10.8_Weight_of_evidence_WOE.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Weight  of evidence\n",
    "\n",
    "Weight of Evidence (WoE) was developed primarily for the credit and financial industries to help build more predictive models to evaluate the risk of loan default. That is, to predict how likely the money lent to a person or institution is to be lost. Thus, Weight of Evidence is a measure of the \"strength” of a grouping technique to separate good and bad risk (default). \n",
    "\n",
    "It is computed from the basic odds ratio: ln( (Proportion of Good Credit Outcomes) / (Proportion of Bad Credit Outcomes))\n",
    "\n",
    "WoE will be 0 if the P(Goods) / P(Bads) = 1. That is, if the outcome is random for that group. If P(Bads) > P(Goods) the odds ratio will be < 1 and the WoE will be < 0; if, on the other hand, P(Goods) > P(Bads) in a group, then WoE > 0.\n",
    "\n",
    "WoE is well suited for Logistic Regression, because the Logit transformation is simply the log of the odds, i.e., ln(P(Goods)/P(Bads)). Therefore, by using WoE-coded predictors in logistic regression, the predictors are all prepared and coded to the same scale, and the parameters in the linear logistic regression equation can be directly compared.\n",
    "\n",
    "The WoE transformation has three advantages:\n",
    "\n",
    "- It establishes a monotonic relationship to the dependent variable.\n",
    "- It orders the categories on a \"logistic\" scale which is natural for logistic regression\n",
    "- The transformed variables, can then be compared because they are on the same scale. Therefore, it is possible to determine which one is more predictive.\n",
    "\n",
    "The WoE also has three drawbacks:\n",
    "\n",
    "- May incur in loss of information (variation) due to binning to few categories (we will discuss this further in the discretisation section)\n",
    "- It does not take into account correlation between independent variables\n",
    "- Prone to cause over-fitting\n",
    "\n",
    "\n",
    "For more details follow this link:\n",
    "\n",
    "http://documentation.statsoft.com/StatisticaHelp.aspx?path=WeightofEvidence/WeightofEvidenceWoEIntroductoryOverview\n",
    "\n",
    "Let's see how to implement WoE in python"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Survived</th>\n",
       "      <th>Cabin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>C85</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>C123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Survived Cabin\n",
       "0         0   NaN\n",
       "1         1   C85\n",
       "2         1   NaN\n",
       "3         1  C123\n",
       "4         0   NaN"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's load again the titanic dataset\n",
    "\n",
    "data = pd.read_csv('titanic.csv', usecols=['Cabin', 'Survived'])\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Survived</th>\n",
       "      <th>Cabin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>Missing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>C85</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>Missing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>C123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>Missing</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Survived    Cabin\n",
       "0         0  Missing\n",
       "1         1      C85\n",
       "2         1  Missing\n",
       "3         1     C123\n",
       "4         0  Missing"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's first fill NA values with an additional label\n",
    "\n",
    "data.Cabin.fillna('Missing', inplace=True)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "148"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Cabin has indeed a lot of labels, here for simplicity, I will capture the first letter of the cabin, \n",
    "# but the procedure could be done as well without any prior variable manipulation\n",
    "\n",
    "len(data.Cabin.unique())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Survived</th>\n",
       "      <th>Cabin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>M</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>M</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>M</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Survived Cabin\n",
       "0         0     M\n",
       "1         1     C\n",
       "2         1     M\n",
       "3         1     C\n",
       "4         0     M"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Now we extract the first letter of the cabin\n",
    "\n",
    "data['Cabin'] = data['Cabin'].astype(str).str[0]\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['M', 'C', 'E', 'G', 'D', 'A', 'B', 'F', 'T'], dtype=object)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the labels\n",
    "data.Cabin.unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Important\n",
    "\n",
    "The calculation of the WoE to replace the labels should be done considering the ONLY the training set, and then expanded it to the test set.\n",
    "See below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((623, 2), (268, 2))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Let's divide into train and test set\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(data[['Cabin', 'Survived']], data.Survived, test_size=0.3,\n",
    "                                                    random_state=0)\n",
    "X_train.shape, X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Cabin\n",
       "A    0.428571\n",
       "B    0.774194\n",
       "C    0.571429\n",
       "D    0.692308\n",
       "E    0.740741\n",
       "F    0.666667\n",
       "G    0.500000\n",
       "M    0.303609\n",
       "T    0.000000\n",
       "Name: Survived, dtype: float64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# now we calculate the probability of target=1 \n",
    "X_train.groupby(['Cabin'])['Survived'].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Survived</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cabin</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A</th>\n",
       "      <td>0.428571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>0.774194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C</th>\n",
       "      <td>0.571429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D</th>\n",
       "      <td>0.692308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>E</th>\n",
       "      <td>0.740741</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F</th>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>G</th>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>M</th>\n",
       "      <td>0.303609</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Survived\n",
       "Cabin          \n",
       "A      0.428571\n",
       "B      0.774194\n",
       "C      0.571429\n",
       "D      0.692308\n",
       "E      0.740741\n",
       "F      0.666667\n",
       "G      0.500000\n",
       "M      0.303609\n",
       "T      0.000000"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# let's make a dataframe with the above calculation\n",
    "\n",
    "prob_df = X_train.groupby(['Cabin'])['Survived'].mean()\n",
    "prob_df = pd.DataFrame(prob_df)\n",
    "prob_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Survived</th>\n",
       "      <th>Died</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cabin</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A</th>\n",
       "      <td>0.428571</td>\n",
       "      <td>0.571429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>0.774194</td>\n",
       "      <td>0.225806</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C</th>\n",
       "      <td>0.571429</td>\n",
       "      <td>0.428571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D</th>\n",
       "      <td>0.692308</td>\n",
       "      <td>0.307692</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>E</th>\n",
       "      <td>0.740741</td>\n",
       "      <td>0.259259</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F</th>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>G</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>M</th>\n",
       "      <td>0.303609</td>\n",
       "      <td>0.696391</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Survived      Died\n",
       "Cabin                    \n",
       "A      0.428571  0.571429\n",
       "B      0.774194  0.225806\n",
       "C      0.571429  0.428571\n",
       "D      0.692308  0.307692\n",
       "E      0.740741  0.259259\n",
       "F      0.666667  0.333333\n",
       "G      0.500000  0.500000\n",
       "M      0.303609  0.696391\n",
       "T      0.000000  1.000000"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# and now the probability of target = 0 \n",
    "# and we add it to the dataframe\n",
    "\n",
    "prob_df = X_train.groupby(['Cabin'])['Survived'].mean()\n",
    "prob_df = pd.DataFrame(prob_df)\n",
    "prob_df['Died'] = 1-prob_df.Survived\n",
    "prob_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Survived</th>\n",
       "      <th>Died</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cabin</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A</th>\n",
       "      <td>0.428571</td>\n",
       "      <td>0.571429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>0.774194</td>\n",
       "      <td>0.225806</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C</th>\n",
       "      <td>0.571429</td>\n",
       "      <td>0.428571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D</th>\n",
       "      <td>0.692308</td>\n",
       "      <td>0.307692</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>E</th>\n",
       "      <td>0.740741</td>\n",
       "      <td>0.259259</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F</th>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>G</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>M</th>\n",
       "      <td>0.303609</td>\n",
       "      <td>0.696391</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T</th>\n",
       "      <td>0.000010</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Survived      Died\n",
       "Cabin                    \n",
       "A      0.428571  0.571429\n",
       "B      0.774194  0.225806\n",
       "C      0.571429  0.428571\n",
       "D      0.692308  0.307692\n",
       "E      0.740741  0.259259\n",
       "F      0.666667  0.333333\n",
       "G      0.500000  0.500000\n",
       "M      0.303609  0.696391\n",
       "T      0.000010  1.000000"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# since the log of zero is not defined, let's set this number to something small and non-zero\n",
    "\n",
    "prob_df.loc[prob_df.Survived == 0, 'Survived'] = 0.00001\n",
    "prob_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Survived</th>\n",
       "      <th>Died</th>\n",
       "      <th>WoE</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cabin</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>A</th>\n",
       "      <td>0.428571</td>\n",
       "      <td>0.571429</td>\n",
       "      <td>-0.287682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>0.774194</td>\n",
       "      <td>0.225806</td>\n",
       "      <td>1.232144</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C</th>\n",
       "      <td>0.571429</td>\n",
       "      <td>0.428571</td>\n",
       "      <td>0.287682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D</th>\n",
       "      <td>0.692308</td>\n",
       "      <td>0.307692</td>\n",
       "      <td>0.810930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>E</th>\n",
       "      <td>0.740741</td>\n",
       "      <td>0.259259</td>\n",
       "      <td>1.049822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F</th>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.693147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>G</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>M</th>\n",
       "      <td>0.303609</td>\n",
       "      <td>0.696391</td>\n",
       "      <td>-0.830169</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>T</th>\n",
       "      <td>0.000010</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-11.512925</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Survived      Died        WoE\n",
       "Cabin                               \n",
       "A      0.428571  0.571429  -0.287682\n",
       "B      0.774194  0.225806   1.232144\n",
       "C      0.571429  0.428571   0.287682\n",
       "D      0.692308  0.307692   0.810930\n",
       "E      0.740741  0.259259   1.049822\n",
       "F      0.666667  0.333333   0.693147\n",
       "G      0.500000  0.500000   0.000000\n",
       "M      0.303609  0.696391  -0.830169\n",
       "T      0.000010  1.000000 -11.512925"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# now we calculate the WoE\n",
    "\n",
    "prob_df['WoE'] = np.log(prob_df.Survived/prob_df.Died)\n",
    "prob_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'A': -0.2876820724517809,\n",
       " 'B': 1.2321436812926321,\n",
       " 'C': 0.28768207245178085,\n",
       " 'D': 0.81093021621632877,\n",
       " 'E': 1.0498221244986774,\n",
       " 'F': 0.69314718055994518,\n",
       " 'G': 0.0,\n",
       " 'M': -0.83016897812423662,\n",
       " 'T': -11.512925464970229}"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# and we create a dictionary to re-map the variable\n",
    "\n",
    "prob_df['WoE'].to_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# and we make a dictionary to map the orignal variable to the WoE\n",
    "# same as above but we capture the dictionary in a variable\n",
    "\n",
    "ordered_labels = prob_df['WoE'].to_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# replace the labels with the WoE\n",
    "\n",
    "X_train['Cabin_ordered'] = X_train.Cabin.map(ordered_labels)\n",
    "X_test['Cabin_ordered'] = X_test.Cabin.map(ordered_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Cabin</th>\n",
       "      <th>Survived</th>\n",
       "      <th>Cabin_ordered</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>857</th>\n",
       "      <td>E</td>\n",
       "      <td>1</td>\n",
       "      <td>1.049822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>D</td>\n",
       "      <td>1</td>\n",
       "      <td>0.810930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>386</th>\n",
       "      <td>M</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.830169</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>D</td>\n",
       "      <td>0</td>\n",
       "      <td>0.810930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>578</th>\n",
       "      <td>M</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.830169</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Cabin  Survived  Cabin_ordered\n",
       "857     E         1       1.049822\n",
       "52      D         1       0.810930\n",
       "386     M         0      -0.830169\n",
       "124     D         0       0.810930\n",
       "578     M         0      -0.830169"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the results\n",
    "\n",
    "X_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x5af86f2978>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xd4FWXa+PHvnYSeUJPQIYQWitQIKB10xYqFVRDFCqKi\n69p3f7uu7+v6rqvuWkFEbKCCXXEXO6CAtID0GmpCS2gh1LT798cMeIypcE7mnOT+XFeunJl5Zp57\n5pwz95n2PKKqGGOMMQBhXgdgjDEmeFhSMMYYc5olBWOMMadZUjDGGHOaJQVjjDGnWVIwxhhzmiWF\nECAij4vIO35c3jYRueAM5+0rIhv8FUshdbwlIn8vYvoREYk/g+XGiYiKSMTZRVgxiMhEEflrCcvO\nEZHbC5kWFNs9WOIIdpYUOL2TTBORGj7jbheROR6GFRTcL1GrU8OqOldV23oZk6pGquqWsqyzIu5Q\nVHWsqj7hdRxl5Wx+LPmp/iJ/DJUVSwq/CAf+cLYLEYdn27Ui7bRM4IhIuNcxhJryss0sKfziGeBB\nEald0EQROV9ElohIhvv/fJ9pc0TkSRGZDxwD4t1xfxeRn9zTHV+ISD0ReVdEDrvLiPNZxgsikuJO\nWyoifUsStIgMEJFUEXlERPYAb7rjLxOR5SJyyI2hUyHz9xCRBW653SLysohUdqf96BZb4a7Ddafq\n85m/nbuuh0RkjYhc4TPtLREZLyL/FZFMEVkkIi3daSIiz7lHaIdFZJWIdPQJrU5B87nznj56ceuY\nKCLfumV/EJHmxWy2W0Vkl7u+D/osN0xEHhWRzSKyX0Q+EJG67uRT2+KQuy3OE5HtItLdnXekG1cH\nd/g2EfmsBMtFRHq579EhEVkhIgN8ps0RkSdEZL67ft+ISHQh7+U6EbnMZzhCRNJFpJs7/KGI7HE/\nwz+eitVnO74iIjNF5CgwUHx+uYpIHRH5j7u8g+7rJvlCaCkii93383PfdcwXZy0Red3d/jvF+Z4U\nuEMt6vPpTlcRGSsim9wy40VE3GnhIvKsiOwTkS3ApQXV4ZadCjQDvnDf34fPcJvVE+e7fuo7/ncR\nmeczT4L7WT0gIhtE5Fp3/BhgJPCwW/8XhcUacKpa4f+AbcAFwCfA391xtwNz3Nd1gYPAjUAEMMId\nrudOnwPsADq40yu545KBlkAtYC2w0a0nApgCvOkTww1APXfaA8AeoKo77XHgnUJiHwDkAP8EqgDV\ngK5AGtAT5wjoJncdq/iur/u6O9DLrTcOWAfc57N8BVrlqy/VfV3JXcc/A5WBQUAm0Nad/hawH+jh\nLv9dYLo77SJgKVAbEKAd0LC4+fLH5JbNBPq56/8CMK+QbRXnzjsNqAGcA6T7bIs/AAuBJu6yXgWm\n5Zs3wmd5U4AH3NeTgM3AnT7T/liC5TZ21/USnB9pF7rDMT6frc1AG/e9nQM8Vcj6PQa86zN8KbDO\nZ/hWIMqN4Xlguc+0t4AMoLcbR1V33KnvQz3gGqC6u4wPgc985p8D7AQ6utv2Y9zPbP5tB3zqboMa\nQCywGLijkHUqyefzPzifo2bu+znEnTYWWA80xfkOz87/Hha0H8g3rrTbbLr7Vx1oD6Tgfh7d9U0B\nbnHXpyuwD2jvs7y/e74/9DqAYPjjl6TQ0X2TY/h1UrgRWJxvngXAzT5fiP/NN30O8P98hv8FfOkz\nfLnvB6yAmA4Cnd3Xj1N0UsjCTSDuuFeAJ/KV2wD0913fQpZ3H/Cpz3BRSaEvTvIK85k+DXjcff0W\nMNln2iXAevf1IJwk2ct3/uLmyx+TW9Y3YUQCuUDTAtYtzp03wWfc08Dr7ut1wGCfaQ2BbH7ZIeVP\nCrcBM3zmvZ1fkt52oFsJlvsIMDVfnF8DN/l8jv7iM+0u4KtC3rtWOAmyujv8LvBYIWVru+tTy2c7\nTingfShwJwV0AQ7m+7w/5TPcHudzGe677YD6wEmgmk/ZEcDsEn5XC/p89vEZ/gB41H09CxjrM+13\n+d/DgvYDRdRd5DZz1zUb90eRO+7v/JIUrgPm5lvmq8DfitveZflnp498qOpqnF8dj+ab1AjnS+5r\nO86vvFNSCljkXp/XxwsYjjw1ICIPuof/GSJyCOfoosDTBAVIV9UTPsPNgQfcw+lD7vKauuvxKyLS\nxj0VsEdEDgP/V4p6GwEpqprnMy7/dtnj8/oY7jqr6izgZWA8kCYik0SkZnHzFeL0tlfVI8ABCljX\ngsq78Z4q2xz41GebrcNJMPULWc4PQF8RaYizQ/gA6C3OacFawPISLLc58Pt871UfnMRxSom2haom\nu8u+XESqA1cA78HpUylPuaewDuPsAOHX73VBn2Hc+auLyKvinDI7jHM6rXa+0z75t2slfvtZau6O\n3+2zvq/iHDEUVG9JPp+FbZ9GBcRUYmewzWJwEl9KIdObAz3zvdcjgQaliSvQLCn81t+A0fx6x7YL\n5w311QzncPkUPdMKxbl+8DBwLVBHVWvjHLFICReRv+4U4ElVre3zV11VpxUw7ys4h9itVbUmzqmg\nkta7C2gqv76wnn+7FB606ouq2h3nV2Ub4KES1ptf01MvRCQS51TBrpKUx4n3VNkU4OJ8262qqu6k\ngPfX3QkfA+4BflTVwzg7qDE4vw7zSrDcFJwjBd9pNVT1qTPZEDhHaiOAocBaN0aA691xF+AkrDh3\nvO97XdRn+AGgLdDT/Zz0K2D+/Ns1G+f0iK8UnCOFaJ/1ramqHSjY2Xw+dxcQU1Hyr39pt1k6zqlc\n32stvvWnAD/ke68jVfXOQur3hCWFfNwv0fvAvT6jZwJtROR69+LddTg7sv/4qdoonA9TOhAhIo8B\nNYuepUivAWNFpKc4aojIpSISVUjdh4EjIpIA3Jlv+l6gsGcCFuHsFB8WkUriXCC9HOecapFE5Fw3\nvkrAUeAEkFfMbIW5RET6uBcgnwAWqmqhv3qBv7q/fDvgnN993x0/EXhS3AvVIhIjIkPdaelufPm3\nxQ/AOPc/OKdRfIeLW+47OL/sL3J/mVYV52J+/ou4JTUd5zTJnbhHCa4onJ3xfpzz3f9XyuVG4Rzd\nHhLnAvLfCihzg4i0d49S/hf4SFVzfQuo6m7gG+BfIlJTnIvwLUWkfxH1FvX5LMoHwL0i0kRE6vDb\nMwD55f+sl2qbuev6CfC4+/lKAEb5FPkPzn7kRvf7Usn9HrQrpH5PWFIo2P/iXBQCQFX3A5fh/Fra\nj/Or/jJVzf8r6Ex9DXyFc459O84OsqidWpFUNQnnaOdlnGsTycDNhRR/EOcXUSZOMnk/3/THgbfd\nw91r89WThZMELsb5RTgBGKWq60sQZk23voM467wf5w6wM/Eezk7qAM6FyRuKKf8Dzjb5HnhWVb9x\nx78AzAC+EZFMnIvDPQFU9RjwJDDf3Ra9fJYVxS93J+UfLm65KTi/Rv+Mk3hScI6Yzui76e50FwDn\n8+v3cgrOdt6Jc9PDwlIu+nmcC9373Hm/KqDMVJzz4ntwLrreW0AZcHaUld04DgIf8evTZb6K+3wW\n5TWc79YKYBnODrso/wD+4r6/D3Jm22wczlHFHpztMQ0nsaCqmTgJezjO0ekefrlBBOB1oL1b/2cl\nXEe/E/cChzEhSUTewrnw/RevYzEmPxH5J9BAVW/yOpaSsiMFY4zxE/c5hE7uadseOHeofep1XKVh\nT78aY4z/ROGcMmqEc43gX8DnnkZUSnb6yBhjzGl2+sgYY8xpAT19JCJDcO68CMd5QvWpfNNr4dyS\n18yN5VlVfbOoZUZHR2tcXFxgAjbGmHJq6dKl+1Q1prhyAUsK7pOO43HackkFlojIDFVd61PsbpwH\nbC4XkRhgg4i8697qWKC4uDiSkpICFbYxxpRLIlKiJ7oDefqoB5Csqlvcnfx0nPuxfSkQJSKC82j6\nAZyHuIwxxnggkEmhMb9+ACuVXzcdAc7DVe1wHuRYBfwhXzs6xhhjypDXF5ovwmk0rBFOq4sv52sU\nDXDaGheRJBFJSk9PL+sYjTGmwghkUtjJrxuDasJvG0q7BfhEHcnAViAh/4JUdZKqJqpqYkxMsddJ\njDHGnKFAJoUlQGsRaeE2VDYcp/0XXzuAwQAiUh+nFcYy7XvXGGPMLwJ295Gq5ojIOJwGqcKBN1R1\njYiMdadPxGnR8i0RWYXTHO0jfmxkzhhjTCkF9DkFVZ2J0+y077iJPq934bQaaIwxJgh4faG5XFq9\nM4OFW/Z7HYYxxpSaJQU/y8rJY/SUJG6YvIhFlhiMMSHGkoKffbZ8J7szThBVNYK73l3GzkPHvQ7J\nGGNKzJKCH+XmKRPnbKZDo5p8OPZ8snLyuGNqEsezcouf2RhjgoAlBT/6es0etuw7yp0DWtIqNpLn\nh3dhza7DPPrJSqyJcmNMKLCk4CeqyvjZybSIrsHFHZ3uZge3q88DF7bh8+W7eG2uPX5hjAl+lhT8\n5MdN+1iz6zBj+8cTHianx989sBWXnNOAp75czw8brYkOY0xws6TgJxNmJ9OwVlWu6trkV+NFhGeG\ndaZN/SjueW8Z2/Yd9ShCY4wpniUFP1i6/QCLth7g9r7xVI747SatUSWC10YlEhYmjJ6SxJGT1jq4\nMSY4WVLwgwmzN1OneiVG9GhaaJmmdasz/vpubNl3lPvfX05enl14Lg/SM0/yQVIKY6cupceT33H/\n+8tZu+uw12EZc8YC2sxFRbBu92G+X5/G/Re2oXrlojdn71bR/PmSdjzxn7W8OGsT913QpoyiNP6i\nqqzZdZjv16Uxa/1eVqRmANCgZlW6N6/DV2v28MnPO+nTKprR/eLp1zoapw8pY0KDJYWz9MqczdSo\nHM5N58WVqPytveNYu+swz3+3iXYNa3JRhwaBDdCctaMnc5iXvI/Z69OYtT6NtMyTiECXprV58Hdt\nGJRQn3YNoxARMo5l897iHbz101ZuemMxbetHcXvfFlzRpRFVIsK9XhVjiiWhdv98YmKiBksfzdv3\nH2Xgs3MY3TeeP13SrsTzncjO5bpXF5CcdoRP7+5Nm/pRAYzSnImUA8f4ft1eZm1IZ+Hm/WTl5hFV\nJYJ+bWIYlBDLgLYx1IusUuj8WTl5fLHCuRV5/Z5MYqOqcHPvOEb2aE6t6pXKcE2McYjIUlVNLLac\nJYUz96dPVvHxslTmPTyQ2JpVSzXv7ozjXP7SfCKrhPP53X1sR+GxnNw8lm4/yCz3aGBT2hEA4mNq\nMKhtLIPaxXJuXF0qhZfuMpyqMnfTPl6bu4W5m/ZRvXI41yY25bY+LWhat3ogVsWYAllSCLC9h0/Q\n95+z+X1iE5686pwzWkbStgOMeG0hveLr8dYtPX71fIMJvINHs/hhYzqz1qcxZ0Mah0/kUClc6Nmi\nHoMSYhmUEEtcdA2/1bd212Emz9vCFyt2kZunXHxOQ0b3jadL09p+q8OYwlhSCLAn/7uW1+dtZc6D\nA2lW78x/8b23aAd//nQVd/Qr3SkoU3qqysa9R/h+/V5mrUtj2Y6D5ClER1ZmYNtYBreLpXeraKKq\nBvaobU/GCd76aRvvLtpO5okcesTVZXS/eAYnxBJmPwxMgJQ0KdiF5jNw6FgW7y7aweWdG51VQgC4\nvmcz1uzK4NUft9C+UU2GdmnspygNONdvFmzZz6x1zmmhU63Wdmxck3GDWjM4IZZzGtcq051xg1pV\nefTiBMYNasX7S1J4Y95WRk9JIj66Brf1bcE13ZpQtZJdlDbeCOiRgogMAV7A6Y5zsqo+lW/6Q8BI\ndzACaAfEqOqBwpYZDEcKL3y3iee+28hX9/UloUHNs15eVk4eIycvZGVqBh/feT4dG9fyQ5QV156M\nE+61gb3MS97Hiew8qlcOp3eraAYnxDIwIZb6pbwGFEg5uXl8uXoPr83dwsrUDOrWqMyNvZoz6rzm\nRV7MNqY0PD99JCLhwEbgQiAVWAKMUNW1hZS/HPijqg4qarleJ4WjJ3Po/c9ZJDavw+SbzvXbctMz\nT3LFy/MQYMY9fYi2nUGJ5eYpK1IPMXt9Gt+vS2PtbufhsaZ1qzE4oT4DE2Lp2aJu0P/6VlUWbz3A\na3O38N26NKpEhHFN9ybc1qcFLWMivQ7PhLhgOH3UA0hW1S1uQNOBoUCBSQEYAUwLYDx+MW3xDg4d\ny+auga38utyYqCpMujGRYRN/4q53l/Hu7T1LfadLRXL4RDZzN+47fZF4/9EswsOE7s3r8KeLExiU\nEEur2MiQenBMROgZX4+e8fVITjvC6/O28NHSVKYt3sHghPqM6RfPuXF1QmqdTOgJ5JHCMGCIqt7u\nDt8I9FTVcQWUrY5zNNGqoFNHIjIGGAPQrFmz7tu3bw9IzMU5mZNLv6dn0yK6BtPHnBeQOj77eSf3\nvb+cUec153+HdgxIHaFqS/oRZrlHA0u2HSAnT6ldvRID2sQwqF19+reOKXe39u47cpIpC7YzdcE2\nDh7LpnPT2ozu24IhHRoQYT8aTCkEw5FCaVwOzC/sWoKqTgImgXP6qCwD8/Xpsp3sPXySZ4Z1Dlgd\nV3ZtzJpdGbw2dysdGtXkunObBayuUHA8K5eXZ29i5qo9bHVbmG1bP+r03TpdmtYu1zvH6Mgq3H9h\nG+7s35KPlqXy+twtjHvvZ5rUqcZtfVpwbWJTalQJlq+xKQ8C+WnaCfi2ENfEHVeQ4QT5qaPcPGXi\nD5s5p3Et+raODmhdjwxJYP2eTP7y2WpaxUbRvXmdgNYXrDbuzWTce8vYuPcI/dvEcGvvOAYmxNKk\nTsV76Kta5XBu7NWc63s047t1e3ntxy38zxdree7bjYzs1Zybz48LqovnJnQF8vRRBM6F5sE4yWAJ\ncL2qrslXrhawFWiqqsV2NuDVheYvVuzinmk/88rIblx8TsOA13foWBZXvDyf49m5fDGuDw1qVZwv\nvKry/pIUHv9iDZFVIvj3tV3o1ybG67CCzrIdB5k8dwtfrd5DeJgwtEtjRveNp20DazbF/Jbndx+5\nQVwCPI9zS+obqvqkiIwFUNWJbpmbca49DC/JMr1ICqrKJS/OIysnl2//2L/M7mnfsCeTqybMp3X9\nKN4f0yvo757xh8wT2fz509V8sWIXvVvV47nruhAbVXES4pnYvv8ob8zbygdJqRzPzqVfmxjG9I2n\nd6t6dlHanBYUSSEQvEgKs9encctbS3hmWCd+n1h4nwmB8NXqPYx9ZynDujfhmWGdyvWXfGXqIe6Z\n9jOpB49z/4VtGNu/pTX9UQqnHqp8c/429h05SbuGNRndtwWXdWpUYOdPpmIpaVKwT0oJTJiTTKNa\nVT152nhIxwbcO7g1Hy1N5a2ftpV5/WVBVXl93laueeUnsnPymD6mF3cPbGUJoZRqV6/M3QNbMf/R\ngTw9rBM5uXnc/8EK+j09m1d/2MzhE9leh2hCgN22UIzFWw+wZNtBHr+8vWe/tu4b3Jp1uw/z9/+u\no239KM5vFdgL3WXpwNEsHvpwBd+vT+PC9vV5Zlgnalev7HVYIa1KhNMS6++7N2HOxnQmz93CP75c\nz4vfb2J4j2bc0juuQl6sNyVjp4+KcfObi1mVmsG8RwZRrbJ35/QzT2Rz1YSf2H/kJDPG9SkXzS4v\n2rKfP0xfzoGjWfz5kgRuOj+uXJ8e89LqnRlMnruFL1buJkzgxeFdy+SGCRM87PSRH6zZlcGcDenc\n2qeFpwkBIKpqJV4blUhunjJ6ShLHsnI8jeds5OYpL3y3iRGvLaRqpTA+uet8bu7dwhJCAHVsXIvn\nh3dl7sMD6dSkNvdM+5mv1+zxOiwThCwpFGHCnM1EVonghl7NvQ4FgBbRNXhxRFc27s3koQ9XEmpH\neeD0Q3HD5EU8991GrujciP/c29caACxDjWpX461bzqVj41qMe28Z36/b63VIJshYUijE1n1H+XLV\nbm48rzm1qgVP0wkD2sby8JAE/rtqNxPmbPY6nFKZsyGNS16Yy/KUQzw9rBPPXdeFSHsat8xFVa3E\n27f2oF3Dmtz5zjJmb0jzOiQTRCwpFOLVHzZTKTyMW3u38DqU37ijXzyXd27Es99sYNb64P+ll52b\nxz9mruPmN5cQE1WFL+7pzbWJTe10kYdqVavE1Ft70rp+JHdMXcqPG9O9DskECUsKBdidcZyPl6Vy\nbWJTYqKCrwlrEeHpazrRrkFN/jBtOZvTj3gdUqFSDhzj9xMX8OqPWxjZsxmf3d2bVrH2xG0wqFW9\nEu/c1pOWMZGMnpLE/OR9XodkgoAlhQJMnruVPIUx/eK9DqVQ1SqHM2lUdypFhDF6SlJQ3oM+c9Vu\nLnlxLpvTjzD++m48edU5FeKp7FBSp0Zl3r29J3H1anDb20tYsHm/1yEZj1lSyOfA0SzeW7SDoZ0b\nBf1tn03qVGf89d3Yvv8Yf5y+nLy84LjwfCI7l//36SruencZLWMimXlvXy7tZLc/Bqu6NSrz7uie\nNK1TndveXsLirYV2fGgqAEsK+bz10zaOZ+dy54CWXodSIue1rMdjl7Xn+/Vp/PvbjV6HQ3JaJleO\nn8+7i3ZwR794Phx7XtAnV+M00f3u6J40qFWVW95czNLtlhgqKksKPo6czOHtn7bxu/b1aV0/dM57\njzqvOdcmNuHl2cnMXLXbkxhUlQ+SUrj8pfmkZZ7kzVvO5U+XtLPe40JIbFRVpo3uRWzNqtz0xhJ+\n3nHQ65CMB+wb6+O9RdvJOO7/rjYDTUR44sqOdG1Wmwc/XMH6PYfLtP4jJ3P44/vLefijlXRpWpsv\n/9CXgW1jyzQG4x/1a1blvdE9qVujMqNeX8zK1ENeh2TKmCUF14nsXCbP3UrvVvXo0rS21+GUWpWI\ncCbe0J3IKhGMnpLEwaNZZVLv6p0ZXP7SPGas2MX9F7bhndt7WmcvIa5hrWpMG9OLWtUrccPkRaze\nmeF1SKYMWVJwfbwslbTMk9w1ILSOEnzVr1mVV2/szt6Mk4ybtoyc3LyA1aWqvDV/K1dP+InjWblM\nG92Lewe3tpZNy4nGtasxbXQvoqpW4obXF7F2V9kefRrvWFIAcnLzePWHLXRuWpvzW9bzOpyz0rVZ\nHf5+VUfmJ+/nH1+uD0gdh45lMWbqUh7/Yi19W0cz8w996Rkf2tvN/FbTutWZNroX1SqFM3LywjI/\nLWm8YUkB+O+q3ew4cIy7BrQsF0/ZXpvYlJvPj+P1eVv5eGmqX5e9ZNsBLnlhLnM2pPHXy9oz+aZE\n6tawpq7Lq2b1nMRQOSKMka8tYtPeTK9DMgEW0KQgIkNEZIOIJIvIo4WUGSAiy0VkjYj8EMh4CqKq\nvDJnM61jI7mwXf2yrj5g/t+l7Tgvvh5/+nQVK1LO/mJhbp7y8qxNDJ+0kEoRYXx85/nc1sdaNq0I\n4qJr8N7oXoSFCSNeW0RyWvA+QW/OXsCSgoiEA+OBi4H2wAgRaZ+vTG1gAnCFqnYAfh+oeAoza30a\n6/dkcueAlmXW93JZqBQexviR3YiJrMIdU5eSlnnijJeVlnmCUW8s4tlvNnLJOQ35zz196NQk9C7G\nmzPXMiaSaaN7Asr1ry1k676jXodkAiSQRwo9gGRV3aKqWcB0YGi+MtcDn6jqDgBVLdPmGlWV8bOT\naVy7Gpd3blSWVZeJujUqM2lUdw4dz+Kud5aRlVP6C88/bkznkhfmsnT7QZ66+hxeHN6FqKrB02qs\nKTutYqN4b3QvcvKUEZMWsn2/JYbyKJBJoTGQ4jOc6o7z1QaoIyJzRGSpiIwqaEEiMkZEkkQkKT3d\nf605Ltp6gGU7DjG2f3y5fciqQ6NaPDOsM0nbD/K3GWtKPF92bh7//Go9o95YTN0alZkxrg/DezSz\n00UVXJv6Ubx7e09O5uQyYtJCUg4c8zok42de7wkjgO7ApcBFwF9FpE3+Qqo6SVUTVTUxJibGb5WP\nn51MdGRlfp/Y1G/LDEaXd27EnQNaMm3xDt5ZuL3Y8qkHj3Hdqwt4Zc5mRvRoxud396FNCD3hbQKr\nXcOavHN7T45m5TLitYWkHrTEUJ4EMinsBHz3tk3ccb5Sga9V9aiq7gN+BDoHMKbTVqVmMHfTPm7t\n06JCtNz54O/aMqBtDI/PWFNkg2dfrd7NJS/MZdPeI7w0oiv/uPocz7siNcGnQ6NavHNbTzKOZ3P9\na4vYdei41yEZPwlkUlgCtBaRFiJSGRgOzMhX5nOgj4hEiEh1oCewLoAxnfbKD8lEVQ2erjYDLTxM\neGF4V5rVrc5d7y79zZf4RHYuj32+mrHvLCMuugb/vbdvubzOYvznnCa1mHpbTw4ezeL61xayJ+PM\nb2YwwSNgSUFVc4BxwNc4O/oPVHWNiIwVkbFumXXAV8BKYDEwWVVXByqmU5LTjvDl6j2MOq85NSvQ\nRdNa1SoxaVR3TmTnMWZqEieycwHYnH6Eqyb8xJQF27m9Tws+Gns+zepZy6ameF2a1ubt23qw74iT\nGNIOW2IIdRJqnb8nJiZqUlLSWS3joQ9X8MXKXcx7ZBDRkcHXs1qgfbd2L6OnJjG0cyP6to7hr5+v\npkpEGP+6tjODEsrPsxqm7CRtO8CoNxbTyG0eIxh7LKzoRGSpqiYWV87rC81lbueh43z6806Gn9us\nQiYEgAva1+f+C9rw2fJdPPDhCjo2rsXMP/S1hGDOWGJcXd68+Vx2HjzOyMkL2X/kpNchmTMU4XUA\nZe21H7cAMDqIu9osC3cPbEXmyRxqVavE2P4trSE7c9Z6xtfj9ZsTufWtJYycvIj3RveyJlBCUIU6\nUth/5CTTl+zgyq6NaVy7mtfheCosTPjzJe24e2ArSwjGb85vGc3kUeeydd9Rbpi8iEPHyqYJd+M/\nFSopvDl/Gydz8hjbPzS62jQmFPVpHc2kUYkkpx3hxtcXk3E82+uQTClUmKSQeSKbtxdsY0iHBrSK\njfQ6HGPKtf5tYph4YzfW7znMqDcWc/iEJYZQUWGSwjsLd5B5IiekO9ExJpQMSqjPhJHdWbMzg5ve\nWEymJYaQUCGSwonsXF6ft5W+raM5p0ktr8MxpsK4sH19Xr6+GytTM7jlzSUcPZnjdUimGBUiKXy4\nNJV9R0K7q01jQtWQjg14cXhXfk45xC1vLeFYliWGYFbuk4LT1eZmujarTa/4ul6HY0yFdGmnhvz7\n2s4kbTsYHLyOAAAbT0lEQVTAbW8lcTwr1+uQTCHKfVL4YuUuUg8e5+4BrazZZ2M8NLRLY/51bWcW\nbt3P6Cm/NLNigku5Tgp5ecqE2ZtpWz+KQQmxXodjTIV3VdcmPDOsM/M37+OOqUstMQShcp0Uvlu3\nl01pR7hrYPnqatOYUDasexOeuvocftiYzl3vLuNkjiWGYFJuk4KqMn7OZprVrc6l5zT0OhxjjI/r\nzm3Gk1d1ZNb6NMa99/MZdRVrAqPcJoUFm/ezIuUQY/rFE1FOu9o0JpSN7Nmc/x3agW/X7uXeaT+T\nnWuJIRiU273lhDmbiYmqwrDuTbwOxRhTiFHnxfHYZe35as0e7pu+nBxLDJ4rl62krkg5xLzkffzp\n4oQK0dWmMaHs1j4tyM1Tnpy5jvAw4bnrulgjjR4K6JGCiAwRkQ0ikiwijxYwfYCIZIjIcvfvMX/U\nO2FOMjWrRjCygnS1aUyoG90vnkeGJDBjxS4e+nAFuXmh1flXeRKwIwURCQfGAxcCqcASEZmhqmvz\nFZ2rqpf5q95NezP5es1e7h3Uisgq5fJAyJhy6c4BLcnNy+PZbzYSFiY8fU0nu2vQA4Hca/YAklV1\nC4CITAeGAvmTgl+98sNmqlUK5+beLQJZjTEmAMYNak1OnvL8d5uICBP+cfU59tBpGQvk6aPGQIrP\ncKo7Lr/zRWSliHwpIh0KWpCIjBGRJBFJSk9PL7TClAPH+Hz5Lkb0aGY9PhkTov4wuDV3D2zJ9CUp\njJ+d7HU4FY7Xdx8tA5qpaifgJeCzggqp6iRVTVTVxJiYmEIX9trcLYQJjO5nRwnGhCoR4cHfteXK\nLo149puNfLlqt9chVSiBTAo7gaY+w03ccaep6mFVPeK+nglUEpHoM6ksPfMk7y9J4equTWhYq2J3\ntWlMqBMRnrqmE12b1eb+D1awemeG1yFVGIFMCkuA1iLSQkQqA8OBGb4FRKSBuCcMRaSHG8/+M6ns\nzflbycrN447+8WcZtjEmGFStFM6kGxOpU70St7+dRNrhE16HVCEELCmoag4wDvgaWAd8oKprRGSs\niIx1iw0DVovICuBFYLiqlvpetMMnspm6YDuXdGxIfIx1tWlMeRETVYXJN53L4RPZ1rJqGQnoNQVV\nnamqbVS1pao+6Y6bqKoT3dcvq2oHVe2sqr1U9aczqWfqgu1knszhzgEt/Rm+MSYItG9Uk+eu68LK\nnRk89NFKzuB3oykFry80n7XjWbm8MW8r/dvE0LGxdbVpTHl0UYcGPHRRW75YsYuXZtkdSYEU8k93\nfZCUwv6jWdw90LraNKY8u7N/S5L3HuHf326kZUwkl3ay1o8DIaSPFLJz85j04xYSm9ehRwvratOY\n8kxE+Mc159C9eR0e+HA5q1LtjqRACOmk8PnyXew8dNyOEoypIKpEhPPqjd2pV6MKt09Zwl67I8nv\nQjYp5OUpr8xJpl3DmgxoW/gDbcaY8iU6sgqTb0ok80QOo6ckcTzL7kjypyKTgohkisjhwv7KKsiC\nfLN2D5vTj3LngJbWNooxFUy7hjV5YXhXVu3M4MGPVtgdSX5U5IVmVY0CEJEngN3AVECAkYBnV3lU\nlQlzNhNXz7raNKaiurB9fR4ZksBTX66ndWwk913QxuuQyoWSnj66QlUnqGqm2zTFKzgtnnpifvJ+\nVqZmcEf/ltYZhzEV2B394rmmWxOe/24T/1m5y+twyoWSJoWjIjJSRMJFJExERgJHAxlYUcbPTqZ+\nzSpc3a2gRleNMRWFiPB/V3cksXkdHvhgBStSDnkdUsgraVK4HrgW2Ov+/d4dV+aOZeWyYMt+RveN\np0qEdbVpTEVXJSKciTd2JzqyCqOnJLEnw+5IOhslSgqquk1Vh6pqtKrGqOqVqrotwLEVKD3zJLWr\nV2JEj2ZeVG+MCULRkVV4/eZEjp60O5LOVomSgoi0EZHvRWS1O9xJRP4S2NAKdvhENjefH0cN62rT\nGOMjoUFNXhzRldW7Mnjgw+XkWT/PZ6Skp49eA/4EZAOo6kqcprDLXJgIN58f50XVxpggN7hdff50\ncQIzV+3h+e83eR1OSCrpz+3qqro43/MAOQGIp1jN61WndnXratMYU7DRfePZtPcIL36/iVaxkVzR\nuZHXIYWUkh4p7BORloACiMgwnOcWylyknTYyxhRBRPj7VR3pEVeXhz5cwXK7I6lUSpoU7gZeBRJE\nZCdwHzC26FmMMcYbVSLCeeWGbsREOXck7c447nVIIaOkSWG7ql4AxAAJqtpHVbcHMC5jjDkr9SKr\n8PpN53I8K5fRU5I4luXJGe+QU9KksFVEJgG9gCMlXbiIDBGRDSKSLCKPFlHuXBHJcU9LGWOMX7Rt\nEMVLI7qydtdhHvhghd2RVAIlTQoJwHc4p5G2isjLItKnqBlEJBwYD1wMtAdGiEj7Qsr9E/imNIEb\nY0xJDEyI5c+XtOPL1Xt47ruNXocT9Er68NoxVf1AVa8GugI1gR+Kma0HkKyqW1Q1C5hOwe0l3QN8\nDKSVPGxjjCm52/q04LrEprw0K5nPl+/0OpygVuL+FESkv4hMAJYCVXGavShKYyDFZzjVHee7zMbA\nVcArxdQ9RkSSRCQpPT29pCEbYwzg3JH0xJUd6dGiLg99tJKfdxz0OqSgVdInmrfh3HE0FzhHVa9V\n1Y/9UP/zwCOqmldUIVWdpKqJqpoYE2Md6hhjSq9yRBgTb+hOg5pVGT1lKbsO2R1JBSnpkUInVb1K\nVaepaklbR90JNPUZbuKO85UITHeTzjBggohcWcLlG2NMqdStUZnXb0rkZHYut79tdyQVpLie1x52\nXz4pIi/m/ytm2UuA1iLSQkQq4zSLMcO3gKq2UNU4VY0DPgLuUtXPzmxVjDGmeK3rR/Hi9V1Zv+cw\nf3zf2kjKr7gjhXXu/yScawn5/wqlqjnAOOBrdzkfqOoaERkrIvbgmzHGMwPbxvL/Lm3P12v28q9v\nN3gdTlAprjvOL9yXq1R1WWkXrqozgZn5xk0spOzNpV2+McacqVt7x5Gclsn42ZtpHRvFlV2t0y4o\n+TWFf4nIOhF5QkQ6BjQiY4wpAyLC/1zRkV7xdXn445Us3W53JEHJn1MYCAwE0oFXRWSVV/0pGGOM\nv1SOCOOVkd1pWKsqd0xNYqfdkVTy5xRUdY+qvojTEN5y4LGARWWMMWWkzuk7kvK4/e0kjp6s2Hck\nlfQ5hXYi8riIrAJeAn7CucXUGGNCXqvYKF4e2Y0New5zXwW/I6mkRwpvAAeBi1R1gKq+oqrWLIUx\nptzo3yaGv17Wnm/X7uWZbyruHUnF9ljjNli3VVVfKIN4jDHGMzefH8emtCO8MmczrWMjubpbxTsh\nUuyRgqrmAk3dB9CMMabccu5I6sD5Levx6MerWLr9gNchlbkS96cAzBeRv4rI/af+AhmYMcZ4oVJ4\nGBNGdqNR7aqMmbKU1IPHvA6pTJU0KWwG/uOWj/L5M8aYcqd29cpMvulcsnKdO5KOVKA7kkQ1tK6y\nJyYmalJSktdhGGMqgLmb0rn5zSUMbBvLqzd2JzxMvA7pjInIUlVNLK5cSW9JnS0is/L/nX2YxhgT\nvPq2juFvl7fnu3V7efrr9V6HUyaKvfvI9aDP66rANUDFOZ4yxlRYo86LY+PeTF79YQutY6MY1r18\n35FUoqSgqvlbRJ0vIosDEI8xxgSdv13ega37jvKnT1bSvF51zo2r63VIAVPS00d1ff6iRWQIUCvA\nsRljTFCoFB7GhOu707ROde6YupSUA+X3jqSS3n20FKdPhSScJi7uB24LVFDGGBNsalWvxOSbEskp\n53ckFdfz2rki0sDtIS0e+B9gvfu3tiwCNMaYYBEfE8mEkd1JTj/CH6b9TG45bCOpuCOFV4EsABHp\nB/wDeBvIACYVt3ARGSIiG0QkWUQeLWD6UBFZKSLLRSRJRPqUfhWMMabs9GkdzeNXdOD79Wn886vy\nd0dScReaw1X11HPe1wGTVPVj4GMRWV7UjG6bSeOBC4FUYImIzFBV3yOM74EZqqoi0gn4AEg4kxUx\nxpiycmOv5mzam8mkH7fQKjaSaxObeh2S3xR3pBAuIqcSx2DA99mE4hJKDyBZVbeoahYwHRjqW0BV\nj+gvT8/VAMrfsZgxplx67LL29GkVzWOfr2bv4RNeh+M3xSWFacAPIvI5cByYCyAirXBOIRWlMZDi\nM5zqjvsVEblKRNYD/wVuLWhBIjLGPb2UlJ6eXky1xhgTeBHhYfzfVeeQm6c8/90mr8PxmyKTgqo+\nCTwAvAX08flVHwbc448AVPVTVU0ArgSeKKTMJFVNVNXEmJgYf1RrjDFnrVm96ozs2ZwPklJITjvi\ndTh+UZKmsxe6O+6jPuM2quqyYmbdCfieaGvijiusnh+BeBGJLi4mY4wJFvcMakW1SuE8+3X56Jin\nxH00n4ElQGsRaeH2xTAcmOFbQERaiYi4r7sBVYD9AYzJGGP8ql5kFcb0i+erNXtYtuOg1+GctYAl\nBVXNAcYBXwPrgA9UdY2IjBWRsW6xa4DV7p1M44HrNNSabTXGVHi39WlBdGQVnpq5nlDfhVnT2cYY\n4wdTF27nr5+t5vWbEhncrr7X4fyGX5vONsYYU7Th5zalRXQN/vnV+pB+0tmSgjHG+EGl8DAeuqgt\nG/ce4ZNlqV6Hc8YsKRhjjJ9c3LEBnZvW5t/fbuREdq7X4ZwRSwrGGOMnIsKjQxLYnXGCKQu2eR3O\nGbGkYIwxfnRey3oMaBvD+NmbyTiW7XU4pWZJwRhj/OyRIQkcPpHNhB+SvQ6l1CwpGGOMn7VrWJOr\nujbmzfnb2HXouNfhlIolBWOMCYD7L2wDCs9/t9HrUErFkoIxxgRAkzrVGXVecz5amsrGvZleh1Ni\nlhSMMSZA7h7YihqVI3g6hHpos6RgjDEBUqdGZcYOaMl369JYvPVA8TMEAUsKxhgTQLf2bkH9mlV4\n6st1IdFYniUFY4wJoGqVw/njBW1YtuMQ36zd63U4xbKkYIwxATasexNaxtTg6a/Wk5Ob53U4RbKk\nYIwxARYRHsbDQxLYnH6UD5cGd2N5lhSMMaYM/K59fbo3r8Nz327keFbwNpZnScEYY8qAiPDoxQmk\nZZ7kjflbvQ6nUAFNCiIyREQ2iEiyiDxawPSRIrJSRFaJyE8i0jmQ8RhjjJfOjavLBe3qM3HOZg4e\nzfI6nAIFLCmISDhOv8sXA+2BESLSPl+xrUB/VT0HeAKYFKh4jDEmGDw8pC1Hs3J4eXZwNpYXyCOF\nHkCyqm5R1SxgOjDUt4Cq/qSqB93BhUCTAMZjjDGea1M/imHdmzB1wXZSDhzzOpzfCGRSaAyk+Ayn\nuuMKcxvwZUETRGSMiCSJSFJ6erofQzTGmLL3xwvbIALPfRt8jeUFxYVmERmIkxQeKWi6qk5S1URV\nTYyJiSnb4Iwxxs8a1qrGLb1b8OnynazdddjrcH4lkElhJ9DUZ7iJO+5XRKQTMBkYqqr7AxiPMcYE\njTv7t6Rm1Uo8/XVwNZYXyKSwBGgtIi1EpDIwHJjhW0BEmgGfADeqavAdRxljTIDUql6Juwe2ZM6G\ndH7avM/rcE4LWFJQ1RxgHPA1sA74QFXXiMhYERnrFnsMqAdMEJHlIpIUqHiMMSbYjDovjka1qvLU\nl+uDprE8CZZASioxMVGTkix3GGPKhw+TUnjoo5WMv74bl3ZqGLB6RGSpqiYWVy4oLjQbY0xFdXW3\nJrStH8UzX68nOwgay7OkYIwxHgoPEx65uC3b9h9j+pKU4mcIMEsKxhjjsYFtY+nRoi4vfLeJoydz\nPI3FkoIxxnjsVGN5+46cZPJcbxvLs6RgjDFBoFuzOgzp0IBJP25m35GTnsVhScEYY4LEQ0PaciIn\nj5dneddYniUFY4wJEi1jIrnu3Ka8u2g72/cf9SQGSwrGGBNE7hvcmoiwMJ79xptGHiwpGGNMEImt\nWZXb+rTgixW7WJWaUeb1W1IwxpggM6Z/PHWqV+KfX5V9Y3mWFIwxJsjUrFqJewa1Zl7yPuZuKts+\nZCwpGGNMEBrZqxlN6lTjqS/Xk5dXdm3UWVIwxpggVCUinAd/15Y1uw7zxcpdZVavJQVjjAlSV3Ru\nRLuGNXnm6w2czMktkzotKRhjTJAKC3Oav0g9eJz3Fu0omzrLpBZjjDFnpF/raHq3qsdLs5LJPJEd\n8PoCmhREZIiIbBCRZBF5tIDpCSKyQEROisiDgYzFGGNCkYjwyJAEDhzNYtKPWwJeX8CSgoiEA+OB\ni4H2wAgRaZ+v2AHgXuDZQMVhjDGhrlOT2lzWqSGT524l7fCJgNYVyCOFHkCyqm5R1SxgOjDUt4Cq\npqnqEiDwx0TGGBPCHvxdW7Jz83jh+00BrSeQSaEx4NuNUKo7rtREZIyIJIlIUnp62T7IYYwxwSAu\nugbX92zG9CUpbEk/ErB6QuJCs6pOUtVEVU2MiYnxOhxjjPHEvYNbUzUijGe/2RCwOgKZFHYCTX2G\nm7jjjDHGnIHoyCqM7hfPzFV7+HnHwYDUEciksARoLSItRKQyMByYEcD6jDGm3Lu9bzzRkZX5x5fr\nUfV/8xcBSwqqmgOMA74G1gEfqOoaERkrImMBRKSBiKQC9wN/EZFUEakZqJiMMSbURVaJ4N7BrVm8\n9QBzNvj/GqsEItMEUmJioiYlJXkdhjHGeCYrJ48Ln/uBapXC+e+9fQkPk2LnEZGlqppYXLmQuNBs\njDHmF5Ujwnjooras35PJZz/791KtJQVjjAlBl3RsSKcmtfj3txs5ke2/xvIsKRhjTAgKCxMeHZLA\nzkPHmbpgu/+W67clGWOMKVPnt4qmX5sYXp6dTMZx/zQMYUnBGGNC2CND2nL4RDYTf9jsl+VZUjDG\nmBDWoVEtruzSmDfmbWV3xvGzXp4lBWOMCXH3X9gGVXj+27NvLM+SgjHGhLimdatzQ6/mfLg0hU17\nM89qWZYUjDGmHBg3qBXVK0fw9Ndn11ieJQVjjCkH6taozNj+8Xy7di9J2w6c8XIsKRhjTDlxa58W\nxEZVOavG8iwpGGNMOVG9cgT3XdCGpdsP8u3avWe0DEsKxhhTjlyb2IT46Bo8/fUGcnLzSj2/JQVj\njClHIsLDeHhIW5LTjvDxstRSz29JwRhjypmLOjSgS9PaPPftJo5nla6xPEsKxhhTzogIf7o4gT2H\nT/DmT1tLNa8lBWOMKYd6xtdjcEIsr8zZzMGjWSWeL6BJQUSGiMgGEUkWkUcLmC4i8qI7faWIdAtk\nPMYYU5E8PCSBIydzmDAnucTzBCwpiEg4MB64GGgPjBCR9vmKXQy0dv/GAK8EKh5jjKlo2jaI4ppu\nTXj7p5L3txDII4UeQLKqblHVLGA6MDRfmaHAFHUsBGqLSMMAxmSMMRXKHy9sA8V34XxaIJNCYyDF\nZzjVHVfaMojIGBFJEpGk9PR0vwdqjDHlVePa1bizf8sSlw+JC82qOklVE1U1MSYmxutwjDEmpPzx\nwjYlLhvIpLATaOoz3MQdV9oyxhhjykggk8ISoLWItBCRysBwYEa+MjOAUe5dSL2ADFXdHcCYjDHG\nFCEiUAtW1RwRGQd8DYQDb6jqGhEZ606fCMwELgGSgWPALYGKxxhjTPEClhQAVHUmzo7fd9xEn9cK\n3B3IGIwxxpRcSFxoNsYYUzYsKRhjjDnNkoIxxpjTLCkYY4w5Tc60H0+viEgmsMHrOEogGtjndRAl\nYHH6VyjEGQoxgsXpb21VNaq4QgG9+yhANqhqotdBFEdEkixO/7E4/ScUYgSL099EJKkk5ez0kTHG\nmNMsKRhjjDktFJPCJK8DKCGL078sTv8JhRjB4vS3EsUZcheajTHGBE4oHikYY4wJEEsKxhhjTgup\npCAiV4qIikiC17EURkRyRWS5iKwQkWUicr7XMRVERBqIyHQR2SwiS0VkpoiUvCeOMuCzLde42/MB\nEQm6z6xPnKf+HvU6psIUEGuc1zHlJyL1ReQ9EdnifjYXiMhVXsfly90PveMzHCEi6SLyHy/jyk9E\n6vm813tEZKfPcOWC5gm15xRGAPPc/3/zOJbCHFfVLgAichHwD6C/tyH9mogI8CnwtqoOd8d1BuoD\nG72MLR/fbRkLvAfUJPje+9NxhoCgjtX9bH6G89m83h3XHLjC08B+6yjQUUSqqepx4EKCsIMwVd0P\nnPoOPQ4cUdVni5on6H51FUZEIoE+wG04HfaEgprAQa+DKMBAIDtfM+YrVHWuhzEVSVXTgDHAOHfH\nYcqnQUBWvs/mdlV9ycOYCjMTuNR9PQKY5mEsfhMySQEYCnylqhuB/SLS3euAClHNPTRbD0wGnvA6\noAJ0BJZ6HURpqeoWnA6bYr2OJZ9q8utTMtd5HVARfGP91OtgCtABWOZ1ECU0HRguIlWBTsAij+Px\ni1A6fTQCeMF9Pd0dDsYdm+8pj/OAKSLSUe3e3/IsqE/J5BNKsSIi43HOEGSp6rlex+NLVVe612RG\nkK8zsVAWEklBROriHFaeIyKK82tRReShYN7ZquoCEYkGYoA0r+PxsQYY5nUQpSUi8UAuwbUtjX+t\nAa45NaCqd7vfoRK12+OBGcCzwACgnreh+EeonD4aBkxV1eaqGqeqTYGtQF+P4yqSe5dUOLDf61jy\nmQVUEZExp0aISCcRCdrtKSIxwETg5WD+IWDO2iygqojc6TOuulfBlMAbwP+o6iqvA/GXkDhSwDk8\n+2e+cR+7438s+3CKVE1ElruvBbhJVXO9DCg/VVX3Fr/nReQR4ASwDbjP08B+69S2rATkAFOBf3sb\nUoF833Nwrn0F7W2pwcz9bF4JPCciDwPpOHf6POJtZAVT1VTgRa/j8Cdr5sIYY8xpoXL6yBhjTBmw\npGCMMeY0SwrGGGNOs6RgjDHmNEsKxhhjTrOkYEwhStOSrIjEicjqQqZNFpH2gY3WGP8IlecUjClT\n/mxJVlVv93+ExgSGHSkYU7ACW5IFfhaR792+MlaJyFCfeSJE5F0RWSciH4lIdQARmSMiie7rIyLy\npNs/xEIRqV+ma2VMMSwpGFOwwlqSPQFcpardcBLHv3ya8m4LTFDVdsBh4K4C5q8BLFTVzjhP44/2\ne+TGnAVLCsaUjgD/JyIrge+AxjinlABSVHW++/odnNY988sCTvXOtRSIC1yoxpSeJQVjCrYGKKjP\njpE4rd52d5ug3gtUdaflbzOmoDZksn0a9MvFruuZIGNJwZiCFdiSLNAcSFPVbBEZ6A6f0sztQwPg\nepyuY40JKZYUjCmA+2v+KuAC95bUNTj9bc8EEkVkFTAKWO8z2wbgbhFZB9QBXinjsI05a9ZKqjHG\nmNPsSMEYY8xplhSMMcacZknBGGPMaZYUjDHGnGZJwRhjzGmWFIwxxpxmScEYY8xp/x+KMNHhJfml\nDQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x5af76905f8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the original variable\n",
    "\n",
    "fig = plt.figure()\n",
    "fig = X_train.groupby(['Cabin'])['Survived'].mean().plot()\n",
    "fig.set_title('Normal relationship between variable and target')\n",
    "fig.set_ylabel('Survived')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x5af8b510b8>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEXCAYAAABCjVgAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npCAiInFKCiIiEvf/AR8/y177Kts9AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x5af86ed320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the transformed result: the monotonic variable\n",
    "\n",
    "fig = plt.figure()\n",
    "fig = X_train.groupby(['Cabin_ordered'])['Survived'].mean().plot()\n",
    "fig.set_title('Monotonic relationship between variable and target')\n",
    "fig.set_ylabel('Survived')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "As you can see in the above plot, there is now a monotonic relationship between the variable Cabin and probability of survival. The higher the Cabin number, the more likely the person was to survive.\n",
    "\n",
    "### Note\n",
    "\n",
    "Monotonic does not mean strictly linear. Monotonic means that it increases constantly, or it decreases constantly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  },
  "toc": {
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": "block",
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}