ngMU4yXcCRz43hbejhZa7QkTScTqyC3/pFF52rntbi/M/OkDBQ2//wknSn4vIXpyEUcd97mFglrs/\nLlHVn4HrgSnuIZG1FKxM87d/fKcNBla6sU8GLlPVTFU9CDwGLHC3VeDLX1X3AYOA/8NJWmuBBD/b\nAribYj4POPvrfVWdq6opOP1pM9yEOAvn8OQ24FecDvWyKvz6X3O3vw6nuOGxIgtUzmco4tiJcqbC\nRORqnBPXynwmrtfcX6FpONVBm7yOx4SWhMFJl+HMWhCm2hHnHIE6bnKYBCy35GBMUZYgTHV0Ps7h\npa04x9OHexuO8ZAdQimBHWIyxhjjl7UgjDHG+BUxJ8qJiDV1jDGmHFS1XOeJRFQLwuvTzsPlNm7c\nOM9jCJeb7QvbF7YvSr5VREQlCGOMMaFjCcIYY4xfliAiUEJCgtchhA3bF0fYvjjC9kXliJgyVxHR\nSInVGGPChYig4dhJ7V6daaeILC9hnmfFuYLaLyLSu7j5jDHGhFawDzG9DJxT3JPuCI2dVPVo4Aac\nC+cYY4wJA0FNEKo6H0gtYZbzcUZnRFUXAQ1EpEUwYzLGGBMYr0+Ua0PBIY23udN2+p/dGGNMsVQh\nYyvsWoLuXMr7HwZ6EUL/vE4QZTJ+/Pj8+wkJCVapYIypvjQXUtfBrqWwa4nzd+cSEn/dQ6J74dev\n17cveR2lCHoVk3v1sY/Uz2UyReRF4Gt1L4UoImtwLuRepAVhVUzGmGorJwtSVsPOJT4J4RfI2ld0\n3tpNoPnx0KIPB+N6Uff4K8pdxRSKFkRJ14v9ELgZeFNE+gNp/pKDMcZUG1kHIXlFgVYBySsgx89F\nD+u3geZ9yGlyPNGt+ziJIbYdiPOV61xy8IpyhxLUBCEir+FccrCJiGwGxuFcelJVdbqqfupeK3cd\nsB/nspvGGFM9ZO51WgK+h4n2rAbNKTpvw85OAmjeB1oc79yv25wff9zGqFEfMWFCNwZ3ji+6XAXY\niXLGGBMKB3YdaRHkJYS09UXnkyho0tUnGfSBZr2gVoMCs2VkZPLgg1/z7LOLUIUzzujAV19dXXR1\nFThRLqI6qY0xJuypQsaWoslg37ai80bXgqY9nCSQlxCa9oCadUrcxCefrGX06E/YsiWd6GjhjjtO\nZty40yv9pVgLwhhjyiu/kmhJwYRwaE/ReWvWh+a9nSTgdiLT+DiIrlmmTR46lM2xx05h8+a9nHBC\nK2bMGMbxx7cqdv6KtCAsQRhjTCDKWknk2ypofjw06uwcPqoEc+b8zurVydx220nUqFHyOi1BGGNM\nZco6CMnLC7YKiq0kanukRZCXEGLb5lcSec36IIwxprzyK4l8DhOlrClTJVEwHD6cw4sv/sSoUSdQ\nu7Y3X9WWIIwx1UfAlUTR0LR7wf4CP5VEwfL991u4/vqPWLlyN8nJB3jkkTNCst3CLEEYY6qevEqi\nAv0FSwOoJHITQgCVRMGQnp7JfffN44UXFqMKnTs35owzOoQ8jjyWIIwxkc23ksg3IRxKKTpvzfru\nISKfPoNyVBIFw44d++jbdzrbtmVQo0YUd931Jx544DTq1PEuNksQxpjIkZMFe1YVbBUEWknUog80\n7FRplUSVrUWLevTp04o2beKYMWMYPXt6f+UDq2IyxoSnvEoi31ZB8grIOVx03jCvJApUWtohYmNj\niI6uvCRmZa7GmMjmt5JotXP4qLCGnQt2Hjc/Huo2C33MFbB37yEaNKgdkm1ZmasxJnIc2FW08zgM\nK4mCITMzmyeemM/TT//ATz+NonPnxl6HVCJLEMaY4ChrJVGzngXPPPaokihYvvtuE6NGfcyaNckA\nfPrp79x220keR1UySxDGmIrTXEj9veg5BiVVEvn2FzTuEhaVRMGQlnaIu+/+gunTlwBw7LFNmD59\nGKedVrGrvYWCJQhjTNnkVxL59BfsXua/kqhO04KtgjCvJAqGPXsOMGvWcmrWjOLee0/l3nsHeHZm\ndFlZJ7UxpnhZB5zKoUAriQoPUBeBlUTB8PrrK+jZswXdugVnWI6SWBWTMabiDqXB7l8KHiaqwpVE\n1YVVMRljymb/TrdFsPTIGch7NxSdT6KdzmLfRNCsN9SKC33MYW7Fip289toKHn98IFJFWk2WIIyp\nylQhYzPs9D3zeAns21503mpQSRQMhw5l8+ij3zJx4gKys3Pp27c1F1/c1euwKoUlCGOqirxKosJl\npf4qiWJinZZANakkCpbExD8YNeojfv/d2cejR/flrLOO8jiqymMJwphI5LeS6BfI2l903jpNi/YX\nVLNKomD44IM1XHjhmwB07dqM6dOHcsop8R5HVbmsk9qYcJd1AHYvL9gqsEoizx06lE3//i9x0UXH\ncffdp1CrVnj+3rYqJmOqirxKIt/DRClr/FcSNToamvmecGaVRKGWnZ1b6jWhvWZVTMZEovxKIp/D\nRAFVEuWNSWSVRKGQk5PLH3+k0alT0XGTwj05VJQlCGOCLb+SqPCYRCVVEvn0GTTpbpVEHlm2bAfX\nX/8RO3bsY+XKm4iNreV1SCFlCcKYylSgkmjJkXMNrJIoohw8mMXDD3/DU08tJCdHads2jg0bUunV\nq6XXoYWUJQhjyivn8JGrm+W1DqySKOJ9++0mrr12NuvXpyICt97aj8ceO7PatR7AEoQxgcmvJPLp\nL9jzaymVRD4JoX4bqySKEAcPZrF+fSrduzdnxoxh9O/f1uuQPGNVTMYUZpVE1d4HH6zhvPOOpmbN\naK9DqTArczWmvPbvLNgq2LW0+EqiJl0L9hdYJZGJAFbmakxpyltJlJcQrJKoSsnOzuWZZ37gwIEs\nHnzwdK/DCVvWgjBVT27Okaub+Q5Qdyi16LwxsUcODeX1GVglUZW2ZEkSf/3rhyxduoMaNaJYt+5W\n2rdv6HVYQWMtCFN9Fakkyru6WSCVRH2g4VFWSVRN7N9/mHHjEpk8+Qdyc5X4+AZMnXpelU4OFRX0\nFoSIDAaeBqKAmao6sdDzccArQDwQDUxS1f/4WY+1IKq7slQSxbYreqlLqySq1kaP/pgXX/yZqChh\nzJiTeOSRM6hfP8brsIIubDupRSQKWAsMBLYDi4HhqrrGZ557gThVvVdEmgK/AS1UNbvQuixBVCeH\n0gpe0GbX0pIriZoXGqCubtPQx2zC2rZt6VxxxXs89dTZ9O3b2utwQiacDzH1A35X1U0AIvIGcD6w\nxmceBWLd+7HAnsLJwVRxRSqJlsDejUXnyxuTyCqJTDm0aRNHYuJIr8OIKMFOEG2ALT6Pt+IkDV9T\ngA9FZDtQH7gsyDEZr6hC+qaiA9TtTyo6b3Qt58vf98zjpj2gRu3Qx20iyrp1KYjgd3A9Uzbh0El9\nDrBUVc8UkU7AFyLSU1X3eR2YqYD8SqLCVzcLoJKohTsmUVQ4vD1NpMjKymHSpO95+OFv6NevDV9/\nfTVRUdbnVBHB/gRuw+l8ztPWnebrGuAJAFVdLyIbgS7AT4VXNn78+Pz7CQkJJCQkVG60pnzyKol8\nzzEorZLI9zCRVRKZClq8eBvXX/8Ry5btBKB9+wYcOpRN3brVr1w5MTGRxMTESllXsDupo3E6nQcC\nScCPwOWqutpnnueBXar6sIi0wEkMvVQ1pdC6rJM6HGTtP3J1s7yEEEglUV5CsEoiU8nuu28eEycu\nIDdX6dixIS++OJSzz+7kdVhhI2w7qVU1R0RuAT7nSJnrahG5wXlapwOPAv8RkeXuYncVTg7GI/mV\nRD79Bam/WSWRCStNmtRBBP7+9z8xfnxCtWw1BIudSW0c+3cUbBWUWEnUrWAisEoi46Hs7FxWr95N\njx4tvA4lLIXteRCVyRJEJcmvJCo0QJ2/SqIataFpT6skMmEh7/MvdoiyTML2EJPxWJkqieKgee+C\nZx5bJZEJE7/9lswNN3zMqFEnMGJED6/DqTbs019V5ByG5JWFBqj7BbIPFJ23TrOC4xE1P94qiUxY\nOnw4h3/+cwGPPvotmZk5JCcfYPjw7la+GiKWICJRkUqiJZD8K+RmFZ03tl3RS11aJZGJAN9/v4Xr\nr/+IlSt3A3DNNb156qmzLTmEkPVBhLtDqU5LIKBKomMKdh5bJZGJULm5Su/eL7JixS46d27MtGlD\nOfPMjl6HFZGsk7qqKFBJlHd1M6skMtXTwoVb+PjjtTz44GnUqWOlq+VlCSLSlKeSyPfM46bdrZLI\nGBMQq2IKZ7k5kLr2SCLYvdQqiYxx5eYq//3vL1x8cVfi4mp5HY4pxL55KlPhSqKd7phExVUStSh0\n5rFVEplqZPXq3Ywa9THz52/m55+TmDLlXK9DMoVYgiivvEoi3zOPS6sk8k0I9VtbJZGpljIzs3ni\nifk8/vh3ZGXl0qJFPU4/vb3XYRk/AkoQIhIDxKvquiDHE54OpR65ulleQrBKImPKbN++w/TrN4PV\nq5MBuP76PkyceBaNGtXxODLjT6kJQkTOA/4FxAAdRaQ3ME5VLwx2cJ7Yv6Ngq6C4SqKoGk5ncYEB\n6no51zYwxvhVv34M/fq1ISdHmT59KKef3sHrkEwJSq1iEpGfcYbr/lpVj3enrVDVkJ7vXulVTKqQ\n/kfRAer27yg6r1USGVNp0tMziYmJpnZtO8IdCsGuYspS1bRCA2RFdr1p1kF4ta9zkZvCYuKOHBrK\nSwhWSWRMme3bd5j69WOKTLdqpcgRyLfeahH5MxAlIh2B24AfghtWkCWvcJJDjTrQ9jSrJDKmEuXk\n5DJ16k889NDXzJv3F44/vpXXIZlyCiRB3AI8BOQC7wFzgfuCGVTQpa13/nYcAv/3rrexGFOFrFix\nk1GjPuaHH7YC8PbbqyxBRLBAEsQ5qno3cHfeBBG5CCdZRKa9boJoYJclNKYyHDqUzaOPfsvEiQvI\nzs6ldetYpkwZwoUXHud1aKYCAjmW8oCfafdXdiAhldeCaGgJwpjKkJZ2iOefX0x2di6jR/dl1aqb\nLDlUAcW2IETkHGAw0EZE/uXzVBzO4abIZQnCmErVsmV9XnppGC1b1ueUU+K9DsdUkpIOMe0CfgUO\nASt9pmcA9wQzqKDbawnCmMp28cVdvQ7BVLJAzoOoraqHQhRPSXFUznkQWQfh2bpO2eqYg1a+akwZ\nbN68l6lTF/PYYwPtwj0RItjnQbQRkceArkD+mWGqekx5Nui5dPes6Lj2lhyMCVBOTi5TpvzI/fd/\nxf79WXTs2IhRo07wOiwTZIF8Q/4HeBR4ChgCXEMknyiXZhVMxpTFsmU7+OtfP+Knn7YDcOmlXRk2\nLDJ/H5qyCaSKqa6qzgVQ1fWq+gBOoohM1kFtTMAWLNjMCSdM56efttO2bRwffjict966lFatbMyx\n6iCQFkSmiEQB60XkRmAbELnvDksQxgSsf/+29O3bmn792vDYY2cSG2vDZFQngSSIsUA9nCE2HgMa\nANcGM6igspPkjAlYdHQU3357DTEx0V6HYjxQaoJQ1UXu3QzgKgARaRPMoILKWhDGFKGqbNq0lw4d\nGhZ5zpJD9VViH4SInCgiF4hIU/dxNxGZBSwqabmwlZtz5NoODY/yNhZjwsTGjakMGfIqfftOZ/fu\n/V6HY8JIsQlCRJ4AXgWuAD4TkfHA18AyIDJLGPZtcy4JWrcF1KzndTTGeCo7O5ennlpIt24vMHfu\nenJzlVWrdnsdlgkjJR1iOh/opaoHRaQxsAXooaobQhNaENjhJWMAWL58J9dcM5slS5IAuPzy7kye\nfA4tWtT3ODITTkpKEIdU9SCAqqaIyNqITg5gCcIY18GDWSxdmkR8fAOmTj2Pc8892uuQTBgqKUEc\nJSJ5Q3oLzvWo84f4VtWLghpZMFgFkzEAnHRSW959988MGtTJ71XfjIGSE8TFhR5PCWYgIWEtCGPy\n2XDcpjTFJghVnRfKQELCEoSpRlSVWbOWsXp1MhMmnOV1OCYCBf3iyyIyWETWiMhaEbm7mHkSRGSp\niPwqIl8HJRBVG+bbVBvr1qVw1ln/Y+TI2UycuIClS5O8DslEoKAOZ+oO0TEFGAhsBxaLyGxVXeMz\nTwPgeeBsVd2Wd85FpTuUCpl7nfLWOs2CsgljvJaVlcOkSd/z8MPfcOhQNk2a1GHy5HPo3bul16GZ\nCBRwghCRWqqaWcb19wN+V9VN7jrewCmfXeMzzwjgXVXdBqCqyWXcRmB8Ww9i49ibqunRR7/lkUe+\nBeCqq3oyadLZNGtm5/yY8in1EJOI9BORFcDv7uNeIvJcgOtvg3P+RJ6t7jRfxwCNReRrEVksIlcF\nuO6ysWG+TTUwZkx/Tj65LXPnXsmsWRdacjAVEkgL4llgKPABgKouE5EzKjmGPsCZOIMCfi8i36vq\nusIzjh8/Pv9+QkICCQkJgW/FOqhNNdC4cR0WLLgWsVZytZWYmEhiYmKlrCuQBBGlqpsKveFyAlz/\nNsD3CuZt3Wm+tgLJ7mVND4nIt0AvoMQEUWaWIEwVsnPnPlJTD9GlS9EuO0sO1VvhH88PP/xwudcV\nSBXTFhHpB6iIRIvI7cDaANe/GOgsIu1FJAYYDnxYaJ7ZwKnuuusCJwGrA1x/4OwkOVMFqCozZy6h\nS5fnGT78HbKyAv2tZkzZBdKCGI1zmCke2Al86U4rlarmiMgtwOc4yWimqq4WkRucp3W6qq4RkbnA\ncpyWyXRVXVWO11Iya0GYCLd27R5GjfqIb77ZBDgX80lPz6RJk7oeR2aqKlEt+fLSItJYVVNCFE9J\ncWhpsRYr+xA8UwckGsYchOialRucMUH27LOLuOuuL8jMzKFZs7o888xghg/vboeTTKlEBFUt1xsl\nkBbEYhH5DXgTeE9VM8qzIU/lXQMirr0lBxORYmNjyMzM4dpre/Pkk2fTuHEdr0My1UAgV5TrJCJ/\nwuk/eFhEfgHeUNU3gh5dZbHDSybCjRzZm+7dm3PiiZF7MUcTeQIaakNVF6rqbTjlqOk4FxKKHDbE\nhokg/g6lioglBxNygZwoV19ErhCRj4Afgd3An4IeWWWyk+RMBNi+PYOLL36LF15Y7HUoxgCB9UH8\nCnwE/FNVvwtyPMFhh5hMGMvNVaZP/5m77/6S9PRMfvhhK3/9ax9q1QrqUGnGlCqQd+BRqpob9EiC\nyRKECVOrV+9m1KiPmT9/MwBDhx7D88+fa8nBhIVi34UiMklV7wTeFZEiB0Uj5opymgvpbhVTg47e\nxmJMITfd9Cnz52+mRYt6PPfcEC65pKuVrpqwUdLPlDfdv5F9JbmMbZBzGOo2h5hYr6MxpoDnnhvC\nc88tYsKEs2jUyEpXTXgp6YpyP7p3j1PVAknCPTs6Mq44Z0NsmDDWvXtzpk0b5nUYxvgVSJnrtX6m\nXVfZgQSN9T8Yj6kq7723mp0793kdijFlUlIfxGU4J8d1FJH3fJ6KBdKCHVilsQRhPLR1azo33/wp\nH374G5df3p3XXrvY65CMCVhJfRA/Antwhuh+3md6BrA0mEFVKksQxgM5OblMnfoT9903j4yMw8TF\n1WLAgHhU1TqhTcQoqQ9iI7ARZ/TWyJXfB3GUt3GYaiM7O5eEhP+wYIFzMcULL+zCc88NoU2bOI8j\nM6ZsSjrE9I2qni4iqYBvmavgDNXdOOjRVYa9G5y/1oIwIVKjRhT9+rVh48Y0pkwZwoUXHud1SMaU\nS7HDfYtIlKrmiki0v+dVNaRXKinXcN+HUuH5xlCzHtyaAda0NyGyf/9hsrNzadCgttehmGquIsN9\nF1vF5HP2dDsg2k0IJwM34Fw7Ovyl+RxesuRgguDgwSy/0+vVi7HkYCJeIGWuH+BcbrQT8DJwNPBa\nUKOqLNbTuXMQAAAgAElEQVRBbYJEVXnzzV856qhnSUz8w+twjAmKQBJErqpmARcBz6nqWCAyxh22\nk+RMEGzevJdhw15n+PB32bFjH//97zKvQzImKAIZESxbRC4FrgIucKdFxmXZrAVhKlFOTi5TpvzI\n/fd/xf79WTRoUIsnnxzEddf18To0Y4IikARxLXATznDfG0SkI/B6cMOqJPkJwkpcTcXt23eYiRMX\nsH9/Fpde2pVnnhlMq1Y2vpepuoqtYiowk0gNoLP7cJ2qZgc1Kv8xlL2KaXo8ZGyBa3+HRp1Ln9+Y\nUnz66e/k5OQybNixXodiTEAqUsVUaoIQkQHA/4BtOOdAtASuUtUF5dlgeZU5QWRnwjN1QKJgzEGI\njoyjYsYYU5mCUubqYzJwrqqeoqp/As4DninPxkJq70ZAIS7ekoMpk+TkA9x//zyyskJ6qo8xYSeQ\nPogYVV2V90BVV4tITBBjqhxWwWTKSFV59dUVjB07l+TkA9SvH8O99w7wOixjPBNIglgiIi8Cr7iP\nryASBuuzCiZTBhs2pDJ69Cd8/rnzvjnjjA5ccklXb4MyxmOBJIgbgduAu9zH3wHPBS2iymIJwgRo\n9erdnHDCdA4ezKZRo9pMmnQ2I0f2tlFXTbVXYoIQkR5AJ+B9Vf1naEKqJDaKqwlQly5NOfXUeJo0\nqcvTT59Dixb1vQ7JmLBQ0miu9+FcOW4JcKKIPKKq/w5ZZBWVZqO4msCICB9+eDm1awfSoDam+iip\niukKoKeqXgqcCIwOTUiVQHNtmG/j19at6X6nW3IwpqiSEkSmqu4HUNXdpcwbXvZth5xMqNMMYuxM\nVwO7d+/nyivfo0uXKWzaFDlXzDXGSyX9bDrK51rUAnTyvTa1ql4U1MgqwjqojUtVmTVrGXfc8Tkp\nKQepU6cGP/+cRPv2Db0OzZiwV1KCKHx19SnBDKRSWYIwwMaNqVx//UfMm7cRgEGDjmLq1PPo1Cky\nLoZojNdKuib1vFAGUqnsJDkDHDqUzbffbqJJkzpMnnwOV17Z00pXjSmDoPcriMhgEVkjImtF5O4S\n5jtRRLJEpOKHrmwUVwMcd1wz3nrrUtasuYWrruplycGYMgpq6YaIROEcmhoIbAcWi8hsVV3jZ74J\nwNxK2XBeBZO1IKq9Cy7o4nUIxkSsgFsQIlKrHOvvB/yuqpvcq9K9AZzvZ75bgXeAXeXYRlHWB1Gt\nfPLJWm6//TOvwzCmyik1QYhIPxFZAfzuPu4lIoEOtdEG2OLzeCuFLlcqIq2BC1R1Kk61VMUcSoND\nKVCjLtRrWeHVmfC1Y8c+LrvsHYYOfZ1nnlnEl19u8DokY6qUQFoQzwJDgT0AqroMOKMSY3ga8O2b\nqFiS2OvT/2DHnKskVeWll5Zw3HHP89ZbK6lbtyaTJp1NQkIHr0MzpkoJpA8iSlU3FergC3Sg/G1A\nvM/jtu40X32BN8TZQFNgiIhkqeqHhVc2fvz4/PsJCQkkJCQU3WKaVTBVdVOn/sTNN38KwODBnZk6\n9Tw6dLDzGowBSExMJDExsVLWFcgV5d4FJgIv4gy5cStwijsER2nLRgO/4XRSJwE/Aper6upi5n8Z\n+EhV3/PzXGBXlFv0BMy/D064AxImlT6/iTgHDmQxcOAsbrutH8OHd7fqJGNKUJErygXSghiNc5gp\nHtgJfEmA4zKpao6I3AJ8jnM4a6Z7waEbnKd1euFFAo68OPkVTFbiWlXVrVuThQuvtcRgTJCV2oII\nFwG3IN46E7Z8DRfNgY6Dgx+YCZr09Ey2bNlLt27NvQ7FmIgV1BaEiMzAzy97VR1Vng0GnZW4Vgmz\nZ6/h5ps/pVatGqxYMZq6de264saEWiCHmL70uV8buJCCpavhIzsTMraAREFce6+jMeWwfXsGt946\nh/fec7qp+vVrw549B6hbt4HHkRlT/ZSaIFT1Td/HIvI/YH7QIqqI9D8Ahdj2EB3jdTSmjF57bQWj\nR39Cenom9evH8PjjZ3LTTScSHR05I80bU5WUZ6iNjkCLyg6kUtjhpYhWr15N0tMzGTr0GF544Vza\ntbNWgzFeCqQPIpUjfRBRQApwTzCDKjdLEBHt/PO78N1313DKKe2sQsmYMFBignBPXuvFkZPbcgMr\nJfKIlbhGDFX1mwROPTXez9zGGC+UeHDXTQafqmqOewvf5ADWgogAaWmHuPHGj3n88e+8DsUYU4pA\n+iB+EZHjVXVp0KOpKLtQUNhSVd57bzW33jqHpKR91K8fw80396Nhw9peh2aMKUaxCUJEaqhqNnA8\nznUc1gP7cQbTU1XtE6IYA6O5Rw4xWQsirGzdms4tt3zK7Nm/AfCnP7Vj+vShlhyMCXMltSB+BPoA\n/xeiWCpmXxJkH4I6TaFWnNfRGB9jx85l9uzfiI2NYeLEs7jhhr5ERVkntDHhrqQEIQCquj5EsVTM\nXut/CFdPPjmIGjWieOqpQbRpY8nbmEhRUoJoJiJ3FPekqv4rCPGUnw3zHbY6dGjI669f7HUYxpgy\nKqmKKRqoD8QWcwsv+f0PVuLqlcTEP9i4MdXrMIwxlaSkFkSSqj4SskgqyloQnklJOchdd33BzJlL\nOeecTsyZc4Wd6GZMFVBqH0TEsHMgQk5Veeutldx222fs2rWfmJhoTjmlHbm5SnR0ZL19jDFFlZQg\nBoYsispgCSKkVJVLL32bd991Rl0dMCCe6dOH0aVLU48jM8ZUlmL7IFQ1JZSBVEjmXji0B2rUgXqt\nvI6mWhARevduSYMGtZg+fSiJiSMtORhTxVSNK8rtXAKvnABNusHIX0MbWDV2+HAOe/YcoFWr8KtZ\nMMY4KnJFuaox0L4dXgqqzMxsv9NjYqItORhThVWNBGGjuAbNl19uoFu3F3j//dVeh2KMCbGqkSCs\nBVHpkpMPMHLkBwwa9D/Wr09l2rSfvQ7JGBNi5bmiXPixYTYqjary6qsrGDt2LsnJB6hVK5px407n\nb3/7k9ehGWNCrGokCDtJrtIcPpzDww9/Q3LyAc44owPTpg3l6KObeB2WMcYDkZ8gcg5DxhaQKGjQ\nwetoIl6tWjWYMWMYGzemMnJkbzsj2phqLPLLXFPWwsvHQlx7uP6PkMdljDHhrHqXuVr/Q7ns33+Y\nRx75hv37D3sdijEmTEX+IaY0K3Etq7lz13HjjZ/wxx9ppKdn8tRTZ3sdkjEmDEV+grDrUAds1679\njB07l9deWwFAr14tuOyybh5HZYwJV5GfIOwciIAkJWXQvftUUlIOUrt2DR5+OIGxY/tTs2a016EZ\nY8KUJYhqolWrWM466yhSUg7y4ovn0alTY69DMsaEuciuYlKFZ+tB9kG4JQ1qNfAmuAhx4EAWderU\nsNJVY6qR6lvFtD/JSQ61m1hy8LFjxz6/0+vWrWnJwRgTsMhOEHZ4qYCMjExuv/0zOnZ8hpUrd3kd\njjEmwkV2H4SN4prvk0/WMnr0J2zZkk50tLBgwRa6dWvudVjGmAgW9BaEiAwWkTUislZE7vbz/AgR\nWebe5otIj4BXbi0Idu3az2WXvcPQoa+zZUs6J5zQisWLr2fUqBO8Ds0YE+GC2oIQkShgCs71rbcD\ni0Vktqqu8ZltA3Caqu4VkcHADKB/QBuwBEFmZjaffvo7devW5B//OIPbbjuJGjUi+8ihMSY8BPsQ\nUz/gd1XdBCAibwDnA/kJQlV/8Jn/B6BNwGu3YTZo164Br756ET17tqBDh4Zeh2OMqUKCnSDaAFt8\nHm/FSRrF+SswJ+C12zDfAPzf/x3rdQjGmCoobDqpReQM4Brg1OLmGT9+fP79hD+dSMLBZKhRG+q3\nCn6AHvv++y288spypkw510pVjTHFSkxMJDExsVLWFdQT5USkPzBeVQe7j+8BVFUnFpqvJ/AuMFhV\n1xezroInyu1cCq/0gSZdYeTKYL0Ez6WnZ3LfffN44YXFqMLrr1/M8OHdvQ7LGBMhKnKiXLBbEIuB\nziLSHkgChgOX+84gIvE4yeGq4pKDX9WgxHX27DXcfPOnbNuWQY0aUfz973/i/PPtcJIxJjSCmiBU\nNUdEbgE+xympnamqq0XkBudpnQ48CDQGXhDn2EmWqpbUT+Go4hVM7723mosvfguAfv3aMGPGMHr2\nbOFxVMaY6iTofRCq+hlwbKFp03zuXw9cX+YVV/FhvocNO4ZTTmnHZZd146abTiQ62kpXjTGhFTad\n1GVWxVsQNWtG891311iHtDHGM5H7s7SKJIjMzGx+/dX/uEmWHIwxXorMBJFzGDI2AwJxHbyOptzm\nz99M797TOOusWaSmHvQ6HGOMKSAyE0T6JtBciG0HNWp5HU2ZpaUd4sYbP2bAgJdZsyaZBg1qk5Tk\nf4huY4zxSmT2QeSVuDaMvBLXuXPXcc01s0lK2kfNmlHcc8+p3HffAGrXjsx/hTGm6orMb6UIHmKj\nTp2aJCXt4+ST2zJjxjAbktsYE7YiO0FEYAf1aae156uv/sLpp3cgKso6oY0x4csShAfOOKOj1yEY\nY0ypIrOTOsyH+T50KJsHHviKu+76wutQjDGm3CKvBaEKaXnjMIVfgkhM/INRoz7i999TiI4Wbrml\nH/HxDbwOq1ro0KEDmzZt8joMYzzRvn17/vjjj0pdZ+QliP07IPsA1G4MtcPnAjkpKQe5664vmDlz\nKQBduzZj+vShlhxCaNOmTQRzdGJjwlkwTqyNvAQRpqO4PvDAV8ycuZSYmGgeeGAAd911CrVqRd7u\nNcaYPJH3DRamHdTjxyewfXsGEyacRZcuTb0OxxhjKswSRCVp3rweH3ww3OswjDGm0kReFZPHw3wv\nW7aDVat2e7JtY4wJpchLEB61IA4ezOKee77khBOmc+21s8nJyQ3p9o2palatWsWJJ57odRgRYdeu\nXXTt2pWsrKyQbtcSRAC+/HIDPXpMZeLEBeTmKv36teHw4ZyQbd9UDR06dKBu3brExcXRunVrrrnm\nGg4cOFBgnoULFzJw4EDi4uJo1KgR559/PqtXry4wT0ZGBrfffjvt27cnLi6Oo48+mjvuuIOUlJRQ\nvpwKe+ihh7jrrru8DqNCDh8+zLXXXkuDBg1o3bo1kydPLnH+5557jqOOOoqGDRvSr18/FixYkP/c\n3XffTXx8PA0aNKBjx45MmDAh/7nmzZtz5plnMm3aNH+rDR5VjYgboJqZrvoUqpNrqebmaCjceuun\nCuMVxmv37i/o999vCcl2Tdk5b+fw1aFDB/3qq69UVXXnzp3aq1cvfeCBB/KfX7hwodavX1+fe+45\n3bdvn6ampuoDDzygjRo10o0bN6qq6uHDh7Vv37569tln65o1a1RVdffu3frYY4/pnDlzghZ7dnZ2\npa4vKSlJmzRpopmZmWERT3ndc889etppp+nevXt19erV2rJlS507d67feRctWqT16tXTpUuXqqrq\n1KlTtVmzZpqbm6uqqr/99ptmZGSoqur27du1W7du+v777+cvv2DBAu3evXuxsRT3/nenl+97t7wL\nhvoGqO78xUkQ/+5S7E6qbM8//6PWqvUPfeyxb/Xw4fB4Uxr/IiFBzJs3L//xXXfdpUOHDs1/PGDA\nAL3llluKLDdkyBC9+uqrVVV1xowZ2rJlSz1w4EDA2/3111910KBB2rhxY23ZsqU+8cQTqqo6cuRI\nffDBB/PnS0xM1LZt2xaId+LEidqzZ0+tXbu2Tpw4US+55JIC677tttt0zJgxqqq6d+9eve6667RV\nq1batm1bfeCBB/K//AqbNWuWDho0qMC0CRMmaKdOnTQ2NrbIl+N//vMfPeWUU3Ts2LHapEmT/Lhn\nzpypxx13nDZu3FgHDx6smzZtyl9mzJgx2q5dO42Li9O+ffvqd999F/A+C1Tr1q31yy+/zH/80EMP\n6eWXX+533jfffFNPOumk/Mf79+/XqKgo3bFjR5F5t27dqj169NAnn3wyf1p2drbWrVtXN2/e7Hf9\nwUgQkXWIyYMhNm68sS+rV9/MffcNoGbN6JBt1wTBJKm8WwVt3bqVOXPmcPTRRwNw8OBBFi5cyCWX\nXFJk3j//+c988YUzbMu8efMYPHgwderUCWg7+/btY9CgQZx77rkkJSWxbt06Bg4cWOz8hU+2euON\nN5gzZw5paWkMHz6cOXPmsH//fgByc3N5++23ueKKKwC4+uqriYmJYcOGDSxdupQvvviCl156ye92\nVqxYwbHHFrhUPZ07d2bBggWkp6czbtw4rrzySnbu3Jn//KJFi+jcuTO7du3i/vvvZ/bs2UyYMIEP\nPviA3bt3M2DAAC6//PL8+fv168fy5ctJTU1lxIgRXHrppRw+fNhvPBMnTqRRo0Y0btyYRo0aFbjf\nuHFjv8ukpaWRlJREz54986f16tWLlStX+p1/yJAh5OTk8OOPP5Kbm8vMmTPp3bs3LVq0KBBHbGws\n7dq148CBA4wYMSL/uejoaDp37syyZcv8rj8YIitBBHGY7+zsXHJzi56FGxUldOzYqNK3Z6qnCy64\ngLi4OOLj42nRogXjx48HICUlhdzcXFq1alVkmVatWpGcnAzAnj17/M5TnI8//phWrVpx++23ExMT\nQ7169crUMTxmzBhat25NrVq1iI+Pp0+fPrz//vuAk6zy1rdz507mzJnD5MmTqV27Nk2bNuX222/n\n9ddf97vetLQ0YmNjC0y7+OKL878sL730Uo4++mh+/PHH/OfbtGnDTTfdRFRUFLVq1WLatGnce++9\nHHPMMURFRXHPPffwyy+/sGXLFgBGjBhBw4YNiYqKYuzYsWRmZvLbb7/5jefuu+8mNTWVlJQUUlNT\nC9wvrm9n3759iAgNGhwZLSEuLo6MjAy/88fGxnLRRRdx6qmnUrt2bf7xj38wffr0InFkZGSwdOlS\nrrrqqgLrzltHWlqa3/UHQ2SdBxGkDuolS5L4618/5OabT+S66/pU6rpNGLnT+2E4Zs+ezRlnnMF3\n333HiBEjSE5Ozu+QjoqKIikpiWOOOabAMklJSTRt6px82aRJE5KSkgLe3pYtW+jUqfyfl7Zt2xZ4\nfPnll/P6669z5ZVX8vrrr+f/wt28eTNZWVn5ySvvEEV8fLzf9TZq1KjIF+msWbOYPHly/nhC+/fv\nz0+MAO3atSsw/6ZNmxgzZgx33nln/jZFhG3bttGuXTueeuop/v3vf+fvr4yMjALrq6j69esDkJ6e\nnv//2bt3b5HEl+ell17i5ZdfZvXq1XTq1Im5c+dy3nnn8csvv9CyZcsC8/bq1YvPPvuMhx56iEmT\nJuVPz8jIoGHD0A0xFJktiEpKEPv3H+Zvf/ucE0+cwdKlO5gyZbHfVoQxlUXdsaIGDBjA1Vdfnf/l\nVrduXU4++WTefvvtIsu89dZbnHXWWQCcddZZzJ07l4MHA7uGebt27Vi/fr3f5+rVq1egispf4il8\nyOnSSy8lMTGRbdu28f777+cniHbt2lG7dm327NmT/8s7LS2N5cuX+912z549Wbt2bf7jzZs3M2rU\nKF544YX8X/DdunUrMLZW4Vji4+OZNm0aKSkp+dvct28f/fv3Z/78+Tz55JO88847+euLi4srdqyu\nJ554gtjYWOLi4grc8qb507BhQ1q1alXgkM+yZcvo1q2b3/mXLVvGsGHD8hP2OeecQ6tWrVi4cKHf\n+bOzs9mwYUP+45ycHNatW0evXr38zh8U5e28CPUNUJ3R0emkTl7ttzOmLD777Hft0OFphfEaFfWw\n3nHHZ5qRUb6KChMeiLBO6t27d2u9evV0+fLlqqo6f/78/CqmjIwMTUlJ0fvvv18bNWqk69atU1XV\nzMxM7devnw4ZMkTXrFmjubm5mpycrI8//rjfKqaMjAxt3bq1PvPMM5qZmakZGRm6aNEiVXU6vI87\n7jhNSUnRpKQk7d+/v7Zr167YePMMGTJEBw0apH369Ckw/YILLtAxY8Zoenq65ubm6vr16/Wbb77x\nuy927typTZs2za9iWrVqldapU0fXrl2rOTk5+u9//1tr1KihM2fOVFWnk3rAgAEF1vH+++9r9+7d\ndeXKlaqqmpaWpm+//baqqn766afapk0b3bFjh2ZmZurDDz+sNWrU8Pt6KuKee+7RhIQETU1N1VWr\nVmnLli31888/9zvvf//7Xz322GN1w4YNqqr6+eefa7169fS3337T3NxcnTZtmqampqqqU/HUqlUr\nnTJlSv7yCxcu1G7duhUbS3Hvf6pNFdOkaNWnRDXrYLE7KRA5Obnap880hfHau/eLunjxtgqtz4SH\ncE8QHTt2LPIFddNNNxWoDFqwYIEmJCRo/fr1tUGDBjp06FBdtWpVgWXS09N17Nix2q5dO42NjdXO\nnTvrnXfeqSkpKX63u3LlSh04cKA2atRIW7VqpRMnTlRV1UOHDulll12mcXFx2qtXL3366acLJAh/\n8aqq/u9//9OoqCidNGlSkbhGjx6tbdu21YYNG2qfPn30zTffLHZ//PnPfy7w/AMPPKCNGzfWZs2a\n6Z133qkJCQklJghV1VdeeUV79OihDRo00Pj4eL3uuutUVTUnJ0evvfZajYuL09atW+uTTz5Z7Oup\niMzMzPzttGzZUp9++ukCz9evX1/nz5+f/3jcuHEaHx+vcXFx2rVrV3311VdVVTU3N1cHDx6sTZo0\n0djYWD322GN1woQJBdZ1880363PPPVdsLMFIEKLFNLnCjYioPgXUbws3bKnw+n7+eTvz5m1k7Nj+\nVp1URYhIsYcQTPhZvXo1I0eOZNGiRV6HEvZ2795NQkICS5cuJSYmxu88xb3/3enlKr2LvATR9nS4\nLNHrcEwYsgRhqrNgJIjI6qSGMnVQZ2Xl8K9/fU9KSmAdesYYY46IrDJXCDhB/PjjNq6//iOWL9/J\nr7/u4t//Pj/IgRljTNUSeQmilJPkMjIyefDBr3n22UWoQseODRk+vHuIgjPGmKoj8hJECS2IjIxM\nevSYyqZNe4mOFu6442TGj0+gbt2aIQzQGGOqhiqVIGJjazFkSGcWL97OjBnDOP74wIckMMYYU1Bk\nJYhaDaF2yeMiTZp0DjEx0dSoEXn976Zi2rdvX+RsW2Oqi/bt21f6OoOeIERkMPA0TsXUTFWd6Gee\nZ4EhwH5gpKr+4ndlPq2H5OQDNG1at8gsdjip+sobw8cYUzmC+jNbRKKAKcA5QDfgchHpUmieIUAn\nVT0auAF4sdgVNujE4cM5PProt8THT+b77yt+wlwkSkxM9DqEsGH74gjbF0fYvqgcwT4O0w/4XVU3\nqWoW8AZQuN70fGAWgKouAhqISAv8+H5rZ/r0mcaDD37NwYPZfPnlBn+zVXn25j/C9sURti+OsH1R\nOYJ9iKkN4PszfytO0ihpnm3utJ2F5uOUm2NQ3U3nzo2ZNm0oZ57ZsbLjNcYY44qoTuroaOHvfz+F\nBx88jTp1rK/BGGOCKahjMYlIf2C8qg52H9+DM7LgRJ95XgS+VtU33cdrgNNVdWehddkgO8YYUw7l\nHYsp2C2IxUBnEWkPJAHDgcsLzfMhcDPwpptQ0gonByj/CzTGGFM+QU0QqpojIrcAn3OkzHW1iNzg\nPK3TVfVTETlXRNbhlLleE8yYjDHGBCZihvs2xhgTWmF3urGIDBaRNSKyVkTuLmaeZ0XkdxH5RUR6\nhzrGUCltX4jICBFZ5t7mi0gPL+IMhUDeF+58J4pIlohcFMr4QinAz0iCiCwVkV9F5OtQxxgqAXxG\n4kTkQ/e7YoWIjPQgzKATkZkislNE/F8EnHJ+b5b3UnTBuOEkrHVAe6Am8AvQpdA8Q4BP3PsnAT94\nHbeH+6I/0MC9P7g67wuf+eYBHwMXeR23h++LBsBKoI37uKnXcXu4L+4FnsjbD8AeoIbXsQdhX5wK\n9AaWF/N8ub43w60FUakn1kW4UveFqv6gqnvdhz/gnD9SFQXyvgC4FXgH2BXK4EIskH0xAnhXVbcB\nqGpyiGMMlUD2hQKx7v1YYI+qZocwxpBQ1flAagmzlOt7M9wShL8T6wp/6RV3Yl1VE8i+8PVXYE5Q\nI/JOqftCRFoDF6jqVKAqV7wF8r44BmgsIl+LyGIRuSpk0YVWIPtiCtBVRLYDy4AxIYot3JTrezOi\nTpQz/onIGTjVX6d6HYuHngZ8j0FX5SRRmhpAH+BMoB7wvYh8r6rrvA3LE+cAS1X1TBHpBHwhIj1V\ndZ/XgUWCcEsQ24B4n8dt3WmF52lXyjxVQSD7AhHpCUwHBqtqSU3MSBbIvugLvCHOeN9NgSEikqWq\nH4YoxlAJZF9sBZJV9RBwSES+BXrhHK+vSgLZF9cATwCo6noR2Qh0AX4KSYTho1zfm+F2iCn/xDoR\nicE5sa7wB/xD4C+Qf6a23xPrqoBS94WIxAPvAlep6noPYgyVUveFqh7l3jri9EPcVAWTAwT2GZkN\nnCoi0SJSF6dTcnWI4wyFQPbFJuAsAPeY+zFAVR3lUyi+5Vyu782wakGonViXL5B9ATwINAZecH85\nZ6lq4cEQI16A+6LAIiEPMkQC/IysEZG5wHIgB5iuqqs8DDsoAnxfPAr8x6f88y5VTfEo5KARkdeA\nBKCJiGwGxgExVPB7006UM8YY41e4HWIyxhgTJixBGGOM8csShDHGGL8sQRhjjPHLEoQxxhi/LEEY\nY4zxyxKECRsikiMiS9xhqpe4JwIWN297EVlRCdv82h0u+hcR+U5Eji7HOm4QkSvd+1eLSEuf56aL\nSJdKjnORewZ9acuMEZHaFd22qb4sQZhwsl9V+6jq8e7fzaXMX1kn8Vyuqr1xRrt8qqwLq+o0VX3F\nfTgSn0HQVHWUqq6plCiPxDmVwOK8HahbSds21ZAlCBNOigwT4LYUvhWRn9xbfz/zdHV/VS9xf2F3\ncqdf4TN9qnu2eUnb/RbIW3agu9wyEXlJRGq60ye4F+H5RUT+6U4bJyJ3isjFOGNCveIuW9v95d/H\nbWX80yfmq0Xk2XLG+T3Q2mddL4jIj+JcEGecO+1Wd56vRWSeO+1sEVno7sc33WE4jCmWJQgTTur4\nHPyOUnYAAAK+SURBVGJ61522EzhLVfvijLXznJ/lbgSeVtU+OF/QW93DOpcBf3Kn5wJXlLL9/wNW\niEgt4GXgUlXthXMxmtEi0hhnSPHu7i/5R32WVVV9F2cQuBFuC+iQz/PvAhf6PL4MZ3DB8sQ5GPjA\n5/F97hArvYAEEemuqs/hDMaWoKoDRaQJcD8w0N2XPwN3lrIdU82F1VhMpto74H5J+ooBpohzicQc\nwF8fwffA/SLSDnhPVdeJyECcIa8Xu7/Ia+MkG39eFZGDwB84Fx06FtjgMwDif4GbgOeBgyLyEvAJ\nzpXr/CnSAlDVZBFZLyL9cEZVPVZVF4rIzWWMsxbOEN6+l4wcLiLX43yeWwJdgV8pOHhbf3f6Anc7\nNXH2mzHFsgRhwt1YYIeq9hSRaOBg4RlU9XUR+QEYCnziDtYmwH9V9f4AtjFCVZfmPXB/bfv7ks9x\nv+AHApcCt7j3A/UmTmthDfB+3ubKGqd7qGoKcLGIdMBpCZygquki8jJOkilMgM9VtbTWiTH57BCT\nCSf+jr03AJLc+38BoossJNJRVTe6h1U+BHriXJv6EhFp5s7TqISqqMLb/Q1oLyJHuY+vAr5xj9k3\nVNXPgDvc7RSWAcQVs533cS79OBzn8piUM86HgJNE5Bh3W/uADHGGsx7iM3+6Tyw/AKf49M/ULU/F\nlqleLEGYcOKvKukFYKSILMUZy3+/n3n+7HYcLwW6AbNUdTXwAPC5iCzDGRK6pZ9li2xTVTNxhkN+\nx102B3gR58v2Y3fatzitm8L+A7yY10ntu35VTcO5LkO8qv7kTitznG7fxiTg76q6HPjFXe8rwHyf\nZWYAn4nIPPe61NcAr7vbWYhzKM2YYtlw38YYY/yyFoQxxhi/LEEYY4zxyxKEMcYYvyxBGGOM8csS\nhDHGGL8sQRhjjPHLEoQxxhi/LEEYY4zx6/8B/2o8ycN3bnsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb88e320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(features, target, stratify=target,test_size=0.20)\n",
    "\n",
    "rf.fit(X_train, y_train)\n",
    "\n",
    "predictions = rf.predict(X_test)\n",
    "predictions = pd.Series(predictions)\n",
    "\n",
    "tn = len(predictions[(predictions == 0) & (y_test.values==0)])  \n",
    "tp = len(predictions[(predictions == 1) & (y_test.values==1)]) \n",
    "fn = len(predictions[(predictions == 0) & (y_test.values==1)])  \n",
    "fp = len(predictions[(predictions == 1) & (y_test.values==0)]) \n",
    "fpr = fp / (fp + tn)\n",
    "tpr = tp / (tp + fn)\n",
    "\n",
    "print('TPR: ',tpr)\n",
    "print('FPR: ',fpr)\n",
    "fprpoints, tprpoints, thresholds = roc_curve(y_test, predictions)\n",
    "print('CONFUSION MATRIX:')\n",
    "cm = confusion_matrix(y_test,predictions)\n",
    "print('TN: ',cm[0,0], '   FP: ',cm[0,1])\n",
    "print('FN: ',cm[1,0], '   TP: ',cm[1,1])\n",
    "plt.figure()\n",
    "lw = 2\n",
    "plt.plot(fprpoints, tprpoints, color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % auc(fprpoints,tprpoints))\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver operating characteristic example')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## RANDOM FOREST -  CV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TPR:  0.7076647243388615\n",
      "FPR:  0.04693675889328063\n",
      "CONFUSION MATRIX:\n",
      "TN:  1929    FP:  95\n",
      "FN:  3261    TP:  7894\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8ckShDHGGJ/+HxvhKVyE6HQwAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x18edb320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "kf = KFold(features.shape[0], random_state=1)\n",
    "\n",
    "predictions = cross_val_predict(rf, features, target,cv=kf)\n",
    "predictions = pd.Series(predictions)\n",
    "\n",
    "tn = len(predictions[(predictions == 0) & (target==0)])  \n",
    "tp = len(predictions[(predictions == 1) & (target==1)]) \n",
    "fn = len(predictions[(predictions == 0) & (target==1)])  \n",
    "fp = len(predictions[(predictions == 1) & (target==0)]) \n",
    "\n",
    "fpr = fp / (fp + tn)\n",
    "tpr = tp / (tp + fn)\n",
    "\n",
    "print('TPR: ',tpr)\n",
    "print('FPR: ',fpr)\n",
    "fprpoints, tprpoints, thresholds = roc_curve(target, predictions)\n",
    "print('CONFUSION MATRIX:')\n",
    "cm = confusion_matrix(target,predictions)\n",
    "print('TN: ',cm[0,0], '   FP: ',cm[0,1])\n",
    "print('FN: ',cm[1,0], '   TP: ',cm[1,1])\n",
    "plt.figure()\n",
    "lw = 2\n",
    "plt.plot(fprpoints, tprpoints, color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % auc(fprpoints,tprpoints))\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver operating characteristic example')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CONCLUSIONS"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "We were able to get TPR > 70% and FPR < 7% for both RF and LR (aprox)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TO DO NEXT\n",
    "\n",
    "* Understand how each feature affect the output (check LG coefficients). \n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}