ADAMS_ML_Notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Credit Scoring \n",
    "\n",
    "\n",
    "For this task, we will load the data set we used in the EDA session. Recall that this data set was provided by www.creditriskanalytics.com. As a quick reminder: \n",
    "\n",
    "## About the data set\n",
    "\n",
    "The data set consists of 5960 observations and 13 variables. \n",
    "The variables are defined as the following:\n",
    "\n",
    "- BAD: 1=default; 0=non-default\n",
    "- LOAN: amount of the loan request\n",
    "- MORTDUE: amount due on existing mortgage\n",
    "- VALUE: value of current property\n",
    "- REASON: DebtCon=debt consolidation; HomeImp=home improvement\n",
    "- JOB: occupational categories\n",
    "- YOJ: years at present job\n",
    "- DEROG: number of major derogatory reports\n",
    "- DELINQ: number of delinquent credit lines\n",
    "- CLAGE: age of oldest credit line in months\n",
    "- NINQ: number of recent credit inquiries\n",
    "- CLNO: number of credit lines\n",
    "- DEBTINC: debt-to-income ratio\n",
    "\n",
    "<br> \n",
    "This notebook will elaborate on the process of different machine learning techniques and their evaluation, including logistic regression, random forest and gradient boosting.\n",
    "\n",
    "Before applying any algorithm, we will import the necessary packages and load the data. After, we will quickly display some general information about the data set to help us recall. If you are not familiar with this data set, please look at the explanatory data analysis notebook to ensure you understand the data before interpreting the results later. There you can also see the original values of all numeric variables before standardization.\n",
    "\n",
    "An important finding is the imbalance of the data set, which contains only about 20% of 'BAD' observations. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 310,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.rcParams['figure.figsize'] = 5,5\n",
    "\n",
    "#Import packages\n",
    "import importlib\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import statsmodels.api as stat\n",
    "import xgboost as xgb\n",
    "from sklearn import metrics\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import train_test_split, StratifiedKFold, GridSearchCV\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 311,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>BAD</th>\n",
       "      <th>LOAN</th>\n",
       "      <th>MORTDUE</th>\n",
       "      <th>VALUE</th>\n",
       "      <th>YOJ</th>\n",
       "      <th>DEROG</th>\n",
       "      <th>DELINQ</th>\n",
       "      <th>CLAGE</th>\n",
       "      <th>NINQ</th>\n",
       "      <th>CLNO</th>\n",
       "      <th>DEBTINC</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5960.000000</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "      <td>5.960000e+03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.199497</td>\n",
       "      <td>-9.123202e-16</td>\n",
       "      <td>6.588596e-16</td>\n",
       "      <td>-8.776630e-16</td>\n",
       "      <td>1.417397e-16</td>\n",
       "      <td>3.113794e-16</td>\n",
       "      <td>-9.155987e-16</td>\n",
       "      <td>-1.203363e-16</td>\n",
       "      <td>-1.761455e-16</td>\n",
       "      <td>1.181139e-15</td>\n",
       "      <td>1.265040e-15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.399656</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "      <td>1.000084e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>-1.562299e+00</td>\n",
       "      <td>-1.424276e+00</td>\n",
       "      <td>-1.707224e+00</td>\n",
       "      <td>-1.168611e+00</td>\n",
       "      <td>-1.239592e+00</td>\n",
       "      <td>-1.133918e+00</td>\n",
       "      <td>-1.850350e+00</td>\n",
       "      <td>-1.134248e+00</td>\n",
       "      <td>-1.986375e+00</td>\n",
       "      <td>-1.696035e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>-6.699632e-01</td>\n",
       "      <td>-5.995930e-01</td>\n",
       "      <td>-5.960053e-01</td>\n",
       "      <td>-7.818625e-01</td>\n",
       "      <td>-1.183341e-01</td>\n",
       "      <td>-2.672662e-01</td>\n",
       "      <td>-6.773687e-01</td>\n",
       "      <td>-5.668382e-01</td>\n",
       "      <td>-5.983115e-01</td>\n",
       "      <td>-3.744374e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>-2.059485e-01</td>\n",
       "      <td>-1.186451e-01</td>\n",
       "      <td>-1.975027e-01</td>\n",
       "      <td>-2.661978e-01</td>\n",
       "      <td>-1.183341e-01</td>\n",
       "      <td>-2.672662e-01</td>\n",
       "      <td>-4.046820e-02</td>\n",
       "      <td>-5.668382e-01</td>\n",
       "      <td>-4.308588e-02</td>\n",
       "      <td>3.429367e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.186868e-01</td>\n",
       "      <td>4.409223e-01</td>\n",
       "      <td>3.272283e-01</td>\n",
       "      <td>5.072992e-01</td>\n",
       "      <td>-1.183341e-01</td>\n",
       "      <td>-2.672662e-01</td>\n",
       "      <td>6.122157e-01</td>\n",
       "      <td>5.679806e-01</td>\n",
       "      <td>5.121397e-01</td>\n",
       "      <td>7.161369e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>6.361645e+00</td>\n",
       "      <td>7.025058e+00</td>\n",
       "      <td>1.292491e+01</td>\n",
       "      <td>4.245868e+00</td>\n",
       "      <td>1.109425e+01</td>\n",
       "      <td>1.273251e+01</td>\n",
       "      <td>1.077030e+01</td>\n",
       "      <td>9.079122e+00</td>\n",
       "      <td>4.676332e+00</td>\n",
       "      <td>1.095704e+01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               BAD          LOAN       MORTDUE         VALUE           YOJ  \\\n",
       "count  5960.000000  5.960000e+03  5.960000e+03  5.960000e+03  5.960000e+03   \n",
       "mean      0.199497 -9.123202e-16  6.588596e-16 -8.776630e-16  1.417397e-16   \n",
       "std       0.399656  1.000084e+00  1.000084e+00  1.000084e+00  1.000084e+00   \n",
       "min       0.000000 -1.562299e+00 -1.424276e+00 -1.707224e+00 -1.168611e+00   \n",
       "25%       0.000000 -6.699632e-01 -5.995930e-01 -5.960053e-01 -7.818625e-01   \n",
       "50%       0.000000 -2.059485e-01 -1.186451e-01 -1.975027e-01 -2.661978e-01   \n",
       "75%       0.000000  4.186868e-01  4.409223e-01  3.272283e-01  5.072992e-01   \n",
       "max       1.000000  6.361645e+00  7.025058e+00  1.292491e+01  4.245868e+00   \n",
       "\n",
       "              DEROG        DELINQ         CLAGE          NINQ          CLNO  \\\n",
       "count  5.960000e+03  5.960000e+03  5.960000e+03  5.960000e+03  5.960000e+03   \n",
       "mean   3.113794e-16 -9.155987e-16 -1.203363e-16 -1.761455e-16  1.181139e-15   \n",
       "std    1.000084e+00  1.000084e+00  1.000084e+00  1.000084e+00  1.000084e+00   \n",
       "min   -1.239592e+00 -1.133918e+00 -1.850350e+00 -1.134248e+00 -1.986375e+00   \n",
       "25%   -1.183341e-01 -2.672662e-01 -6.773687e-01 -5.668382e-01 -5.983115e-01   \n",
       "50%   -1.183341e-01 -2.672662e-01 -4.046820e-02 -5.668382e-01 -4.308588e-02   \n",
       "75%   -1.183341e-01 -2.672662e-01  6.122157e-01  5.679806e-01  5.121397e-01   \n",
       "max    1.109425e+01  1.273251e+01  1.077030e+01  9.079122e+00  4.676332e+00   \n",
       "\n",
       "            DEBTINC  \n",
       "count  5.960000e+03  \n",
       "mean   1.265040e-15  \n",
       "std    1.000084e+00  \n",
       "min   -1.696035e+00  \n",
       "25%   -3.744374e-01  \n",
       "50%    3.429367e-01  \n",
       "75%    7.161369e-01  \n",
       "max    1.095704e+01  "
      ]
     },
     "execution_count": 311,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load data\n",
    "credit = pd.read_csv('/Users/wiebkeberlin/Desktop/creditnew.csv')\n",
    "credit.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Problem to solve \n",
    "\n",
    "\n",
    "This data set stages a common problem for banks and lenders. It is important, that banks can ensure which people are likely to pay back a loan and which are likely to default, as obviously wrong decisions could lead to high losses. This data set contains information about the customers as well as the outcome variable 'BAD'. This allows us to find generalizable patterns in these two groups ('GOOD' and 'BAD') through supervised machine learning.\n",
    "\n",
    "\n",
    "# Let's get started\n",
    "\n",
    "\n",
    "Before we apply any algorithms to our data, we need to ensure that it is in the right format. Not every algorithm can handle categorical variables. The way to overcome this obstacle is by using dummy variables.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 312,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 5960 entries, 0 to 5959\n",
      "Data columns (total 13 columns):\n",
      " #   Column   Non-Null Count  Dtype  \n",
      "---  ------   --------------  -----  \n",
      " 0   BAD      5960 non-null   int64  \n",
      " 1   REASON   5960 non-null   object \n",
      " 2   JOB      5960 non-null   object \n",
      " 3   LOAN     5960 non-null   float64\n",
      " 4   MORTDUE  5960 non-null   float64\n",
      " 5   VALUE    5960 non-null   float64\n",
      " 6   YOJ      5960 non-null   float64\n",
      " 7   DEROG    5960 non-null   float64\n",
      " 8   DELINQ   5960 non-null   float64\n",
      " 9   CLAGE    5960 non-null   float64\n",
      " 10  NINQ     5960 non-null   float64\n",
      " 11  CLNO     5960 non-null   float64\n",
      " 12  DEBTINC  5960 non-null   float64\n",
      "dtypes: float64(10), int64(1), object(2)\n",
      "memory usage: 605.4+ KB\n"
     ]
    }
   ],
   "source": [
    "credit.info()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dummy Variables: What are they and why do we use them?\n",
    "\n",
    "A dummy variables is a binary vectors of a specific category of a variable. For example, a vector for the variable 'JOB' and category 'Office'. In this vector the value would  a 1 for every observation with an office job, and 0 for all others. \n",
    "\n",
    "We require dummy encoding as categorical data is common and carries valuable information, however, are not able to be interpreted by all algorithms.\n",
    "\n",
    "This can be clarified through the example of a logistic regression. Regression analysis is common practice for using a continuous variable (e.g. 'LOAN') to predict the independent variable ('BAD' in this example). As you cannot rank and/or calculate a meaningful distance between categories, this interpretation is not possible. However, if we split up these variable into the earlier described dummy variables, we create a numeric variable by only using the number 0 and 1. The new variables will then be used as input for the categorical variable. In our case, we would have 6 different dummy variables for the variable 'JOB' as it had 6 categories. \n",
    "\n",
    "(\n",
    "However, regression settings also have problems with multicollinearity. This means, if we incorporate all six new dummy variables, we will encounter a problem. If we would, the sum of the 6 vectors would result in the value 1 for every observation. If this is to abstract for you, let's look at this by using an example. One observation's (all other variables excluded) dummy variables for the category 'JOB'.\n",
    "\n",
    "|  JOB_Mgr \t| JOB_Office  \t| JOB_Other  \t| JOB_ProfExe  \t| JOB_Sales  \t| JOB_Self  \t|\n",
    "|---\t|---\t|---\t|---\t|---\t|---\t|\n",
    "|   0\t|   0\t|   0\t|   0\t|   0\t|   1\t|\n",
    " \n",
    "    \n",
    "However, the value 0 for all other categories ('Mgr', 'Office', 'Other', 'ProxExe' and 'Sales') already indicates that the job must be the category 'Self'. Therefore, one vector is redundant. And that is what multicollinearity is about. Multicollinearity is the high correlations among independent variables in a regression setting. Therefore, we can escape the so called dummy trap by dropping one dummy variable for each category. \n",
    "\n",
    "\n",
    "So, firstly, we will create the dummy variables.\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {},
   "outputs": [],
   "source": [
    "credittemp = pd.get_dummies(credit, drop_first= True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 314,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>REASON_HomeImp</th>\n",
       "      <th>JOB_Office</th>\n",
       "      <th>JOB_Other</th>\n",
       "      <th>JOB_ProfExe</th>\n",
       "      <th>JOB_Sales</th>\n",
       "      <th>JOB_Self</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5960.000000</td>\n",
       "      <td>5960.000000</td>\n",
       "      <td>5960.000000</td>\n",
       "      <td>5960.000000</td>\n",
       "      <td>5960.000000</td>\n",
       "      <td>5960.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.298658</td>\n",
       "      <td>0.159060</td>\n",
       "      <td>0.447483</td>\n",
       "      <td>0.214094</td>\n",
       "      <td>0.018289</td>\n",
       "      <td>0.032383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.457708</td>\n",
       "      <td>0.365763</td>\n",
       "      <td>0.497276</td>\n",
       "      <td>0.410227</td>\n",
       "      <td>0.134004</td>\n",
       "      <td>0.177029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       REASON_HomeImp   JOB_Office    JOB_Other  JOB_ProfExe    JOB_Sales  \\\n",
       "count     5960.000000  5960.000000  5960.000000  5960.000000  5960.000000   \n",
       "mean         0.298658     0.159060     0.447483     0.214094     0.018289   \n",
       "std          0.457708     0.365763     0.497276     0.410227     0.134004   \n",
       "min          0.000000     0.000000     0.000000     0.000000     0.000000   \n",
       "25%          0.000000     0.000000     0.000000     0.000000     0.000000   \n",
       "50%          0.000000     0.000000     0.000000     0.000000     0.000000   \n",
       "75%          1.000000     0.000000     1.000000     0.000000     0.000000   \n",
       "max          1.000000     1.000000     1.000000     1.000000     1.000000   \n",
       "\n",
       "          JOB_Self  \n",
       "count  5960.000000  \n",
       "mean      0.032383  \n",
       "std       0.177029  \n",
       "min       0.000000  \n",
       "25%       0.000000  \n",
       "50%       0.000000  \n",
       "75%       0.000000  \n",
       "max       1.000000  "
      ]
     },
     "execution_count": 314,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "credittemp.select_dtypes(include='uint8').describe()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Have a closer look at this table. What information can we draw from this table? What does the mean indicate?\n",
    "\n",
    "\n",
    "<br> <br> <br>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 315,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 5960 entries, 0 to 5959\n",
      "Data columns (total 17 columns):\n",
      " #   Column          Non-Null Count  Dtype  \n",
      "---  ------          --------------  -----  \n",
      " 0   BAD             5960 non-null   float64\n",
      " 1   LOAN            5960 non-null   float64\n",
      " 2   MORTDUE         5960 non-null   float64\n",
      " 3   VALUE           5960 non-null   float64\n",
      " 4   YOJ             5960 non-null   float64\n",
      " 5   DEROG           5960 non-null   float64\n",
      " 6   DELINQ          5960 non-null   float64\n",
      " 7   CLAGE           5960 non-null   float64\n",
      " 8   NINQ            5960 non-null   float64\n",
      " 9   CLNO            5960 non-null   float64\n",
      " 10  DEBTINC         5960 non-null   float64\n",
      " 11  REASON_HomeImp  5960 non-null   float64\n",
      " 12  JOB_Office      5960 non-null   float64\n",
      " 13  JOB_Other       5960 non-null   float64\n",
      " 14  JOB_ProfExe     5960 non-null   float64\n",
      " 15  JOB_Sales       5960 non-null   float64\n",
      " 16  JOB_Self        5960 non-null   float64\n",
      "dtypes: float64(17)\n",
      "memory usage: 791.7 KB\n"
     ]
    }
   ],
   "source": [
    "# allow input to all algorithms\n",
    "credit = credittemp.astype('float64')\n",
    "credit.info()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Simulation of model application\n",
    "\n",
    "One of the core tasks in model building is the robust model simulation of our model. What does this mean?\n",
    "\n",
    "We build a model on our data. But how do we ensure, that our model finds generalizable patterns in the data opposed to just describing this particular sample. And there are several ways you can do this.\n",
    "\n",
    "Whilst this will become more clear when evaluating the different models,the data needs to be handled beforehand.\n",
    "\n",
    "\n",
    "### From the resubstitution estimate to the split-sample method\n",
    "\n",
    "The most simple way is building a model on the whole data set and using the same data to make predictions. These predictions will be compared to the actuals (target variable column) to assess the model performance. The resubstitution estimate however often leads to the problem mentioned above. \n",
    "\n",
    "One of the most common methods is the split-sample method.\n",
    "The method can by implemented by splitting the data into a minimum of two sets: train and test set. This means, we train a model on a random subset of our data set, the training set. Then we use the remaining instances, the test set, also called hold out data, for model assessment. In order to evaluate the performance, we compute predictions of the test set with the model built on the training set. We compare these predictions to the actuals in the test set. \n",
    "\n",
    "\n",
    "The meta-parameter of the split-sample method is the size of the training and test set\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 316,
   "metadata": {},
   "outputs": [],
   "source": [
    "#data partitioning\n",
    "#split outcome variable and the rest of the variables into two sets\n",
    "y= credit.BAD\n",
    "x = credit.drop('BAD', axis=1) #specify axis when selecting variables\n",
    "\n",
    "# create training and test set\n",
    "# x_train = all columns but 'BAD' of training set (needed to build supervised learning model)\n",
    "# y_train = outcome variable of training observations (needed to build supervised learning model)\n",
    "# x_test = all columns but 'BAD' of test set (needed to predict outcome of test set with training set model)\n",
    "# y_test = outcome variable of test observations (needed for model assessment)\n",
    "x_train, x_test, y_train, y_test = train_test_split(x,y, test_size=0.2 )\n",
    "\n",
    "# 20& of data in hold out data\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 317,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((4768,), (1192,), (4768, 16), (1192, 16))"
      ]
     },
     "execution_count": 317,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train.shape, y_test.shape, x_train.shape, x_test.shape\n",
    "#make yourself familiar with these vectors and what each of them represents"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Supervised Learning Algorithm\n",
    "\n",
    "\n",
    "### Logistic Regression\n",
    "\n",
    "Logistic regression or logit model is a classification technique. The outcome variable in this model is binary, 1 indicating 'BAD'.In this case, the predictions estimate a probability between 0 and 1 of the loan being defaulted on.\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 318,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
      "                   intercept_scaling=1, l1_ratio=None, max_iter=500,\n",
      "                   multi_class='auto', n_jobs=None, penalty='none',\n",
      "                   random_state=None, solver='lbfgs', tol=0.0001, verbose=0,\n",
      "                   warm_start=False)\n",
      "Accuracy on train data: 0.8727\n",
      "Accuracy on test data: 0.8591\n"
     ]
    }
   ],
   "source": [
    "sk_logit= LogisticRegression(penalty='none', fit_intercept=True, max_iter=500)\n",
    "log = sk_logit.fit(x_train, y_train)\n",
    "# todo add code to store all performance values in a dataframe\n",
    "#print('Accuracy: {0}'.format(sk_logit.score(train, y.TARGET)))\n",
    "\n",
    "print(log)\n",
    "accuracy=list()\n",
    "accuracy.append(log.score)\n",
    "\n",
    "print('Accuracy on train data: {:.4f}'.format(sk_logit.score(x_train, y_train)))\n",
    "print('Accuracy on test data: {:.4f}'.format(sk_logit.score(x_test, y_test)));\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluation methods\n",
    "\n",
    "Evaluation methods are those that help us evaluate how well our model managed to predict the given cases. A common practice is comparing the actuals to the predictions. Meaning, just as we did, we predict the outcome variable of the test set, based on each observations attributes. We then compare these results to \"y_test\", our actuals, as we started out with a labeled data set and know the outcome.\n",
    "\n",
    "\n",
    "### Accuracy and its shortcomings\n",
    "\n",
    "Accuracy in particular  is a threshold metric. It compares the model performance over one threshold. If not specified, this threshold usually is 0.5. But predictive accuracy does therefore gives insufficient information. Changing the threshold can change the accuracy a lot, eventhough the model stays the same. This is particularly true in imbalanced data sets like this. Let me give you an examplary scenario: We have 100 observations. 5 observations are of the class 'BAD' and 95 of the class 'GOOD'. So even without finding any generalizable patterns, classifying all observations as good would still result in an accuracy of 95%. To ensure robustness of our model, we want to avoid this. \n",
    "\n",
    "In our case, the accuracy is just under the accuracy of the training data. We've included both for comparison and as it can be a first indicator for overfitting (high score on training data, low on test). In our case however, it looks acceptable.\n",
    "\n",
    "\n",
    "A good way to overcome these issues is by using the ROC curve and AUC. But let's take it step by step. We are starting off with the confusion matrix.\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "### Confusion matrix\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 319,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[912  30]\n",
      " [138 112]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 1. Quick and dirty way\n",
    "cmat = metrics.confusion_matrix(y_test, sk_logit.predict(x_test))\n",
    "print(cmat)\n",
    "# 2. Pretty-print confusion matrix\n",
    "metrics.plot_confusion_matrix(sk_logit, x_test, y_test, labels=[0, 1], display_labels=['Goods(0)', 'Bads(1)'], cmap='Blues',values_format='n')\n",
    "plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What can we see on this table? Do you think, this is a good model? How do you calculate the accuracy from this table? Hint: remind yourself of how the variable BAD is distributed. \n",
    "Are both outcomes the same?\n",
    "\n",
    "A great way to overcome some of these shortcomings is creating the ROC curve and calculating the AUC. For that we will need the True Positive Rate and False Positive Rate. Again, remind yourself that for this confusion matrix we would only consider one cut-off. To create a ROC curve this is done for all cut-offs.\n",
    "The formulas as a quick reminder:\n",
    "We are working with a boolean outcome variable in which 1 indicates 'BAD'. So the TPR and FPR will refer to 'BAD' as a positive. <br> <br>\n",
    "TPR: $113: (113+131)$ | True Positive / (all actual positives)\n",
    "\n",
    "FPR: $36:(912+36)$    | False Positive/ (actual no)\n",
    "(edit later)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ROC Curve and AUC\n",
    "\n",
    "This is one of the most important evaluation metrics, specifically it is a ranking method. Have a look at the graph below. The curve tells us how well the model differntiates the two classes, in regards to their predicted probabilities.\n",
    "A ROC curve has the FPR on the X-axis and FPR on the Y-axis.\n",
    "The curve is made up out of all the different combinations of TPR and FPR at all threshold, creating a line.\n",
    "\n",
    "The AUC stand for the Area under the Curvea and is a ranking indicator. The AUC estimates the probability that a randomly\n",
    "chosen positive instance is correctly ranked higher than a randomly chosen negative (Hanley and McNeil, 1983) . We calculate this area, as our goal is to have a FPR of 0 and a TPR of 1. This would be the perfect model. This point is in the top right corner of a ROC Curve. So the closer we get to that point, the better the model and the bigger the area under it. . It is popular as it allows comparison of different kind of models.\n",
    "\n",
    "So, we can plot the curve for both the training and test set evaluation.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 320,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC on training data: 0.8412\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# simple way\n",
    "print('AUC on training data: {:.4f}'.format(roc_auc))\n",
    "metrics.plot_roc_curve(log, x_test, y_test);\n",
    "plt.title('ROC curve of logit model on test')\n",
    "plt.plot([0, 1], [0, 1], \"r--\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Can you spot any differences? Does a similar AUC indicate anything, considering we have evaluated both training and test set? Remember, evaluating using the training set is equivalent to the resubstitution estimate.\n",
    "\n",
    "To make sure you understand how to extract the necessary metrics, we will compute the ROC Curve manually."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 322,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Example how to 'manually' plot the ROC curve\n",
    "# todo change label to Train and Test set \n",
    "# todo: do aucs differ\n",
    "yhat_logit = log.predict_proba(x_test)[:, 1]\n",
    "fp_rate_logit, tp_rate_logit, _ = metrics.roc_curve(y_test, yhat_logit) #extracting FPR and TPR\n",
    "auc_test_logit = metrics.auc(fp_rate_logit, tp_rate_logit) #extract AUC value\n",
    "_test_logit = metrics.auc(fp_rate, tp_rate) #extract AUC value\n",
    "\n",
    "\n",
    "plt.title('ROC curve of logit model on Training and Test Set')\n",
    "plt.plot(fp_rate_logit, tp_rate_logit, \"b\", label=\"Test Set = %0.4f\" % auc_test_logit)\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.plot([0, 1], [0, 1], \"r--\")\n",
    "plt.ylabel(\"True Positive Rate\")\n",
    "plt.xlabel(\"False Positive Rate\")\n",
    "\n",
    "\n",
    "#save data for both test set (auc_test_logit already saved above)\n",
    "yhat_logit = log.predict_proba(x_train)[:, 1]\n",
    "fp_rate_logit, tp_rate_logit, _ = metrics.roc_curve(y_train, yhat_logit) \n",
    "auc_train_logit = metrics.auc(fp_rate_logit, tp_rate_logit) \n",
    "\n",
    "\n",
    "plt.title('ROC curve of logit model on test')\n",
    "plt.plot(fp_rate_logit, tp_rate_logit, \"b\", label=\"Train Set= %0.4f\" % auc_train_logit, color='green')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.ylabel(\"True Positive Rate\")\n",
    "plt.xlabel(\"False Positive Rate\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's make sure we save our results so we can later establish which is our best model.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Cross-validation (model simulation)\n",
    "\n",
    "Cross validation is an extendend approach of the split sample method. It consists of repeating the split sample method multiple times.\n",
    "\n",
    "Step by step:\n",
    "    -randomly split data in N disjoint folds (E.g., N=5) \n",
    "    -Build model on N-1 folds and validate that model on thw remaining fold \n",
    "- Apply trained model to predict cases in the validation sample (hold-out data) \n",
    "- measure model performance \n",
    "- Repeat N times and average \n",
    "- average model performance over validation folds gives performance estimate \n",
    "- Can also estimate standard deviation (e.g., for statistical hypothesis testing)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 323,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cross-validating the logit model:\n",
      "\tCV iter: 0 with AUC=0.8548667114695341\n",
      "Average AUC on test set during cross-validation: 0.9535\n",
      "\tCV iter: 1 with AUC=0.8472502240143369\n",
      "Average AUC on test set during cross-validation: 0.9035\n",
      "\tCV iter: 2 with AUC=0.8784722222222222\n",
      "Average AUC on test set during cross-validation: 0.8870\n",
      "\tCV iter: 3 with AUC=0.8262596522415421\n",
      "Average AUC on test set during cross-validation: 0.8768\n",
      "\tCV iter: 4 with AUC=0.8770346091883784\n",
      "Average AUC on test set during cross-validation: 0.8710\n",
      "\tCV iter: 5 with AUC=0.8710627187378479\n",
      "Average AUC on test set during cross-validation: 0.8683\n",
      "\tCV iter: 6 with AUC=0.8106771846008555\n",
      "Average AUC on test set during cross-validation: 0.8656\n",
      "\tCV iter: 7 with AUC=0.7992055996889063\n",
      "Average AUC on test set during cross-validation: 0.8641\n",
      "\tCV iter: 8 with AUC=0.8679075774165473\n",
      "Average AUC on test set during cross-validation: 0.8629\n",
      "\tCV iter: 9 with AUC=0.8519329571296219\n",
      "Average AUC on test set during cross-validation: 0.8616\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x864 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#code did not work anymore (from internet)\n",
    "# todo: how to get the same thresholds for the fpr and tpr values \n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#random statehere?\n",
    "plt.figure(figsize=[12,12])\n",
    "plt.xlim([0.0, 1])\n",
    "plt.ylim([0.0, 1])\n",
    "resultscv = list()\n",
    "\n",
    "nfold =10\n",
    "cv = StratifiedKFold(n_splits=nfold)\n",
    "cv_auc_test = np.zeros(nfold)\n",
    "cv_auc_test[cv_auc_test == 0] = 'nan' #for now, to avoid 0 values to be taken into account for, help ? average\n",
    "cv_auc_train = np.zeros(nfold)\n",
    "fpr_cv = {}\n",
    "tpr_cv= {}\n",
    "thresholds = {}\n",
    "\n",
    "\n",
    "print('Cross-validating the logit model:')\n",
    "for i, (tr_idx, vl_idx) in enumerate(cv.split(x_train, y_train)):\n",
    "    xtr, xvl = x_train.iloc[tr_idx], x_train.iloc[vl_idx]\n",
    "    ytr, yvl = y_train.iloc[tr_idx], y_train.iloc[vl_idx]\n",
    "    sk_logit.fit(xtr, ytr)\n",
    "    temp= sk_logit.predict_proba(xvl)[:, 1]\n",
    "    fp_rate_cv, tp_rate_cv, threshold = metrics.roc_curve(yvl, temp.astype('float'))\n",
    "    thresholds[i] = threshold\n",
    "    fpr_cv[i] = fp_rate_cv\n",
    "    tpr_cv[i] = tp_rate_cv\n",
    "    cv_auc_test[i] = metrics.auc(fp_rate, tp_rate)\n",
    "    print('\\tCV iter: {} with AUC={}'.format(i, cv_auc[i]))\n",
    "    print('Average AUC on test set during cross-validation: {:.4f}'.format(np.nanmean(cv_auc_test)))\n",
    "    \n",
    "    # Print ROC curve\n",
    "    plt.plot(fp_rate_cv,tp_rate_cv)\n",
    "    plt.plot([0, 1], [0, 1], \"r--\")\n",
    "    \n",
    "    #save training performance for evaluation\n",
    "    temp=sk_logit.predict_proba(xtr)[:, 1]\n",
    "    fp_rate, tp_rate, _ = metrics.roc_curve(ytr, temp.astype('float'))\n",
    "    cv_auc_train[i] = metrics.auc(fp_rate, tp_rate)\n",
    "\n",
    "\n",
    "mean_fp_rate_cv = pd.DataFrame(np.vstack(fp_rate_cv))\n",
    "mean_tp_rate_cv = pd.DataFrame(np.vstack(tp_rate_cv))\n",
    "\n",
    "plt.plot(mean_fp_rate_cv, mean_tp_rate_cv, color='black')    \n",
    "plt.show()\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save results for comparison later! build average\n",
    "auc_train_cv = np.nanmean(cv_auc_train)\n",
    "auc_test_cv =np.nanmean(cv_auc_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking at the curves of all models that have been developed, it should further deepen your understanding of the importance of model simulation and that different data leads to different results. You can always get a lucky sample. but make sure your model is robust."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Regularization\n",
    "\n",
    "Regularization is a method that aims to avoid overfitting. Compley models tend to show a high variance. Model complexity and bias are closely connected (low complexity -> high bias and vice versa). Introducing bias can help prevent overfitting, as it reduced variance. This is done by adding a complexity penatly. \n",
    "\n",
    "## Regularized Logit and Lasso Penalty\n",
    "\n",
    "In a regression setting, large coefficients are indicators of complex, unstable models. Possible causes include high dimensionality and multicollinearity. The aim of the model is to minimize the magnitude the coefficients have on the model. Therefore it is included in the loss function. \n",
    "\n",
    "$$ 𝜷←min⁡ℒ(𝜷)+𝜆||(𝜷)||{_1} $$\n",
    "\n",
    "This penatly produces sparser models, as it forces the coefficients to zero. Furthermore, we also have a new metaparameter $𝜆$ (equivalent to as 'C' in the code below), which requires tuning. Therefore we use a grid search to identify its best setting. The higher $𝜆$ , the higher the penalty for complexity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 325,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Lasso test set AUC: 0.8460\n",
      "Grid-searching regularization parameter\n",
      "Lasso test set AUC with best paramter: 0.8508\n"
     ]
    }
   ],
   "source": [
    "#random seed\n",
    "# Trying to improve performance using regularized logit\n",
    "lasso = LogisticRegression(penalty='l1', C=0.1, fit_intercept=True, max_iter=200, solver='liblinear')\n",
    "lasso.fit(x_train, y_train)\n",
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_test, lasso.predict_proba(x_test)[:, 1])\n",
    "print('Lasso test set AUC: {:.4f}'.format(metrics.auc(fp_rate, tp_rate)))\n",
    "\n",
    "# Try finding a better setting for penalty parameter C\n",
    "penalty_settings = {'C': [0.001, 0.003, 0.005, 0.01, 0.03, 0.05, 0.1, 0.3, 0.5, 1, 2, 3, 3, 4, 5, 10, 20, 22, 24, 26, 28, 30, 32, 34, 36]}\n",
    "print('Grid-searching regularization parameter')\n",
    "gs_lasso = GridSearchCV(lasso, penalty_settings, scoring='roc_auc', refit=True, cv=nfold)\n",
    "gs_lasso.fit(x_train, y_train)\n",
    "# Find test set auc of the best lasso model\n",
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_test, gs_lasso.predict_proba(x_test)[:, 1])\n",
    "print('Lasso test set AUC with best paramter: {:.4f}'.format(metrics.auc(fp_rate, tp_rate)))\n",
    "\n",
    "\n",
    "#save results\n",
    "auc_test_reglog =metrics.auc(fp_rate, tp_rate)\n",
    "\n",
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_train, gs_lasso.predict_proba(x_train)[:, 1])\n",
    "                             \n",
    "auc_train_reglog =metrics.auc(fp_rate, tp_rate)                             \n",
    "                             \n",
    "                             "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 326,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'C': 0.3}"
      ]
     },
     "execution_count": 326,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Which was the best penalty?\n",
    "gs_lasso.best_params_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Both regularized regression show a slight improvement in our predictions. But can we do more?\n",
    "\n",
    "\n",
    "\n",
    "## Meta-parameter tuning\n",
    "\n",
    "Meta-paramters are those that we have to choose before building a model, just the complexity meta-parameter in the regularized logistic regression above. If you have a look at the difference, it can have an impact on our model output. There are many different meta-parameters and they are model specific. \n",
    "\n",
    "### Meta-parameter tuning through grid search\n",
    "Grid search is a grid of meta parameters. It specifies different values for each meta-parameter. By implementing this into the building of our model, we can test different settings to improve our model output. These settings depend on the data and therefore has to be done for every new model. By using the grid search, we build multiple models by testing all possible combinations of meta-parameters in the grid matrix.\n",
    "\n",
    "Whilst above we only had one meta-parameter, we will continue with an algorithm that requires more tuning.\n",
    "\n",
    "\n",
    "## Random Forest\n",
    "\n",
    "The random forest algorithm is an ensemble algorithm, with the goal to predicts the outcome variable by learning simple decision rules from the data features. The random forest algorithm is based on several models, built using bootstrap samples of the training set(random samples with replacement) and different random sample of attributes (random subspace). This means instead of using all observations and features in our data, we build a variety of models on smaller subsets of both the data AND the attributes. This allows the algorithms to pick up on more rules, as it forces models to find rules only based on the given subset of attributes. This way we might discover rules that a normal model would overshadow. This happens when one attribute has a high impact on the outcome variable.. The base models are supposed to stay simple to avoid overfitting. \n",
    "\n",
    "<br>\n",
    "\n",
    "#### Meta-parameters in our Grid search\n",
    "n_estimators: number of trees(models) in forest (ensemble)<br>\n",
    "criterion: splitting function to use, e.g. gini coefficient<br>\n",
    "max_features : maximum features in random subspace<br>\n",
    "max_depth: pruning parameter, maximum depth of decision tree<br>\n",
    "\n",
    "<br><br><br>\n",
    "\n",
    "More possible meta-parameters in a Random Forest setting:<br> <br>\n",
    "min_samples_split: minimum number of samples required in leaf node before another split is made. If it is less, this node wont split.<br>\n",
    "min_samples_leaf: minimum number of samples required to be at a leaf node.<br>\n",
    "max_leaf_nodes: maximum number of leaf nodes in a tree<br>\n",
    "n_jobs: parallelization of model building<br>\n",
    "random_state:\n",
    "This parameter is used to define the random selection.\n",
    "It is used for comparison between various models.<br><br>\n",
    "\n",
    "If meta-parameters are not specified, they will be set to their default."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 327,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuning random forest classifier\n",
      "Fitting 5 folds for each of 36 candidates, totalling 180 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done 180 out of 180 | elapsed:  2.3min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best CV AUC: 0.9748\n",
      "Optimal RF meta-parameters:\n",
      "{'criterion': 'entropy', 'max_depth': 20, 'max_features': 1, 'n_estimators': 200}\n",
      "RF test set AUC: 0.9766\n"
     ]
    }
   ],
   "source": [
    "# Performance of tree-based ensemble algorithms in the form of random forest\n",
    "# 1. Random forest\n",
    "print('Tuning random forest classifier')\n",
    "rf = RandomForestClassifier(random_state=123)\n",
    "\n",
    "# Define meta-parameter grid of candidate settings\n",
    "# The following settings are just for illustration\n",
    "# todo good grid parameters?\n",
    "param_grid = {'n_estimators': [100, 200],\n",
    "              'max_features': [1, 2, 4],\n",
    "              'max_depth': [5, 10, 20],\n",
    "              'criterion' :['gini', 'entropy'],\n",
    "              }\n",
    "\n",
    "# Set up the grid object specifying the tuning options\n",
    "gs_rf = GridSearchCV(rf, param_grid, cv=5, scoring='roc_auc', verbose=1)\n",
    "gs_rf.fit(x_train, y_train)\n",
    "print(\"Best CV AUC: %0.4f\" % gs_rf.best_score_)\n",
    "print(\"Optimal RF meta-parameters:\")\n",
    "print(gs_rf.best_params_)\n",
    "# Find test set auc of the best random forest classifier\n",
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_test, gs_rf.predict_proba(x_test)[:, 1])\n",
    "print('RF test set AUC: {:.4f}'.format(metrics.auc(fp_rate, tp_rate)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 328,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9766369426751594"
      ]
     },
     "execution_count": 328,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_test, gs_rf.predict_proba(x_test)[:, 1])\n",
    "metrics.auc(fp_rate, tp_rate)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9758772836952775\n",
      "1.0\n"
     ]
    }
   ],
   "source": [
    "#save results\n",
    "#todo add ab line\n",
    "auc_test_rf = metrics.auc(fp_rate, tp_rate)\n",
    "\n",
    "print(auc_test_rf)\n",
    "\n",
    "\n",
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_train, gs_rf.predict_proba(x_train)[:, 1])\n",
    "auc_train_rf = metrics.auc(fp_rate, tp_rate)\n",
    "print(auc_train_rf)\n",
    "\n",
    "#todo should this be left or not? otherwise will just store the value of the training set auc for later"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9753413015493395\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#define classifier here todo\n",
    "classifier = gs_rf.best_estimator_\n",
    "print(gs_rf.best_score_)\n",
    "\n",
    "metrics.plot_roc_curve(classifier, x_test, y_test)\n",
    "plt.plot([0, 1], [0, 1], \"r--\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# XGBoost\n",
    "\n",
    "\n",
    "XGBoost (Extreme Gradient Boosting) is a state-of-the-art algorithm. It incorporates both boosting and gradient descent.\n",
    "\n",
    "Boosting is an ensemble method, in which weak learners are combined iteratively to enhance model performance (bias-variance trade-off). With each new model, we add weights on instances with wrong predictions/high errors. This should force the next model to find generalizable patterns in these, more difficult, cases. \n",
    "\n",
    "Next up, gradient descent: \"To find a\n",
    "local minimum of a function using gradient descent, one takes steps proportional to the negative of the gradient (or of the approximate gradient) of the function at the current point.\"(source: https://en.wikipedia.org/wiki/Gradient_descent) <br> Whilst this sounds rather complicated, the idea behind it is quite simple. The sequentielly added model will aim to predict the error function instead of the target variable. By this we try and identify the loss function and each new model aims to compensate the error of the prior model, slowly moving towards the minimum. To avoid overfitting the error function, we implement the learning rate. It is a weighting factor for the correction models, to reduce the impact of the individual model. The learning rate is an extra parameter. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 20 candidates, totalling 100 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done 100 out of 100 | elapsed:  1.7min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best CV AUC: 0.9606\n",
      "Optimal XGB meta-parameters:\n",
      "{'colsample_bytree': 0.7, 'early_stopping_rounds': 10, 'learning_rate': 0.2, 'max_depth': 5, 'n_estimators': 200}\n",
      "XGB test set AUC with optimal meta-parameters: 0.9535\n"
     ]
    }
   ],
   "source": [
    "# todo add some comments on scikit-learn implementation versus xgboost module\n",
    "    #still to do: main difference default parameter settings \n",
    "    #anything else?\n",
    "    \n",
    "# XGBoost module recommends storing data in a different manner\n",
    "DM_train = xgb.DMatrix(data=x_train, label=y_train)\n",
    "DM_test = xgb.DMatrix(data=x_test, label=y_test)\n",
    "#Setting up the grid of xgb hyperparameters\n",
    "#todo choose parameters\n",
    "xgb_param_grid = {\n",
    "    'colsample_bytree': np.linspace(0.5, 0.9, 5),\n",
    "    'n_estimators': [200],\n",
    "    'max_depth': [2,5],\n",
    "    'learning_rate': [0.2, 0.1],\n",
    "    'early_stopping_rounds': [10]}\n",
    "gs_xgb = GridSearchCV(estimator=xgb.XGBClassifier(), param_grid=xgb_param_grid, scoring='roc_auc', cv=5, verbose=1)\n",
    "gs_xgb.fit(x_train, y_train)\n",
    "print(\"Best CV AUC: %0.4f\" % gs_xgb.best_score_)\n",
    "print(\"Optimal XGB meta-parameters:\")\n",
    "print(gs_xgb.best_params_)\n",
    "# Find test set auc of the best random forest classifier\n",
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_test, gs_xgb.predict_proba(x_test)[:, 1])\n",
    "print('XGB test set AUC with optimal meta-parameters: {:.4f}'.format(metrics.auc(fp_rate, tp_rate)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_test, gs_xgb.predict_proba(x_test)[:, 1])\n",
    "\n",
    "#plot auc on test\n",
    "plt.title('ROC curve of XGBoost model on test')\n",
    "plt.plot(fp_rate, tp_rate, \"b\", label=\"AUC = %0.2f\" % roc_auc)\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.plot([0, 1], [0, 1], \"r--\")\n",
    "plt.ylabel(\"True Positive Rate\")\n",
    "plt.xlabel(\"False Positive Rate\")\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9534781109962083\n",
      "0.9999948187319747\n"
     ]
    }
   ],
   "source": [
    "# save data\n",
    "auc_test_xgb = metrics.auc(fp_rate, tp_rate)\n",
    "fp_rate, tp_rate, _ = metrics.roc_curve(y_train, gs_xgb.predict_proba(x_train)[:, 1])\n",
    "auc_train_xgb = metrics.auc(fp_rate, tp_rate)\n",
    "print(auc_test_xgb)\n",
    "print(auc_train_xgb)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Score Training Set</th>\n",
       "      <th>Score Test Set</th>\n",
       "      <th>Score_diff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logit</td>\n",
       "      <td>0.850300</td>\n",
       "      <td>0.846173</td>\n",
       "      <td>0.004127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CV</td>\n",
       "      <td>0.851585</td>\n",
       "      <td>0.852473</td>\n",
       "      <td>0.000888</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Regularized Logit</td>\n",
       "      <td>0.851246</td>\n",
       "      <td>0.848080</td>\n",
       "      <td>0.003166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>RF</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.975877</td>\n",
       "      <td>0.024123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>XGBoost</td>\n",
       "      <td>0.999995</td>\n",
       "      <td>0.953478</td>\n",
       "      <td>0.046517</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Model  Score Training Set  Score Test Set  Score_diff\n",
       "0              Logit            0.850300        0.846173    0.004127\n",
       "1                 CV            0.851585        0.852473    0.000888\n",
       "2  Regularized Logit            0.851246        0.848080    0.003166\n",
       "3                 RF            1.000000        0.975877    0.024123\n",
       "4            XGBoost            0.999995        0.953478    0.046517"
      ]
     },
     "execution_count": 268,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Model Comparison\n",
    "#todo add comments here\n",
    "\n",
    "models = pd.DataFrame({    \n",
    "\n",
    "    'Model': ['Logit', 'CV', 'Regularized Logit', 'RF', 'XGBoost'],\n",
    "    \n",
    "    'Score Training Set': [auc_train_logit, auc_train_cv, auc_train_reglog, auc_train_rf, auc_train_xgb]\n",
    "                    ,\n",
    "    'Score Test Set': [auc_test_logit, auc_test_cv, auc_test_reglog, auc_test_rf, auc_test_xgb]\n",
    "                    })\n",
    "models['Score_diff'] = abs(models['Score Training Set'] - models['Score Test Set'])\n",
    "models\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Score Training Set</th>\n",
       "      <th>Score Test Set</th>\n",
       "      <th>Score_diff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logit</td>\n",
       "      <td>0.850300</td>\n",
       "      <td>0.846173</td>\n",
       "      <td>0.004127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>CV</td>\n",
       "      <td>0.851585</td>\n",
       "      <td>0.852473</td>\n",
       "      <td>0.000888</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Regularized Logit</td>\n",
       "      <td>0.851246</td>\n",
       "      <td>0.848080</td>\n",
       "      <td>0.003166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>RF</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.975877</td>\n",
       "      <td>0.024123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>XGBoost</td>\n",
       "      <td>0.999995</td>\n",
       "      <td>0.953478</td>\n",
       "      <td>0.046517</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Model  Score Training Set  Score Test Set  Score_diff\n",
       "0              Logit            0.850300        0.846173    0.004127\n",
       "1                 CV            0.851585        0.852473    0.000888\n",
       "2  Regularized Logit            0.851246        0.848080    0.003166\n",
       "3                 RF            1.000000        0.975877    0.024123\n",
       "4            XGBoost            0.999995        0.953478    0.046517"
      ]
     },
     "execution_count": 269,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "models['Score_diff'] = abs(models['Score Training Set'] - models['Score Test Set'])\n",
    "models\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot\n",
    "# keep ? \n",
    "#todo add comments \n",
    "plt.figure(figsize=[25,6])\n",
    "xx = models['Model']\n",
    "plt.tick_params(labelsize=14)\n",
    "plt.plot(xx, models['Score Training Set'], label = 'Score Training Set')\n",
    "plt.plot(xx, models['Score Test Set'], label = 'Score Test Set')\n",
    "plt.legend()\n",
    "plt.title('Score of 5 different algorithm')\n",
    "plt.xlabel('Models')\n",
    "plt.ylabel('Score, %')\n",
    "plt.xticks(xx, rotation='vertical')\n",
    "plt.savefig('graph.png')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Feature importance and interpretable machine learning\n",
    "\n",
    "Interpretability of machine learning models is a big topic, because usually, the more complicated the model, the harder it will be to understand how our final predictions are computed. Especially for so called black-box models, methods have been develop to understand the importance of specific features in these models. \n",
    "\n",
    "### Permutation-based feature importance\n",
    "\n",
    "Permuation-based feature importance is a learner-agnostic way to judge the relevance of features. It produces an ordinal feature ranking. In order to do this, we permute one variable by shuffling its values across all observations. This means, each observation will receive a new value. This permutation breaks the relationship to the target variable. Then we make an assumption: the higher the increase of the error, the more important this variable was in building the model. This way we can compute a ranking. \n",
    "\n",
    "Let's give it a try using our Random Forest model.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Computing permutation-based feature importance...\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.rcParams['figure.figsize'] = 20, 10\n",
    "\n",
    "# Permutation-based feature importance\n",
    "print('Computing permutation-based feature importance...')\n",
    "from sklearn.inspection import permutation_importance\n",
    "\n",
    "\n",
    "perm_imp = permutation_importance(classifier, x_test, y_test, scoring='roc_auc', random_state=123)\n",
    "sorted_idx = perm_imp.importances_mean.argsort()\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.boxplot(perm_imp.importances[sorted_idx].T,\n",
    "           vert=False, labels=x_test.columns[sorted_idx])\n",
    "ax.set_title(\"Permutation importance (test set) of {}\".format(str(classifier)[0:str(classifier).find('(')]))\n",
    "fig.tight_layout()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(add text here about importance of variable job for example. What about the dummy that we left out? Do we know anything about it? Why do you think specific variables might have a high impact in the model building.\n",
    "\n",
    "\n",
    "## Partial dependence plot\n",
    "\n",
    "Whilst the permuation-based feature importance ranking has given us more insight on the which variables have been important, it is lacking insight as in how they affected the model. Therefore, we will continue with the partial dependence plot(PDP).\n",
    "PDP shows what type of effect a feature has on our predicted probability. For example, according to the plot above, 'DEBTINC' was very important in the model building process. But how did it affect the model? Did a higher value lead to a lower or higher credit score? We will find out shortly. \n",
    "<br> But firstly, what is the theory behind these plots.\n",
    "PDP's indicate the chosen values on the x-axis and the estimated probabilities on the y-axis. PDP's are computed by setting one variable to a specific value for all observations. Then you calculate the probabilities and average these. You then increase this value and repeat this step. Each average marks one point in the graph. By increasing the values, we can identify how the model output changes based on an increase in the values of each variable. \n",
    "<br> To understand it better, we can have a look at the output:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Computing partial dependence plots...\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x2880 with 17 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.rcParams['figure.figsize'] = 20,40\n",
    "# Partial dependence\n",
    "from sklearn.inspection import plot_partial_dependence\n",
    "\n",
    "print('Computing partial dependence plots...')\n",
    "features = x_train.columns\n",
    "plot_partial_dependence(classifier, x_train, features, n_jobs=3, grid_resolution=20)\n",
    "# print(\"done in {:.3f}s\".format(time() - tic))\n",
    "fig = plt.gcf()\n",
    "fig.suptitle('Partial dependence')\n",
    "fig.subplots_adjust(wspace=0.4, hspace=0.3)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br> Which variables show interesting findings? Can you explain the results? Why do you think specific variables have the impact they do (e.g. 'DEBTINC' or 'NINQ')? What characteristics do the plots of the dummy variables show?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Things to do:  \n",
    "            Save data appropriately for evaluation <br>\n",
    "             <br>  Algorithm evaluation\n",
    "              <br> Proof read\n",
    "              <br> Format markdown text consistently*\n",
    "              consistency for fpr tpr and similar. fp_rate or fpr? find these inconsistencies\n",
    "              \n",
    "              introduce fpr and tpr as temp variables or save fully for plotting roc curves together in the end\n",
    "             \n",
    "              plot logit train and test roc curve together \n",
    "              \n",
    "              add legend in cv plot ? and elsewhere! + titles etc\n",
    "              \n",
    "              "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}