Justin_Lange_code.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Mining Cup 2019 - Justin Lange\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"http://crisp-dm.eu/wp-content/uploads/2013/07/newcrispdiagram.gif\"/>"
      ],
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "url = 'http://crisp-dm.eu/wp-content/uploads/2013/07/newcrispdiagram.gif'\n",
    "Image(url=url)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1. Data Adquisition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df_test = pd.read_csv('./data/test.csv', delimiter=\"|\")\n",
    "df_train = pd.read_csv('./data/train.csv', delimiter=\"|\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Business Case understanding"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "An established food retailer has introduced a self-scanning system that allows customers to scan their items using a handheld mobile scanner while shopping. This type of payment leaves retailers open to the risk that a certain number of customers will take advantage of this freedom to commit fraud by not scanning all of the items in their cart.\n",
    "\n",
    "To minimize losses, the food retailer hopes to identify cases of fraud using targeted follow-up checks. The challenge here is to keep the number of checks as low as possible to avoid unnecessary added expense as well as to avoid putting off innocent customers due to false accusations. The objective  is to create a model to classify the scans as fraudulent or non-fraudulent. The classification does not take into account whether the fraud was committed intentionally or inadvertently. \n",
    "\n",
    "In addition to this general task, the retailer has defined his own domain specific costs outcomes for each possible prediction. This needs to be considered when evaluation possible prediction models.\n",
    "\n",
    "- True Positive Cost: 5€\n",
    "- False Positive Cost: -25 €\n",
    "- True Negative Cost: 0€\n",
    "- False Negative Cost: -5€\n",
    "\n",
    "Thus, the retailer receives a profit of € 5 for every correctly identified fraud attempt (TP). However, for every fraud case that goes unexposed he loses € 5 (FN). A costumer falsely accused of fraud (FP), might not return to this store, which is denoted by a € 25 loss for the retailer. An honest customer identified (TN) correctly means neither profit nor loss for the retailer."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Data Understanding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "scrolled": true
   },
   "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>trustLevel</th>\n",
       "      <th>totalScanTimeInSeconds</th>\n",
       "      <th>grandTotal</th>\n",
       "      <th>lineItemVoids</th>\n",
       "      <th>scansWithoutRegistration</th>\n",
       "      <th>quantityModifications</th>\n",
       "      <th>scannedLineItemsPerSecond</th>\n",
       "      <th>valuePerSecond</th>\n",
       "      <th>lineItemVoidsPerPosition</th>\n",
       "      <th>fraud</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1268</th>\n",
       "      <td>5</td>\n",
       "      <td>1169</td>\n",
       "      <td>35.52</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>0.024808</td>\n",
       "      <td>0.030385</td>\n",
       "      <td>0.206897</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>820</th>\n",
       "      <td>6</td>\n",
       "      <td>1192</td>\n",
       "      <td>51.26</td>\n",
       "      <td>11</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.023490</td>\n",
       "      <td>0.043003</td>\n",
       "      <td>0.392857</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>806</th>\n",
       "      <td>5</td>\n",
       "      <td>1701</td>\n",
       "      <td>88.92</td>\n",
       "      <td>2</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0.008818</td>\n",
       "      <td>0.052275</td>\n",
       "      <td>0.133333</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>315</th>\n",
       "      <td>5</td>\n",
       "      <td>598</td>\n",
       "      <td>13.13</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0.005017</td>\n",
       "      <td>0.021957</td>\n",
       "      <td>3.666667</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1577</th>\n",
       "      <td>1</td>\n",
       "      <td>1823</td>\n",
       "      <td>24.46</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>0.014811</td>\n",
       "      <td>0.013417</td>\n",
       "      <td>0.148148</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      trustLevel  totalScanTimeInSeconds  grandTotal  lineItemVoids  \\\n",
       "1268           5                    1169       35.52              6   \n",
       "820            6                    1192       51.26             11   \n",
       "806            5                    1701       88.92              2   \n",
       "315            5                     598       13.13             11   \n",
       "1577           1                    1823       24.46              4   \n",
       "\n",
       "      scansWithoutRegistration  quantityModifications  \\\n",
       "1268                         1                      4   \n",
       "820                          0                      0   \n",
       "806                          8                      1   \n",
       "315                          3                      2   \n",
       "1577                         4                      5   \n",
       "\n",
       "      scannedLineItemsPerSecond  valuePerSecond  lineItemVoidsPerPosition  \\\n",
       "1268                   0.024808        0.030385                  0.206897   \n",
       "820                    0.023490        0.043003                  0.392857   \n",
       "806                    0.008818        0.052275                  0.133333   \n",
       "315                    0.005017        0.021957                  3.666667   \n",
       "1577                   0.014811        0.013417                  0.148148   \n",
       "\n",
       "      fraud  \n",
       "1268      0  \n",
       "820       0  \n",
       "806       0  \n",
       "315       0  \n",
       "1577      1  "
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train.sample(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the following, I will state a few core expects of the features. This is necessary to get a general feeling of the data and extract potential new features in the next step.\n",
    "\n",
    "* **Trust Level** ~ A customers trust level rated between 1-6\n",
    "* **Total scaned Item in sec** ~ total time in seconds between first and last product scanned\n",
    "* **Grand total** ~ Total of products scanned\n",
    "* **Line items voids** ~ Nr. of voided scanes\n",
    "* **Scans without reg**~ Attempts to scan something without actually scanning something\n",
    "* **Quantity Modification** ~ Number of modified quantities \n",
    "* **Scanned lined items per seconds** ~ Average number of scanned products\n",
    "* **Value per send** ~ Average total value per second\n",
    "* **Line items void per seconds** ~ Avg nr of item voids per total nr of all scanned and not canceled product\n",
    "* **Fraud**~ Classification\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4. Data Preperation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.1 Data Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "trustLevel                     int64\n",
       "totalScanTimeInSeconds         int64\n",
       "grandTotal                   float64\n",
       "lineItemVoids                  int64\n",
       "scansWithoutRegistration       int64\n",
       "quantityModifications          int64\n",
       "scannedLineItemsPerSecond    float64\n",
       "valuePerSecond               float64\n",
       "lineItemVoidsPerPosition     float64\n",
       "fraud                          int64\n",
       "dtype: object"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 1. Checking data types\n",
    "df_train.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of NaN values: 0\n"
     ]
    }
   ],
   "source": [
    "# 2. Checking if rows exists where cells are NaN\n",
    "print(\"Number of NaN values:\", len(df_train) - len(df_train.isna().any(axis=1)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The provided data of DMC 2019 seems as \"clean\" as a could be. All features have appropriate data types and no NaN value exists. Therefore, I will leave the data set as it is and continue with selecting and extracting appropriate features. After selecting and potentially generating new features, I will however consider detecting outliners."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2 Feature Extraction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In general, within a data set, there is many \"hidden\" information which needs to be explored by carefully looking at the data. Referring back to part 3 (Data Understanding), many possible interesting features can be generated. In the following steps I will try to enrich the data set by extracting potential values features. Their valuebility will afterwards be evaluated by measuring correlation with the output variable.\n",
    "\n",
    "1. Extracting **total amount of items** by multiplying **\"scannedLineItemsPerSecond\"** with **\"totalScanTimeInSeconds\"**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train[\"totalItems\"] = df_train[\"scannedLineItemsPerSecond\"] * df_train[\"totalScanTimeInSeconds\"]\n",
    "df_test[\"totalItems\"] = df_test[\"scannedLineItemsPerSecond\"] * df_test[\"totalScanTimeInSeconds\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "2. Extracting **seconds per Euro** by multiplying **\"grandTotal\"** with **\"scannedLineItems\"**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train['secondsPerEuro'] = df_train['grandTotal'] / df_train['totalItems']\n",
    "df_test['secondsPerEuro'] = df_test['grandTotal'] / df_test['totalItems']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Swap target to last column (just for better Understandability...)\n",
    "df_temp_1 = df_train.pop('totalItems') \n",
    "df_temp_2 = df_train.pop('fraud') \n",
    "df_train['totalItems']=df_temp_1 \n",
    "df_train['fraud']=df_temp_2 "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2.1 Automated Feature Extraction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# featuretools for automated feature engineering\n",
    "import featuretools as ft"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Featuretools is an open-source Python library for automatically creating features out of a set of related tables using a technique called Deep Feature Synthesis. Automated feature engineering, like many topics in machine learning, is a complex subject built upon a foundation of simpler ideas. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Entityset: DMC\n",
       "  Entities:\n",
       "    DMC [Rows: 1832, Columns: 6]\n",
       "    DMC2 [Rows: 2, Columns: 6]\n",
       "  Relationships:\n",
       "    DMC.fraud -> DMC2.fraud"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# creating and entity set 'es'\n",
    "es = ft.EntitySet(id = 'DMC')\n",
    "\n",
    "# adding a dataframe to the entity\n",
    "es = es.entity_from_dataframe(entity_id = 'DMC', dataframe = df_train, make_index = True, index = 'index')\n",
    "\n",
    "# create new datafrom from existing entity\n",
    "es.normalize_entity(base_entity_id='DMC', new_entity_id='DMC2', index = 'fraud', \n",
    "                    additional_variables = ['lineItemVoids', 'scansWithoutRegistration', 'scannedLineItemsPerSecond', 'valuePerSecond', 'lineItemVoidsPerPosition'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/svg+xml": [
       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\r\n",
       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\r\n",
       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\r\n",
       "<!-- Generated by graphviz version 2.38.0 (20140413.2041)\r\n",
       " -->\r\n",
       "<!-- Title: DMC Pages: 1 -->\r\n",
       "<svg width=\"241pt\" height=\"288pt\"\r\n",
       " viewBox=\"0.00 0.00 241.00 288.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\r\n",
       "<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 284)\">\r\n",
       "<title>DMC</title>\r\n",
       "<polygon fill=\"white\" stroke=\"none\" points=\"-4,4 -4,-284 237,-284 237,4 -4,4\"/>\r\n",
       "<!-- DMC -->\r\n",
       "<g id=\"node1\" class=\"node\"><title>DMC</title>\r\n",
       "<polygon fill=\"none\" stroke=\"black\" points=\"10.5,-158.5 10.5,-279.5 222.5,-279.5 222.5,-158.5 10.5,-158.5\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"116.5\" y=\"-264.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">DMC (1832 rows)</text>\r\n",
       "<polyline fill=\"none\" stroke=\"black\" points=\"10.5,-256.5 222.5,-256.5 \"/>\r\n",
       "<text text-anchor=\"start\" x=\"18.5\" y=\"-241.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">index : index</text>\r\n",
       "<text text-anchor=\"start\" x=\"18.5\" y=\"-226.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">trustLevel : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"18.5\" y=\"-211.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">totalScanTimeInSeconds : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"18.5\" y=\"-196.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">secondsPerEuro : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"18.5\" y=\"-181.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">totalItems : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"18.5\" y=\"-166.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">fraud : id</text>\r\n",
       "</g>\r\n",
       "<!-- DMC2 -->\r\n",
       "<g id=\"node2\" class=\"node\"><title>DMC2</title>\r\n",
       "<polygon fill=\"none\" stroke=\"black\" points=\"0,-0.5 0,-121.5 233,-121.5 233,-0.5 0,-0.5\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"116.5\" y=\"-106.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">DMC2 (2 rows)</text>\r\n",
       "<polyline fill=\"none\" stroke=\"black\" points=\"0,-98.5 233,-98.5 \"/>\r\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-83.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">fraud : index</text>\r\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-68.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">lineItemVoids : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-53.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">scansWithoutRegistration : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-38.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">scannedLineItemsPerSecond : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-23.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">valuePerSecond : numeric</text>\r\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-8.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">lineItemVoidsPerPosition : numeric</text>\r\n",
       "</g>\r\n",
       "<!-- DMC&#45;&gt;DMC2 -->\r\n",
       "<g id=\"edge1\" class=\"edge\"><title>DMC&#45;&gt;DMC2</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M116.5,-158.365C116.5,-158.365 116.5,-131.734 116.5,-131.734\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"120,-131.734 116.5,-121.734 113,-131.734 120,-131.734\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"101.5\" y=\"-133.849\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">fraud</text>\r\n",
       "</g>\r\n",
       "</g>\r\n",
       "</svg>\r\n"
      ],
      "text/plain": [
       "<graphviz.dot.Digraph at 0x263b8d66da0>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "es.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Deep Feature Synthesis with Default Primitives¶"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "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>name</th>\n",
       "      <th>type</th>\n",
       "      <th>description</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>is_weekend</td>\n",
       "      <td>transform</td>\n",
       "      <td>Determines if a date falls on a weekend.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>longitude</td>\n",
       "      <td>transform</td>\n",
       "      <td>Returns the second tuple value in a list of LatLong tuples.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>multiply_boolean</td>\n",
       "      <td>transform</td>\n",
       "      <td>Element-wise multiplication of two lists of boolean values.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>not_equal</td>\n",
       "      <td>transform</td>\n",
       "      <td>Determines if values in one list are not equal to another list.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>month</td>\n",
       "      <td>transform</td>\n",
       "      <td>Determines the month value of a datetime.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>isin</td>\n",
       "      <td>transform</td>\n",
       "      <td>Determines whether a value is present in a provided list.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>cum_mean</td>\n",
       "      <td>transform</td>\n",
       "      <td>Calculates the cumulative mean.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>divide_numeric_scalar</td>\n",
       "      <td>transform</td>\n",
       "      <td>Divide each element in the list by a scalar.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>equal</td>\n",
       "      <td>transform</td>\n",
       "      <td>Determines if values in one list are equal to another list.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>num_characters</td>\n",
       "      <td>transform</td>\n",
       "      <td>Calculates the number of characters in a string.</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     name       type  \\\n",
       "22             is_weekend  transform   \n",
       "23              longitude  transform   \n",
       "24       multiply_boolean  transform   \n",
       "25              not_equal  transform   \n",
       "26                  month  transform   \n",
       "27                   isin  transform   \n",
       "28               cum_mean  transform   \n",
       "29  divide_numeric_scalar  transform   \n",
       "30                  equal  transform   \n",
       "31         num_characters  transform   \n",
       "\n",
       "                                                        description  \n",
       "22                         Determines if a date falls on a weekend.  \n",
       "23      Returns the second tuple value in a list of LatLong tuples.  \n",
       "24      Element-wise multiplication of two lists of boolean values.  \n",
       "25  Determines if values in one list are not equal to another list.  \n",
       "26                        Determines the month value of a datetime.  \n",
       "27        Determines whether a value is present in a provided list.  \n",
       "28                                  Calculates the cumulative mean.  \n",
       "29                     Divide each element in the list by a scalar.  \n",
       "30      Determines if values in one list are equal to another list.  \n",
       "31                 Calculates the number of characters in a string.  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# List the primitives (transformation) in a dataframe\n",
    "primitives = ft.list_primitives()\n",
    "pd.options.display.max_colwidth = 100\n",
    "\n",
    "primitives[primitives['type'] == 'transform'].head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Without using any domain knowledge we can make thousands of features by using the default primitives in featuretools. This first call will use the default aggregation and transformation primitives, a max depth of 2, and calculate primitives for the entity. We will only generate the features themselves (the names and not the values) which we can save and inspect."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Built 35 features\n",
      "EntitySet scattered to 3 workers in 5 seconds                                                                          \n",
      "Elapsed: 00:01 | Progress: 100%|███████████████████████████████████████████████████████████████████████████████████████\n"
     ]
    }
   ],
   "source": [
    "# DFS with  primitives\n",
    "feature_matrix, feature_names = ft.dfs(entityset=es, target_entity = 'DMC', max_depth = 2, verbose = 1, n_jobs = 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['trustLevel', 'totalScanTimeInSeconds', 'secondsPerEuro', 'totalItems',\n",
       "       'fraud', 'DMC2.lineItemVoids', 'DMC2.scansWithoutRegistration',\n",
       "       'DMC2.scannedLineItemsPerSecond', 'DMC2.valuePerSecond',\n",
       "       'DMC2.lineItemVoidsPerPosition', 'DMC2.SUM(DMC.trustLevel)',\n",
       "       'DMC2.SUM(DMC.secondsPerEuro)', 'DMC2.SUM(DMC.totalScanTimeInSeconds)',\n",
       "       'DMC2.SUM(DMC.totalItems)', 'DMC2.STD(DMC.trustLevel)',\n",
       "       'DMC2.STD(DMC.secondsPerEuro)', 'DMC2.STD(DMC.totalScanTimeInSeconds)',\n",
       "       'DMC2.STD(DMC.totalItems)', 'DMC2.MAX(DMC.trustLevel)',\n",
       "       'DMC2.MAX(DMC.secondsPerEuro)', 'DMC2.MAX(DMC.totalScanTimeInSeconds)',\n",
       "       'DMC2.MAX(DMC.totalItems)', 'DMC2.SKEW(DMC.trustLevel)',\n",
       "       'DMC2.SKEW(DMC.secondsPerEuro)',\n",
       "       'DMC2.SKEW(DMC.totalScanTimeInSeconds)', 'DMC2.SKEW(DMC.totalItems)',\n",
       "       'DMC2.MIN(DMC.trustLevel)', 'DMC2.MIN(DMC.secondsPerEuro)',\n",
       "       'DMC2.MIN(DMC.totalScanTimeInSeconds)', 'DMC2.MIN(DMC.totalItems)',\n",
       "       'DMC2.MEAN(DMC.trustLevel)', 'DMC2.MEAN(DMC.secondsPerEuro)',\n",
       "       'DMC2.MEAN(DMC.totalScanTimeInSeconds)', 'DMC2.MEAN(DMC.totalItems)',\n",
       "       'DMC2.COUNT(DMC)'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_matrix.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "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>trustLevel</th>\n",
       "      <th>totalScanTimeInSeconds</th>\n",
       "      <th>secondsPerEuro</th>\n",
       "      <th>totalItems</th>\n",
       "      <th>fraud</th>\n",
       "      <th>DMC2.lineItemVoids</th>\n",
       "      <th>DMC2.scansWithoutRegistration</th>\n",
       "      <th>DMC2.scannedLineItemsPerSecond</th>\n",
       "      <th>DMC2.valuePerSecond</th>\n",
       "      <th>DMC2.lineItemVoidsPerPosition</th>\n",
       "      <th>...</th>\n",
       "      <th>DMC2.SKEW(DMC.totalItems)</th>\n",
       "      <th>DMC2.MIN(DMC.trustLevel)</th>\n",
       "      <th>DMC2.MIN(DMC.secondsPerEuro)</th>\n",
       "      <th>DMC2.MIN(DMC.totalScanTimeInSeconds)</th>\n",
       "      <th>DMC2.MIN(DMC.totalItems)</th>\n",
       "      <th>DMC2.MEAN(DMC.trustLevel)</th>\n",
       "      <th>DMC2.MEAN(DMC.secondsPerEuro)</th>\n",
       "      <th>DMC2.MEAN(DMC.totalScanTimeInSeconds)</th>\n",
       "      <th>DMC2.MEAN(DMC.totalItems)</th>\n",
       "      <th>DMC2.COUNT(DMC)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.0</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "      <td>1832.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>3.396834</td>\n",
       "      <td>933.483624</td>\n",
       "      <td>5.519836</td>\n",
       "      <td>15.746725</td>\n",
       "      <td>0.056769</td>\n",
       "      <td>7.227074</td>\n",
       "      <td>0.283843</td>\n",
       "      <td>0.027869</td>\n",
       "      <td>0.049769</td>\n",
       "      <td>0.251694</td>\n",
       "      <td>...</td>\n",
       "      <td>0.019464</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.001575</td>\n",
       "      <td>7.052402</td>\n",
       "      <td>1.965066</td>\n",
       "      <td>3.396834</td>\n",
       "      <td>5.519836</td>\n",
       "      <td>933.483624</td>\n",
       "      <td>15.746725</td>\n",
       "      <td>1635.807860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.711672</td>\n",
       "      <td>529.815609</td>\n",
       "      <td>6.948540</td>\n",
       "      <td>8.513185</td>\n",
       "      <td>0.231463</td>\n",
       "      <td>0.925852</td>\n",
       "      <td>1.157315</td>\n",
       "      <td>0.001447</td>\n",
       "      <td>0.008679</td>\n",
       "      <td>0.042056</td>\n",
       "      <td>...</td>\n",
       "      <td>0.173335</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001892</td>\n",
       "      <td>20.600210</td>\n",
       "      <td>3.934871</td>\n",
       "      <td>0.552775</td>\n",
       "      <td>0.866075</td>\n",
       "      <td>58.999961</td>\n",
       "      <td>2.544404</td>\n",
       "      <td>375.895959</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>0.001111</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.027514</td>\n",
       "      <td>0.014403</td>\n",
       "      <td>0.241379</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.686889</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.001111</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.144231</td>\n",
       "      <td>1.990506</td>\n",
       "      <td>919.013310</td>\n",
       "      <td>15.122685</td>\n",
       "      <td>104.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>2.000000</td>\n",
       "      <td>476.500000</td>\n",
       "      <td>1.647291</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.027514</td>\n",
       "      <td>0.051898</td>\n",
       "      <td>0.241379</td>\n",
       "      <td>...</td>\n",
       "      <td>0.061976</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.001111</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.532407</td>\n",
       "      <td>5.732249</td>\n",
       "      <td>919.013310</td>\n",
       "      <td>15.122685</td>\n",
       "      <td>1728.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>3.000000</td>\n",
       "      <td>933.500000</td>\n",
       "      <td>3.241556</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.027514</td>\n",
       "      <td>0.051898</td>\n",
       "      <td>0.241379</td>\n",
       "      <td>...</td>\n",
       "      <td>0.061976</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.001111</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.532407</td>\n",
       "      <td>5.732249</td>\n",
       "      <td>919.013310</td>\n",
       "      <td>15.122685</td>\n",
       "      <td>1728.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>5.000000</td>\n",
       "      <td>1399.000000</td>\n",
       "      <td>6.007321</td>\n",
       "      <td>23.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.027514</td>\n",
       "      <td>0.051898</td>\n",
       "      <td>0.241379</td>\n",
       "      <td>...</td>\n",
       "      <td>0.061976</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.001111</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.532407</td>\n",
       "      <td>5.732249</td>\n",
       "      <td>919.013310</td>\n",
       "      <td>15.122685</td>\n",
       "      <td>1728.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>6.000000</td>\n",
       "      <td>1831.000000</td>\n",
       "      <td>42.040000</td>\n",
       "      <td>30.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>11.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.033766</td>\n",
       "      <td>0.051898</td>\n",
       "      <td>0.423077</td>\n",
       "      <td>...</td>\n",
       "      <td>0.061976</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.009286</td>\n",
       "      <td>91.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>3.532407</td>\n",
       "      <td>5.732249</td>\n",
       "      <td>1173.913462</td>\n",
       "      <td>26.115385</td>\n",
       "      <td>1728.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 35 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        trustLevel  totalScanTimeInSeconds  secondsPerEuro   totalItems  \\\n",
       "count  1832.000000             1832.000000     1832.000000  1832.000000   \n",
       "mean      3.396834              933.483624        5.519836    15.746725   \n",
       "std       1.711672              529.815609        6.948540     8.513185   \n",
       "min       1.000000                2.000000        0.001111     1.000000   \n",
       "25%       2.000000              476.500000        1.647291     8.000000   \n",
       "50%       3.000000              933.500000        3.241556    16.000000   \n",
       "75%       5.000000             1399.000000        6.007321    23.000000   \n",
       "max       6.000000             1831.000000       42.040000    30.000000   \n",
       "\n",
       "             fraud  DMC2.lineItemVoids  DMC2.scansWithoutRegistration  \\\n",
       "count  1832.000000         1832.000000                    1832.000000   \n",
       "mean      0.056769            7.227074                       0.283843   \n",
       "std       0.231463            0.925852                       1.157315   \n",
       "min       0.000000            7.000000                       0.000000   \n",
       "25%       0.000000            7.000000                       0.000000   \n",
       "50%       0.000000            7.000000                       0.000000   \n",
       "75%       0.000000            7.000000                       0.000000   \n",
       "max       1.000000           11.000000                       5.000000   \n",
       "\n",
       "       DMC2.scannedLineItemsPerSecond  DMC2.valuePerSecond  \\\n",
       "count                     1832.000000          1832.000000   \n",
       "mean                         0.027869             0.049769   \n",
       "std                          0.001447             0.008679   \n",
       "min                          0.027514             0.014403   \n",
       "25%                          0.027514             0.051898   \n",
       "50%                          0.027514             0.051898   \n",
       "75%                          0.027514             0.051898   \n",
       "max                          0.033766             0.051898   \n",
       "\n",
       "       DMC2.lineItemVoidsPerPosition  ...  DMC2.SKEW(DMC.totalItems)  \\\n",
       "count                    1832.000000  ...                1832.000000   \n",
       "mean                        0.251694  ...                   0.019464   \n",
       "std                         0.042056  ...                   0.173335   \n",
       "min                         0.241379  ...                  -0.686889   \n",
       "25%                         0.241379  ...                   0.061976   \n",
       "50%                         0.241379  ...                   0.061976   \n",
       "75%                         0.241379  ...                   0.061976   \n",
       "max                         0.423077  ...                   0.061976   \n",
       "\n",
       "       DMC2.MIN(DMC.trustLevel)  DMC2.MIN(DMC.secondsPerEuro)  \\\n",
       "count                    1832.0                   1832.000000   \n",
       "mean                        1.0                      0.001575   \n",
       "std                         0.0                      0.001892   \n",
       "min                         1.0                      0.001111   \n",
       "25%                         1.0                      0.001111   \n",
       "50%                         1.0                      0.001111   \n",
       "75%                         1.0                      0.001111   \n",
       "max                         1.0                      0.009286   \n",
       "\n",
       "       DMC2.MIN(DMC.totalScanTimeInSeconds)  DMC2.MIN(DMC.totalItems)  \\\n",
       "count                           1832.000000               1832.000000   \n",
       "mean                               7.052402                  1.965066   \n",
       "std                               20.600210                  3.934871   \n",
       "min                                2.000000                  1.000000   \n",
       "25%                                2.000000                  1.000000   \n",
       "50%                                2.000000                  1.000000   \n",
       "75%                                2.000000                  1.000000   \n",
       "max                               91.000000                 18.000000   \n",
       "\n",
       "       DMC2.MEAN(DMC.trustLevel)  DMC2.MEAN(DMC.secondsPerEuro)  \\\n",
       "count                1832.000000                    1832.000000   \n",
       "mean                    3.396834                       5.519836   \n",
       "std                     0.552775                       0.866075   \n",
       "min                     1.144231                       1.990506   \n",
       "25%                     3.532407                       5.732249   \n",
       "50%                     3.532407                       5.732249   \n",
       "75%                     3.532407                       5.732249   \n",
       "max                     3.532407                       5.732249   \n",
       "\n",
       "       DMC2.MEAN(DMC.totalScanTimeInSeconds)  DMC2.MEAN(DMC.totalItems)  \\\n",
       "count                            1832.000000                1832.000000   \n",
       "mean                              933.483624                  15.746725   \n",
       "std                                58.999961                   2.544404   \n",
       "min                               919.013310                  15.122685   \n",
       "25%                               919.013310                  15.122685   \n",
       "50%                               919.013310                  15.122685   \n",
       "75%                               919.013310                  15.122685   \n",
       "max                              1173.913462                  26.115385   \n",
       "\n",
       "       DMC2.COUNT(DMC)  \n",
       "count      1832.000000  \n",
       "mean       1635.807860  \n",
       "std         375.895959  \n",
       "min         104.000000  \n",
       "25%        1728.000000  \n",
       "50%        1728.000000  \n",
       "75%        1728.000000  \n",
       "max        1728.000000  \n",
       "\n",
       "[8 rows x 35 columns]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_matrix.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.3 Feature Selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since not all features are relevant, I will filter irrelevant features out for my models. The filtering is applied using a correlation matrix (Pearson correlation). I will only select those features which have a correlation of above 0.02 (taking absolute value) with the output variable \"**fraud**\".\n",
    "\n",
    "As we can see in the following, there are only a few features which we can classify as correlated. Most of the features are correlated around the notion of 0. Therefore, I will remove those features, which have practically no correlation, setting the absolute boundary at 0.02.\n",
    "\n",
    "Furthermore, the extracted features **total items** and **seconds per euro** are comparable strongly correlated. Wuhu!\n",
    "\n",
    "In the following code section, I will remove all non-relevant features given the above defined boundary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x720 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#Using Pearson Correlation\n",
    "plt.figure(figsize=(12,10))\n",
    "cor = df_train.corr()\n",
    "sns.heatmap(cor, annot=True, cmap=plt.cm.Reds)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "trustLevel                   0.319765\n",
       "totalScanTimeInSeconds       0.110414\n",
       "lineItemVoids                0.063496\n",
       "scansWithoutRegistration     0.074123\n",
       "scannedLineItemsPerSecond    0.023085\n",
       "valuePerSecond               0.028873\n",
       "lineItemVoidsPerPosition     0.090116\n",
       "secondsPerEuro               0.100765\n",
       "totalItems                   0.298423\n",
       "fraud                        1.000000\n",
       "Name: fraud, dtype: float64"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cor_target = abs(cor[\"fraud\"])\n",
    "#Selecting highly correlated features\n",
    "relevant_features = cor_target[cor_target>0.02]\n",
    "relevant_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Only keep relevant features\n",
    "df_train = df_train[relevant_features.index.values] # Training Set\n",
    "df_test = df_test[relevant_features.iloc[:-1].index.values] # Testing Set"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.4. Outlier Detection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_boxplots(df):\n",
    "    fig, axs = plt.subplots(nrows=2, ncols=3,squeeze = True)\n",
    "    fig.tight_layout()\n",
    "    fig.set_figheight(10)\n",
    "    fig.set_figwidth(10)\n",
    "\n",
    "    sns.boxplot(x=df['trustLevel'], ax=axs[0,0])\n",
    "    sns.boxplot(x=df['totalScanTimeInSeconds'], ax=axs[1,0])\n",
    "    sns.boxplot(x=df['lineItemVoidsPerPosition'], ax=axs[0,2])\n",
    "    sns.boxplot(x=df['secondsPerEuro'],ax=axs[0,1])\n",
    "    sns.boxplot(x=df['totalItems'], ax=axs[1,1])\n",
    "    sns.boxplot(x=df['fraud'],ax=axs[1,2])\n",
    "    \n",
    "plot_boxplots(df_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the following, I will remove the outliers in **'secondsPerEuro'** and **'LineItemVoidsPerPosition'**. This will be done by keeping only the rows that are within +3 to -3 standard deviations in the column **'LineItemVoidsPerPosition'** and **'secondsPerEuro'**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "# Training Set\n",
    "df_train = df_train[np.abs(df_train.secondsPerEuro-df_train.secondsPerEuro.mean()) <= (3*df_train.secondsPerEuro.std())]\n",
    "# df_train = df_train[np.abs(df_train.lineItemVoidsPerPosition-df_train.lineItemVoidsPerPosition.mean()) <= (3*df_train.lineItemVoidsPerPosition.std())]\n",
    "\n",
    "plot_boxplots(df_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.5. Handling Imbalanced data using SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting Data Distribution\n",
    "sns.set(style=\"ticks\", color_codes=True)\n",
    "sns.catplot(x=\"fraud\", kind=\"count\", palette=\"ch:.25\", data=df_train);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we can see, the data is highly imbalanced. This will require sample techniques such as SMOTE, Tomek Links, etc.: \n",
    "- oversampling the minority class\n",
    "- undersampling the majority class.\n",
    "\n",
    "Since the overall dataset is sufficient to train a model, it is unfortunately not sufficient to undersampling the majority class. Therefore, I will only consider oversampling methods of the minority class.\n",
    "\n",
    "In the following, I will split the available data into test/evaluation and train sets before using sampling techniques! Sampling before splitting the data can allow the exact same observations to be present in both the test/evaluation and train sets. This can allow the model to simply memorize specific data points and cause overfitting and poor generalization to the test/evaluation data.\n",
    "\n",
    "Since python provides a variety of packages, their are also multiple ways to apply oversampling techniques. I will  consider the SMOTE technique since this was also suggested in the appendix of this task."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.5.1 Oversampling Minority class using SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import warnings\n",
    "from imblearn.over_sampling import SMOTE\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "warnings.simplefilter(action='ignore', category=FutureWarning) # for readability purposes\n",
    "\n",
    "# X_train, X_test, y_train, y_test = train_test_split(\n",
    "#     df_train.drop(columns=['fraud']), df_train['fraud'], test_size=0.2, random_state=np.random.seed(1304)) \n",
    "\n",
    "# sm = SMOTE(random_state=np.random.seed(1304), ratio=1.0)\n",
    "# X_train, y_train = sm.fit_sample(X_train, y_train)\n",
    "\n",
    "sm = SMOTE(random_state=np.random.seed(1304), ratio=1.0)\n",
    "X_train, y_train = sm.fit_sample(df_train.drop(columns=['fraud']), df_train['fraud'])\n",
    "\n",
    "X_test = df_test\n",
    "y_test = pd.read_csv('./data/testY.csv', delimiter=\"|\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# check new class counts\n",
    "sns.set(style=\"ticks\", color_codes=True)\n",
    "sns.catplot(x=\"fraud\", kind=\"count\", palette=\"ch:.25\", data=pd.DataFrame(data=y_train, columns=['fraud']));"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Besides using sampling techniques, we also need to consider appropriate evaluation metrics based on True/False Positives and Negatives such as:\n",
    "- Precision\n",
    "- Recall\n",
    "- Domain-specific cost function"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 5. Modelling\n",
    "## 5.1 General domain-specific consideration\n",
    "### 5.1.1 Domain-specific costs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As in the data mining cup 2019 pdf-files states, evaluation need to consider cost in terms of prediction outcome:\n",
    "- **True Positive** Cost: 5€\n",
    "- **False Positive** Cost: -25 €\n",
    "- **True Negative** Cost: 0€\n",
    "- **False Negative** Cost: -5€\n",
    "\n",
    "Thus, the retailer receives a profit of € 5 for every correctly identified fraud attempt (TP). However, for every fraud case that goes unexposed he loses € 5 (FN). A costumer falsely accused of fraud (FP), might not return to this store, which is denoted by a € 25 loss for the retailer. An honest customer identified (TN) correctly means neither profit nor loss for the retailer.\n",
    "\n",
    "The sum of the costs or profit of all scans is the monetary value to be considered in the evaluation. Since predicting a fraud comes with considerable costs, it is important to be sensible in predicting Positives as a hole.\n",
    "\n",
    "In the following, I will create two functions. One for calculating the domain specific cost, and one for plotting the confusion matrix."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "def calc_domain_specific_metrics(y_true, y_pred):\n",
    "    tp_cost = 5\n",
    "    fp_cost = -25\n",
    "    tn_cost = 0\n",
    "    fn_cost = -5\n",
    "    \n",
    "    tn, fp, fn, tp = confusion_matrix(y_true, y_pred).ravel()\n",
    "    print(\"Precision: \\t\\t\", format(tp/(tp+fp), '.2f'))\n",
    "    print(\"Recall: \\t\\t\",  format(tp/(tp+fn), '.2f'))\n",
    "    print(\"Domain Specific return: \" , ((tp_cost*tp) + (fp_cost*fp) + (tn_cost*tn) + (fn_cost*fn)), '€')    \n",
    "    \n",
    "def calc_return_only(y_true, y_pred):\n",
    "    tp_cost = 5\n",
    "    fp_cost = -25\n",
    "    tn_cost = 0\n",
    "    fn_cost = -5\n",
    "    \n",
    "    tn, fp, fn, tp = confusion_matrix(y_true, y_pred).ravel()\n",
    "    \n",
    "    return ((tp_cost*tp) + (fp_cost*fp) + (tn_cost*tn) + (fn_cost*fn))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1.2 Maximum Possible Profit\n",
    "The best possible outcome in this task is the highest possible return. To measure  this we need the following\n",
    "1. Training data's true targets: **y_train**\n",
    "2. Domain specific return function: **calc_return_only**\n",
    "\n",
    "Then we can compute the maximum possible return assuming we have predicted all outcomes correct:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximimum possible return in training spit: 118635 €\n"
     ]
    }
   ],
   "source": [
    "print(\"Maximimum possible return in training spit:\", calc_return_only(y_true=y_test, y_pred=y_test),\"€\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximimum possible loss in training spit: -11978485 €\n"
     ]
    }
   ],
   "source": [
    "print(\"Maximimum possible loss in training spit:\", calc_return_only(y_true=y_test, y_pred=(~y_test.astype(bool)).astype(int)),\"€\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Given the training set which I have splitting using random 80-20 split, we can achieve a profit within the boundaries of **[-11.978.485 € : 118.635 €]**. This means that there is \"not much to win\". Given this insight it will be important to be extremely careful and sensible in predicting positive fraud, since False Positive will result in huge profit loss."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " ### 5.1.3 Precision favoured over Recall\n",
    "\n",
    "Given the above domain specific costs, we can see that we need to consider the trade-off between high precision, and high recall values. However, since False Positive is extremely costly for the retailers, and True Positives provides long-term profit benefits, I will prefer models with high precision score since precision is trying to minimize False Positive while maximizing True Positives. Thus, if precision is high, we can assume False Positives are low and True Positives are High.\n",
    "\n",
    "Given the importance, the following function will be used for plotting a confusion matrix for every single model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import plot_confusion_matrix\n",
    "\n",
    "def plot_confusion_mat(X_test, y_test, cls, txt):\n",
    "    title = \"Confusion Matrix of \" + txt\n",
    "    disp = plot_confusion_matrix(cls, X_test, y_test,\n",
    "                                     display_labels=['No Fraud', 'Fraud'],\n",
    "                                     cmap=plt.cm.Blues,\n",
    "                                     )\n",
    "    disp.ax_.set_title(title)\n",
    "\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1.4 Algorithm considerations\n",
    "Since our domain specific problem creates high costs from false predictions (in particular false positive) it is important to consider appropriate model design decisions. In general, no predefined classifier or solution exists for any problem (free lunch theorem). Two general approaches however can be considered for huge costs caused by false classification.\n",
    "\n",
    "#### 1. Weight update scheme of algorithm\n",
    "Since facing a supervised problem, the learning rate for the target variable can be set differently. In particular, setting the learning rate for fraud=1 higher compared to fraud=0 would increase sensibility towards predicting fraud.\n",
    "\n",
    "#### 2. Optimal decision threshold\n",
    "Another option to increase sensibility toward predicting fraud is setting a probability threshold. As mention above, we cannot gain much \"profit\" anyway (max = 80€). However, loosing profit can be achieved very fast (min = -8630€)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.2 Modelling the Baseline: Naïve Bayes\n",
    "Fundamentally, a baseline is a model that is both simple to set up and has a reasonable chance of providing decent results. Usually, a baseline model consists of relatively few trainable parameters and can be quickly fitted to the data without too much work. Using such a simple baseline, it is easier to identify where a simple baseline model fails for a given task and to make conclusions about where improvements are necessary. \n",
    "\n",
    "The Naïve Bayes classifier is simple to understand, fast to train, and only has a limited amount of hyperparameters to optimize. It is also one of the most straightforward yet most effective classification techniques."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.naive_bayes import GaussianNB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [],
   "source": [
    "cls_nb = pipeline.Pipeline([\n",
    "    ('scal', StandardScaler()),\n",
    "    ('clf', GaussianNB())\n",
    "])\n",
    "\n",
    "cls_nb = cls_nb.fit(X_train, y_train)\n",
    "predY_nb = cls_nb.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.35\n",
      "Recall: \t\t 0.96\n",
      "Domain Specific return:  -929970 €\n"
     ]
    }
   ],
   "source": [
    "plot_confusion_mat(X_test, y_test, cls_nb, \"Baseline: Naive Bayer\")\n",
    "calc_domain_specific_metrics(y_test, predY_nb)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The baseline model returns high recall score, low precision, and an overall negative return in the domain specific use case. The increase the precision, we need to find a way which is explicitly increasing precision. Precision increases if False Positives increases and/or True Negatives. Since we have a True Negative occurrence, we need to focus on the increasing True Positives. This will ultimately result in the reduction of costly False Positives.\n",
    "\n",
    "One idea mentioned above is to modify the weights update scheme of an algorithm. One algorithm that can explicitly model weights classes is SVM with the *class_weight* parameter.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.2 Weight update scheme of algorithm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2.1 Support Vector Machine - Weight configuration"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "SVM has two core advantages in our use case. It I san algorithm that can explicitly model weights classes with *class_weight* parameter and be can specify the level of sensibility towards misclassification using the hyperparameter *C*."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC\n",
    "\n",
    "cls_svm = pipeline.Pipeline([                          # Setting up Pipeline for SVM\n",
    "    ('scale', StandardScaler()),\n",
    "    ('clf', SVC(C=0.1, kernel='linear', class_weight={0:1.0, 1:2.0}, probability=True, random_state = np.random.seed(1304)))\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "cls_svm = cls_svm.fit(X_train, y_train)                # Train on training set\n",
    "predY_svm = cls_svm.predict(X_test)                    # Predict on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.61\n",
      "Recall: \t\t 1.00\n",
      "Domain Specific return:  -267760 €\n"
     ]
    }
   ],
   "source": [
    "plot_confusion_mat(X_test, y_test, cls_svm, \"SVM\")\n",
    "calc_domain_specific_metrics(y_test, predY_svm)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2.3 Random Forest - Weight configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "cls_rforest = pipeline.Pipeline([                       # Setting up Pipeline for RF\n",
    "    ('scale', StandardScaler()),\n",
    "    ('clf', RandomForestClassifier(class_weight={0:1.0, 1:20.0}, n_estimators=10, random_state = np.random.seed(1304)))\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "cls_rforest = cls_rforest.fit(X_train, y_train)         # Train on training set\n",
    "predY_rforest = cls_rforest.predict(X_test)             # Predict on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEXCAYAAABWNASkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nO3deVxUVf/A8c8Mmxoi8jwgrvxKc0+tcA3B1GQZEEEtE9G0R9Nya8EASTN3JTUzxSytjEpzATPAUlNzNyv3rExUQBZFRCC2mfv7g4d5JJABRZTr9+3rvnTO3b5nRr6cOffcczWKoigIIYRQFe29DkAIIUTVk+QuhBAqJMldCCFUSJK7EEKokCR3IYRQIUnuQgihQpLcq5ler2fNmjX4+/vj6+uLl5cXCxcuJD8//46OOW7cONzd3fn8888rvf+JEyeYOHHibZ//n3r37k2nTp3Izs4uUb5p0yZatWpFXFxcufvfuHGD4cOH33K9r68vmZmZdxxnee/bpk2bePLJJ/H19cXX15f+/fvTu3dvpkyZQl5e3h2fu9hLL73Epk2bqux4NwsMDKR3797GOhQvd5upz09UD/N7HcCD5u233+b69et8+umn1K1bl5ycHN544w2mTp3KwoULb+uYKSkp7N27l19//RUzM7NK7//YY4+xdOnS2zr3rdSvX5/vv/+eAQMGGMuioqL497//bXLf69evc+LEiVuuj46OrpIYTb1vzs7OrFy50vg6Ly+P559/ns2bNzNkyJAqieFumzJlCh4eHtV6TlOfn6ge0nKvRgkJCXzzzTfMmTOHunXrAlCnTh1mzJhB3759gaJWzxtvvIG3tzc+Pj4sWLCAwsJCoCgJv//++wwZMoTevXvzxRdfkJWVxX/+8x8KCwvx9/fn4sWLtGrVivT0dON5i19nZ2czceJEfH198fPzIywsDIPBwKFDh/D29r6t899K//792bJli/F1YmIiOTk5PPLII8ayDRs2MHjwYAYMGMDTTz9tPF5ISAi5ubn4+vqi1+tp3749kyZNwt3dnRMnThjrs2zZMoYMGYJeryctLQ0XFxcOHjxYKpaffvqJZ599Fh8fH/z9/dmzZ0+Z75spGRkZZGVlUa9ePQB++OEHhgwZgr+/P7169WLJkiUAHDp0iCFDhhAUFMSAAQPw9vbm6NGjQNEvlJEjR6LT6Rg9ejRpaWnlxglF3yLGjh3LmDFj8Pb25sUXX2Tbtm0EBgbSs2dPVq9ebTL2f0pOTmbs2LH4+Pjg7e3NRx99BBT9H3Vzc2PUqFG4u7uTmprKzz//zNChQ/Hz82PgwIH88MMPAKSlpTFq1Cj8/Pzw8/Mz1v+fn5+4RxRRbeLi4pSBAweWu82UKVOUmTNnKgaDQcnLy1NGjRqlrFy5UlEURWnZsqWydu1aRVEU5cSJE0r79u2V3Nxc5dKlS0qnTp2Mx2jZsqVy9erVUq83b96sjBo1SlEURSksLFSmTp2qxMfHKwcPHlR0Ot1tn/+fnn76aeXo0aNK9+7dlZSUFEVRFOWDDz5Q1q5dqwwbNkyJjY1VsrKylGeffVZJT09XFEVRfvnlF2MdyqrP5s2bS9WnsLBQCQgIUFauXKm88MILyooVK0rFkp6ernTv3l359ddfFUVRlN9//13p0qWLcvHixVLnudnGjRuVJ554Qunfv7/i7u6udO3aVXnuueeUL7/8UlEURTEYDMqwYcOU8+fPK4qiKMnJyUqbNm2Uq1evKgcPHlTatGmjnD59WlEURfn444+VgIAARVEU5eWXX1YWL16sKIqixMfHK506dVI2btxYbpwbN25UnnzySSUpKUnR6/WKl5eXMmHCBEWv1ytnzpxRHnvsMUWv15eqw7Bhw5Snn35a6d+/v3HZtWuXoiiKEhAQoKxevVpRFEXJzMxUfHx8lK1btyqXLl1SWrZsqRw5ckRRFEXJyMhQ+vXrp1y6dMlYT1dXVyUxMVFZtmyZ8tZbbymKoijZ2dnK5MmTlczMzHLfV1F9pOVejbRaLQaDodxt9uzZw7Bhw9BoNFhaWjJkyBBjCw6gT58+ALRr1478/HxycnIqfP4nn3ySP//8k8DAQD788ENGjBiBk5PTXTm/hYUF7u7ubN26FYDY2FjjtwOAhx56iIiICHbv3s2SJUuIiIgoty7Ozs6lyszMzAgPD2fVqlUoisJLL71Uapvjx4/TrFkzOnbsCMCjjz7KE088weHDh295rpvPGR0dTUxMDMOGDSMjI8PYxaHRaIiIiODUqVMsW7aMefPmoSgKf//9NwCNGjWiTZs2ALRt25br168DsH//fvz9/QFwcnKia9euFYrzscceo2HDhmi1Wpo0aYKLiwtarZamTZuSl5dnPO8/TZkyhejoaOPi5uZGTk4OP//8MwEBAQDUrVu3xDcFc3NzOnXqBMCvv/5KWloar7zyCr6+vowZMwaNRsPZs2fp2bMn3333HaNHj2bdunW8/vrrxm+k4t6T5F6NOnTowF9//UVWVlaJ8pSUFMaMGUNubi4GgwGNRmNcZzAYjN0iAFZWVgDGbRQTUwPdfKG2adOmfP/994wZM4asrCxGjhzJzp07S2xflecfMGAAW7Zs4eeff+bhhx/G1tbWuC45OZkBAwaQmJjIk08+yeTJk8utR506dcosT0xMxMrKiosXLxoT6M30en2J+hTHfHOdTNFqtYwfP57GjRsTHBwMQE5ODn5+fpw6dYq2bdsyZcoUzM3Nje9HrVq1jPtrNBpj+c3/hqJEWpE4LS0tS6wr3u92GAyGUp/bzZ+zpaVlibiaN29e4hfEunXrcHFxoUOHDuzYsYPnnnuOxMREBg8ezMmTJ287LlG1JLlXowYNGuDj40NoaKgxwWdlZfH2229ja2tLrVq1cHFx4fPPP0dRFPLz81m/fj09evSo1Hns7OyMF7SKW84AX3zxBSEhIbi4uBAUFISLiwunT58usW9VnL9Yx44dyc3NZfHixfj5+ZVYd/LkSezs7Hj55ZdxcXEx9uPq9XrMzc3R6/Umf3FlZmYSFBTEvHnz8Pb2ZurUqaW26dSpE3/99RfHjx8H4I8//uDIkSN06dKl0vWZPn06+/btY/v27Vy4cIGsrCwmT55M7969OXToEPn5+Sa/mfXs2ZN169YBkJSUxKFDh6o8TlOsra3p2LEjkZGRQNF1lqioqDI/506dOnHhwgWOHDkCwJkzZ3B3dyclJYXw8HCWL19O3759mTp1Ki1atOCPP/6o8Ocn7i4ZLVPNpk+fzvLlyxkyZAhmZmbk5+fTt29fJkyYAEBYWBizZs3Cx8eHgoICevbsydixYyt1jrCwMN555x1sbGzo0aMH9vb2QFFL+vDhw3h5eVG7dm0aNmxIYGAgv/32W4l97/T8N/P19SUyMpKePXuWKH/qqafYsGEDHh4eaDQaunTpgp2dHRcuXMDJyYkOHTqg0+mMCehW9ezVqxcuLi506dKFQYMGERkZaexugKJfdO+99x4zZ84kNzcXjUbD3Llzefjhh0lISKhUXZo1a8bo0aOZO3cuW7dupVevXnh6emJpaUnLli1p0aIFFy5cKNXKvtn06dMJCQnB09MTR0dHWrdubTLOX375pVJxVkR4eDjvvPMOmzZtIj8/33gRNzExscR2dnZ2LF26lAULFpCXl4eiKCxYsIAmTZowYsQIgoOD8fb2xtLSklatWqHT6TAzMyvx+dWvX7/K4xemaRT59SqEEKoj3TJCCKFCktyFEEKFJLkLIYQKSXIXQggVktEylZSbm8vJkyext7e/rXlchBDlK55Oon379iXuF6is4ukiKsLa2rrEfRhqIMm9kk6ePFliqJ0Q4u6IjIws887kisjIyMC561OYUbGb1erVq8d3332nqgQvyb2SiseMJ1s6o9fefqtCVK9jW2bc6xBEBaUkJzNyeIDxZ+12ZGVlYUYhKbW6UKgp/+fUXMmF64fJysqS5P4gK+6K0WtrodfWvsfRiIpq3LjJvQ5BVFJVdHsWamuh15Y9dYVR+TcV11iS3IUQ6qXRFi2mtlEhSe5CCPXSaIoWU9uokCR3IYR6acxAa6J7R1HnqDdJ7kII9dJoKtAtIy13IYSoWaRbRgghVEguqAohhApJy10IIVRIWu5CCKFCWm0FRstIchdCiBqmAi13lU6OK8ldCKFeWk3RYmobFZLkLoRQL+lzF0IIFZLRMkIIoUIVuaCqlZa7EELULNItI4QQalSBbhmkW0YIIWoWabkLIYQKaajABdVqiaTaSXIXQqiXtNyFEEKFtBV4WIep9TWUJHchhHpJy10IIVRI+tyFEEKNZOIwIYRQH5l+QAghVOgB7nNXZ62EEALQaLUVWipr/vz5BAcHA3DmzBn8/f1xd3dn6tSpFBYWApCUlERAQAAeHh6MGzeO7OxsADIzMxkzZgyenp4EBASQlpYGQH5+PkFBQXh6euLn58e5c+cAUBSF+fPn4+HhgZeXF0ePHq1QjJLchRCqVdQrozGxVO6YBw4cYPPmzcbXQUFBTJs2jW3btqEoCuvXrwdgxowZDB06lLi4ONq3b8/y5csBWLJkCc7OzsTGxjJ48GBmz54NwNq1a6lduzaxsbGEhoYSEhICwLZt2zh37hwxMTF88MEHhISEGH+BlEeSuxBCvTQVXCooIyODxYsXM3bsWAASExPJzc2lU6dOAPj7+xMXF0dBQQFHjhzB3d29RDnArl278PHxAcDb25s9e/ZQUFDArl276N+/PwCdO3cmPT2dpKQkdu/ejZeXF1qtlocffpiGDRvyyy+/mIxV+tyFEKpV3Do3tQ1AcnJyqXU2NjbY2NgYX0+bNo1XX32Vy5cvA5Camoq9vb1xvb29PSkpKVy7dg1ra2vMzc1LlP9zH3Nzc6ytrUlPTy/zWMnJyaSmpuLg4FCq3BRJ7kII1dJQgeT+36Z7QEBAqXXjx49nwoQJAHz99dc0bNiQ7t27s2nTJgAMBkOJ4yuKgkajMf5d4jy3iENRFLRabal9isvLOoe2AtcJJLkLIVRLq9WimEiExYkyMjISR0fHEutubrXHxMSQlpaGr68v169fJycnB41GY7wgCnDlyhUcHByws7Pjxo0b6PV6zMzMSEtLM7a+HRwcuHLlCo6OjhQWFpKdnY2trS0NGjQgNTWVZs2alTiWo6Mjqamppc5hsu4mtxBCiJqqEn3ujo6ONGnSpMRyc3Jfs2YNW7duJTo6mokTJ9K7d2/mzp2LlZWVcQRLdHQ0rq6uWFhY4OzsTExMDABRUVG4uroC4ObmRlRUFFD0C8PZ2RkLCwvc3NyIjo4G4KeffsLKyopGjRrh6urKN998g16v58KFC8THx/PYY4+ZrLq03IUQ6lWBPvc7vYkpPDycsLAwsrKyaNeuHcOHDwdg+vTpBAcHs2LFCho2bMiiRYsAmDRpEsHBweh0OurWrUt4eDgAgYGBTJs2DZ1Oh6WlJQsWLADAw8OD48ePGy+2zp49m1q1apmMS6MoinJHNXvAJCQk0KdPHxJruaDX1r7X4YgKunZk2b0OQVRQYmICXv36sGPHDpo0aXJbxyj+Ob3SYSIGK9tyt9XmZfDv40vv6Hz3I2m5CyFUqzKjZdRGkrsQQrWKb2IytY0aSXIXQqiWRqNBo5WWuxBCqIp0ywghhApJchdCCDWqyNwx6sztktyFEOolLXchhFCjariJ6X4lyV0IoVpajem5ZTQqfRKTJHchhHpJn7sQQqiPRoPJbheV9spIchdCqJem6BZV09uokCR3IYRqSXIXNZaXWwciZgynWa83SpS7dWnFzEl+xte1rCx41KkBvQLnc+y3S7d1rtpWFiwNC+CxVk3QajW8/X40MbuPAzD6WTdeH+VO6tVMALKyc/Eas+Q2a6VOYYs3EbXjF+rb1AHgUacGrJ47qsQ2B349R+iiDRQWGqhlZcGCoME83tbpts+Zk5vPxFmRnDibgMGg8PYEX3S9OgIQtf1nwlcXPdT5X7bWLA4ZQvNmph8CUZNotBowNf2AifU11V1J7sXTba5evZqnnnrKWN67d28+++yzCk+r2apVK1q3bl2i7J133qFjx45VGm9l47pfPNLUnpmT/Mpseew+fBbXgHnG15/Me5GtPxy77cQOEDzGi6y/8+j27CyaNKjPttWv8+uZiySlZtC1w8OELd7Ehm0/3fbx1e7wib/4ePZIunZ85JbbvDTtU5a9FYBr51Zs/eEYL7+9lgPrw277nPM+jMG6thWHvn6LS8npuI96l05tmmFuZsarc79i7xchNG5Qnw/X72bKwvVsfH/8bZ/rfiQt97vAwsKCt956iy1btmBtbX3bxyl+MokoqbaVBSvfGUHYkk18OPOFcrd91rMzzRr+ixenrjGWvT7SHZ/endBqNVxMusob89eTfOW6cf2bo70AmL8qxlim69WR0W99AkBCyjV2Hf6NAX2fYPkXO+nS4RHqPlSLySOeIflKJtPe28zpc0lVVt+aLi+/gONnE1i6djvn56bRopkDs18bSFNHuxLb6fUGMm7kAJCVk4uVlYVxXfjqOL7Z+SsGg0KzRv8i/M1naWj/v7nK5334LQDBY3TGsm93HWPVzBcAaOpoR68urYna/jOvBPTh921zsTA3o7BQz6XL6dSv99Bdqv29VIFx7iodLnPXkruDgwM9evRg/vz5zJw5s9T6iIgItmzZgpmZGU899RRBQUGYmZlV6Njvv/8+v/76K5cvX2bYsGG0aNGCxYsXk5ubS2ZmJiEhIfTt25fg4GC6dOmCv78/UPRN4OzZs2RkZBAUFERycjLNmzcnLy+vzPNkZmaSmZlZoqwiTx2vDotDn+eTTXs5+UdiudtZmJvx1ss+jA77BL3eAMBzXl1o26IRfV5YiF5vYITfUywNG8qzk1eUe6zGDeqTmHLN+DopJYNGDrbUqWXJ7/EpvPfZ9+z/5RwD+j7O+vfG0XXwTLL/zr/zyqpActp1ejq3ZOpYb9o0b8j7n+8g4PUP2f35myWSz7JpAQx7YxUh727k+o0cNi0rakl/9e0hTv+ZxI5PgjA3N+OTTXuZOOsLvn7v5XLPm5hyjcYN6htfN2pgS1JqBlD0f+OX0xcY8moEf+cVsGFp+ceqiSrSclfrcJm72uceHByMj48P+/btK9E9s3v3bnbu3MnGjRuxsLBgwoQJfPXVV2U+fdzX19f4765duxIaGgpAfn6+8fmEEydOZNasWTRv3pwDBw4wZ84c+vbte8u4li5dStu2bVm1ahVHjhwhNja2zO0+/fRTli27/57g8+KgnhTqDUR+c5CmDe3K3da3z+PEJ17l4LG/jGXuLu15sp0TP3w2BQAzrZbatYpaiGsXjMap8b9w+FfRsyN1vTpwIfEqgVNWodVquPm5XRqNBoPBQE5uPoMmfmAsj9r+C0EvevJ4Wyf2Hv2jqqpdozk1/neJRDxhWB/CP47jYtJVnBr/G4DUq5lMmv0lW1dO4vG2Tny76xgvBH/MkY3T2Lb3JEdPXeDp4UWPXtMbDPydWwDAsKAPuZB41Xi949tdx3Fq/C8+XzgGg0EpkbsURTE+EBrg8bZOnN02l+37T/PcqxH8GvU29erWudtvR/WRce53h7W1NTNnzjR2zxQ7ePAgOp2O2rWLHlM3cOBAoqKiykzut+qW6dChg/HfCxcu5IcffiAuLo5jx46RnZ1dblyHDx/m3XffBaBz5840bdq0zO1GjBiBn59fibLk5OQy46xOz3t3pU4tS/ZEBmNhbkZtKwv2RAbz7KQVJbpWAPyeeYLIbw6WKDMz0/LeZ9+zeuNeACwtzLG1KfosAqesAsrulklIvobjv+uRln4DAEf7epz4PYGmjvXxcO3AqvW7jdtqNFBYqK/imtdcJ/9I5OQfiQzx6mIsUxQFc/P/fVs98Os5mjraGS+g6np1JGTRRn4/n4xeb2DS8Gd4cVBPoKibJyPzbwA+XzgGKLtbpoljfZKvXDf+sk5Ou85jLZtwOS2D038m0ad7WwD69mhL3YdqcT7hCp3aNLtbb0O1e5Bb7nf9vlsXFxdj90wxg8FQarvCwsJKHffmB8QOHTqU48eP0759e8aOHWss12g0FD8itqCgoMxy4JbdQTY2NqWehu7o6FipOO+Gvi+E02PIHFwD5vHs5BX8nVeAa8C8UokdoMfjLdhz5GyJsp0HzxDo24O6DxW9h6Ev6YiYMcLkeWP2HOcFv6JvYI0cbOnTvQ3bfjxJ9t/5hI3z5on/JqVnerSldi1Ljp66cKdVVQ2tRkNw+NdcSLwCwMcbfqRti8YlukzatWjEmb+S+PNCCgA/nYzn79x8mjdzoHe3NqyN3k9mVlFCn7PyW8ZO/9Tkeb1cO/DJ5n1AURfNjgNncO/Znry8Ql6cuoa/LqUB8ONPv6PXG2j58L3//12VtFrQajUmlnsd5d1RLUMhi7tn0tKK/iN169aNFStW8Nxzz2Fubs7GjRvp1q3bbR07IyOD+Ph4vvjiCywtLQkPD0evL2ox2tra8ueffwKwfft24z7du3cnOjqa1q1bc/z4cS5evHiHNbw/OP67HuvfG2dswf/L1pqH6lgZ+1iLfRa1n4b29fhu9esoQGJyOq/MWFtim5tb7MXmrfyWd4OHsH/dVMy0GqYvjSL+v8lqZMhqFocOwcLCnBtZuQQGraJAWu5GbVs0Yv4bgxny2koMBgONHGz5aPYLXE7L4NlJK1j/3jhaODVgUfAQhr/5ERqNhjq1LPlswWhsrGszfEAPLqddp9+od9EAjR3t+GB6YIlz3NxiN5a9pOP1eV/R/dlZ6A0KMyYO4OEm9gAsDRtadC6gXt3afLnoJerUsqyGd6P6PMgtd41ycxO2iiQkJDB8+HB27txpLNu7dy8vvvii8Qnjy5cv59tvv6WwsBAXFxdCQkIwNy/5u6b4Aug/vf/++wBMmDABgLlz57Jjxw7Mzc3p1q0bsbGx/PDDD1y5coXJkydTUFBgLN+7dy9ZWVkEBwdz/vx5HnnkEc6ePcvq1asrNBSyeJhnYi0X9Nrad/I2iWp07cj9d+1ElC0xMQGvfn2MueJ2FP+cKs9Mgzr/Kn/jnKtovn/njs53P7oryV3NJLnXTJLca46qTO48Mx0eMpHcs6/C9zNUl9zlDlUhhHpVoFdGUWevjCR3IYR6abUak9MLKFoNpYd41HyS3IUQqiXJXQghVKgig2XkJiYhhKhhKvKAbLUOhZTkLoRQMdPJXVFp012SuxBCtR7ge5gkuQsh1Kt4ioHyN1JndpfkLoRQraKWuzwgWwghVEW6ZYQQQoUqMlpGHrMnhBA1jLTchRBCleQZqkIIoToVGS2jyGgZIYSoWR7kbhmVPmBKCCH+d0HV1FIZ7733Hl5eXuh0OtasWQPA/v378fHxoV+/fixevNi47ZkzZ/D398fd3Z2pU6caHyealJREQEAAHh4ejBs3zvjc58zMTMaMGYOnpycBAQHGp9fl5+cTFBSEp6cnfn5+nDt3zmScktyFEKpV3HI3tVTU4cOHOXjwIFu2bGHjxo2sXbuW3377jdDQUJYvX05MTAwnT55k9+6ih8UHBQUxbdo0tm3bhqIorF+/HoAZM2YwdOhQ4uLiaN++PcuXLwdgyZIlODs7Exsby+DBg5k9ezYAa9eupXbt2sTGxhIaGkpISIjJWCW5CyFUq6pb7l26dOGzzz7D3Nycq1evotfryczMxMnJiaZNm2Jubo6Pjw9xcXEkJiaSm5tLp06dAPD39ycuLo6CggKOHDmCu7t7iXKAXbt24ePjA4C3tzd79uyhoKCAXbt20b9/fwA6d+5Meno6SUlJ5cYqfe5CCNWqzDj35OTkUutsbGywsbEpUWZhYcHSpUtZvXo1Hh4epKamYm9vb1zv4OBASkpKqXJ7e3tSUlK4du0a1tbWxmdGF5cDJfYxNzfH2tqa9PT0Mo+VnJxMo0aNblkvSe5CCNXSaEyPljH8N7kHBASUWjd+/HgmTJhQqnzixImMHj2asWPHEh8fX+IXiKIoaDQaDAZDmeXFf/8zzrIoioJWqy21T3F5eSS5CyFUqzKjZSIjI3F0dCyx7p+t9nPnzpGfn0+bNm2oXbs2/fr1Iy4uDjMzM+M2aWlpODg44OjoaLwgCnDlyhUcHByws7Pjxo0b6PV6zMzMjNtDUav/ypUrODo6UlhYSHZ2Nra2tjRo0IDU1FSaNWtW4ljlkT53IYRqFU8cVv5StK2joyNNmjQpsfwzuSckJBAWFkZ+fj75+fns2LGDIUOGcP78eS5cuIBer2fr1q24urrSuHFjrKysOHr0KADR0dG4urpiYWGBs7MzMTExAERFReHq6gqAm5sbUVFRAMTExODs7IyFhQVubm5ER0cD8NNPP2FlZVVulwxIy10IoWJVPc7dzc2N48ePM2DAAMzMzOjXrx86nQ47OzsmTJhAXl4ebm5ueHh4ABAeHk5YWBhZWVm0a9eO4cOHAzB9+nSCg4NZsWIFDRs2ZNGiRQBMmjSJ4OBgdDoddevWJTw8HIDAwECmTZuGTqfD0tKSBQsWmK67oihKWSsyMjLK3dHW1rbi74iKJCQk0KdPHxJruaDX1r7X4YgKunZk2b0OQVRQYmICXv36sGPHDpo0aXJbxyj+OW00IhxzG/tyty3MTCPp0zfu6Hz3o1u23Lt162bs/P8njUbDmTNn7mpgQghxpzQVmH5A86BNP/Dbb79VZxxCCFHltJh+0JJaLzyarJfBYODjjz8mODiYrKwsVq5ciV6vr47YhBDijtyN6QdqCpMXVBcsWEB6ejonTpxAURR+/PFH0tLSCAsLq474hBDitsnEYeU4cOAA8+bNw8rKirp167J69Wr27dtXHbEJIcQd0VTwjxqZbLmbm5uXuBPK0tLSeNusEELcz7SaCvS5qzO3m07uLVu2JDIyEr1ez19//cUnn3xC69atqyM2IYS4Iw/yaBmT3TJTp07l1KlTXL16leeff57s7GxCQ0OrIzYhhLgjWo2mQosamWy5W1tbM2fOnOqIRQghqpRcUC3H1atXee211+jatSsuLi6EhoaSmZlZHbEJIcQd0VCBoZAqvaBqMrmHhYXRtGlTNmzYwOeff069evWYNm1adcQmhBB3pKqfxFSTmOyWSUxMZMWKFTa4DOsAAB+eSURBVMbXb775pvFJIUIIcT/TasDMRPZW6fVU0y13BwcHLl26ZHydnJxc4okgQghxv5I7VMswduxYANLT0xkwYAA9evRAq9Vy6NAhWrVqVW0BCiHE7ZJx7mUofnjrP/Xq1etuxSKEEFWqMs9QVZtbJnc/P78yyxVF4cKFC3ctICGEqCoP8lBIkxdUv/rqKxYsWMDff/9tLLOzs5P5ZYQQ9z1puZfjww8/ZM2aNaxYsYLJkyfzww8/kJycXB2xCSHEHdFqNJiZ6FRX6x2qJkfL2Nra0rFjR9q0acPVq1cZN24cR44cqY7YhBDijmgquKiRyeRubm7O9evXcXJy4vjx4wDysA4hRI3wIM8tYzK5P/vss7z00kv06tWLdevW4e/vzyOPPFIdsQkhxB2RO1TLMWjQILy8vKhTpw7r1q3jxIkT9OzZszpiE0KIOyIXVMuwZs2aW+70xRdfMHLkyLsSkBBCVJmKtMzVmdtvndx///336oxDCCGqnJnW9GgZU+trqlsm97lz51ZnHDXOsS0zaNy4yb0OQ1RQXoEMAqgp8gsMVXYsDaa7XdSZ2ivQ5y6EEDWVFtOjRkyOKqmhJLkLIVRLLqgKIYQKaSowK6RKc7vpbyQGg4GPPvqIN998k6ysLFauXCk3MQkhaoTiC6qmFjUy2XJfsGAB6enpnDhxAoAff/yRtLQ0wsLC7npwQghxJx7k+dxNttwPHDjAvHnzsLKywtramtWrV8uMkEKIGkHuUC1vA3NztNr//Q6wtLTE3Fy66oUQ9z9NBeaOeWAvqLZs2ZLIyEj0ej1//fUXn3zyCa1bt66O2IQQ4o48yEMhTdZr6tSpnDp1iqtXr/L888+TnZ1NaGhodcQmhBB3pOgmJhPLvQ7yLjHZcre2tmbOnDnVEYsQQlQpmX6gHLNmzSqzXEbLCCHudzLOvRy2trbG5aGHHuLw4cPVEZcQQtyxB/lhHSZb7uPHjy/xevTo0YwbN+6uBSSEEFWlIkMdVZrbK3+h2NramtTU1LsRixBCVKnim5hMLRW1bNkydDodOp2OBQsWALB//358fHzo168fixcvNm575swZ/P39cXd3Z+rUqRQWFgKQlJREQEAAHh4ejBs3juzsbAAyMzMZM2YMnp6eBAQEkJaWBkB+fj5BQUF4enri5+fHuXPnKlZ3UxvMnDmTWbNmMWvWLGbOnMmQIUPkMXtCiBpBU8E/FbF//3727t3L5s2biYqK4tSpU2zdupXQ0FCWL19OTEwMJ0+eZPfu3QAEBQUxbdo0tm3bhqIorF+/HoAZM2YwdOhQ4uLiaN++PcuXLwdgyZIlODs7Exsby+DBg5k9ezYAa9eupXbt2sTGxhIaGkpISEiF4jWZ3OvXr2/sc69fvz79+/dn4cKFFTq4EELcS2YaMNeWv5hVsOVub29PcHAwlpaWWFhY0Lx5c+Lj43FycqJp06aYm5vj4+NDXFwciYmJ5Obm0qlTJwD8/f2Ji4ujoKCAI0eO4O7uXqIcYNeuXfj4+ADg7e3Nnj17KCgoYNeuXfTv3x+Azp07k56eTlJSksl4Tfa5X7x40fj1QwghapLKTPmbnJxcap2NjQ02NjYAPProo8by+Ph4YmNjGTZsGPb29sZyBwcHUlJSSE1NLVFub29PSkoK165dw9ra2niXf3E5UGIfc3NzrK2tSU9PL/NYycnJNGrUqNx6mUzuv/32G4qiqPYWXSGEelVm4rCAgIBS68aPH8+ECRNKlP3xxx+89NJLTJkyBTMzM+Lj443rinOlwWAokTOLy8vKpbfKrYqioNVqS+1TXG6KyeRub2+PTqejY8eOPPTQQ8ZyGecuhLjfVWa0TGRkJI6OjiXWFbfaix09epSJEycSGhqKTqfj8OHDxgufAGlpaTg4OODo6Fii/MqVKzg4OGBnZ8eNGzfQ6/WYmZkZt4eiVv+VK1dwdHSksLCQ7OxsbG1tadCgAampqTRr1qzEsUy5ZfrPz88H4PHHH8fLy4vGjRuXGPMuhBD3u6KbmMof416c3B0dHWnSpEmJ5ebkfvnyZV555RXCw8PR6XQAdOzYkfPnz3PhwgX0ej1bt27F1dWVxo0bY2VlxdGjRwGIjo7G1dUVCwsLnJ2diYmJASAqKgpXV1cA3NzciIqKAiAmJgZnZ2csLCxwc3MjOjoagJ9++gkrKyuTXTJQTsv9ueeeY/PmzaXGuQshRE1hpi1aTG1TER9//DF5eXnMmzfPWDZkyBDmzZvHhAkTyMvLw83NDQ8PDwDCw8MJCwsjKyuLdu3aMXz4cACmT59OcHAwK1asoGHDhixatAiASZMmERwcjE6no27duoSHhwMQGBjItGnT0Ol0WFpaVvga6C2Tu6IoFauxEELcp7Ro0JoY6mhqfbGwsLBbdkdv2bKlVFnr1q3ZsGFDqfLGjRuzdu3aUuW2trZERESUKreysmL+/PkVivFmt0zueXl5nD59+pZJvl27dpU+mRBCVKcH+Q7VWyb3S5cuMWHChDKTu0ajYceOHXc1MCGEuFMaKjBxWLVEUv1umdxbtGhh7NwXQoiaqCITgz2wE4cJIURNJd0yZXB2dq7OOIQQosrJwzrKIDcpCSFqOg2mJ9BSZ2qXbhkhhIpVZm4ZtZHkLoRQLQ2mW+bqTO2S3IUQKiajZYQQQoWk5S6EECqk0WjQmhgNI33uQghRw2gxPVqm0g+SriEkuQshVEtGywghhApJn7sQQqhQ0fQDplru1RRMNZPkLoRQLa1Gg5kMhRRCCHWRbhkhhFAhmRVSCCFUqCofs1fTSHIXQqiWtNyFEEKFNP/9Y2obNZLkLoRQLRktI4QQKiTdMkIIoUIaKpDcqyWS6ifJXQihWtLnLoQQKqTVFC2mtlEjSe5CCNXSVOBJTGqdFVKtUxk/ML7ddYymbq+Xue5ScjrPvboC14B59Bgymx0HTt/RufR6AyHvbqDLoJk84fc2qzf+aFz340+/8/Tw+bgMncszI8M5eir+js6lVoqiMGHm5yz/YmeZ6z/6eg9PDZlNnxELGDvtU65lZt/R+a5cy+L51yLoOXQObgFzOXLifKltYncfp3nfKXd0nvuVpoJ/1Oi+brknJCTg4eFB8+bNS5RHRETQsGHDKjtPq1atOHv2bJUdr7qcu5jKW+9tRlGUMtc//9pKRvq78OKgnhw/e4n+45ZyNm4OVpYWt3W+NZv28ufFVPZ/FUpWTh79Rr1Lx9ZNeaxlE0aFrmbj+6/QoVVT4n48wdhpn3Fk47Q7qZ7q/B6fTEj4Bn4+fYE2zRuVWr/36B988Pl2vl31Go0cbPk69ghvzFvHx3NG3fY5Q979mm4dH2HSorGc/D2BYW98yP71YdSpZQnAX5dSmbEs+pb/h2o66Za5jzk4OBAdHX2vw7jv5OTm89K0T5k12Z8xb31Sav2Jswlcu57Ni4N6AtChVVNiV72GVlv0ZW1t9H4+3vAjBoOCXb2HWDBlMC3/z9G4/xffHGTv0T9Y/nagsezbXccY4fcU5uZm2NrUwb/fE6yPPcKT7f6P0zGzsTA3Q1EULiRepX69h+7uG1ADrdm4l4D+3WjsWL/M9cfPXqJn51Y0crAFwKtXB16f9yX5BYUAzFq+hQO/nENvMNC+ZRNmvzqQug/VMu4/cVYkPR5vwRBdVwAKC/Vs33eKua8PBqB9yyY83NSeHw6eQderIzm5+bwy43NmTBzAuLc/u5tVv2eKJg4zdUFVne775F6W4OBgMjIyuHDhAkFBQeTl5bFmzRpyc3PJz89nzpw5PPHEEwQGBjJ+/Hi6du1KQkICw4cPZ+fOnSQkJBAUFEROTg4dO3a819W5La/O+ZIX/F1o/2jjMtf/eTGVZo3+xdTFGzl07C/MzMwIGeNFm+YN2Xf0D7789hAxq16lTi1Ldh48Q2DQKg59/Va550xMyaBxg/8lpkYO9Tn1RxIAFuZmpF7NpFfgfK5mZLN6zsiqq6xKzH19EAC7Dpf9LfGJtk58/PVuLl1Op2lDO7769hD5BXquXc/m8y0HMDMz47s1b6DRaJgT8Q2zlm9hftCztzxf+vVsDIrCv+tbG8sa2dfjcmoGAFPmryNwQA/atCj9LUItZJz7fSw1NRVfX1/jax8fHwBsbW2JiIjAYDAwcuRIIiIisLOzY8OGDXz44YdERETc8pgzZ87E39+fwYMHExUVxbp168rcLjMzk8zMzBJlycnJVVCrO/PR13swN9MyrH93LiZdLXObgkI9h479xYRhfZj96kCOnopn8KTl7PsylG37TnH+Uhruo941bp+RmcO169nMjtjKoWN/cS0zh+y/8+g5dC5WluZs/yQIg2IocfFJURTMzP532cbhXzacjpnNsd8u4fvyUlo97EgLpwZ3741QmW6dmvP6KA9GhXyMRqvheV1X6tvUwcLCnO/3nSIz62/2HCn6xZBfUMi/69cFwPM/i8gvKCQx+Rr7jv7OqvW76fzYw0we0a9Uq1UBtGZa1mz8ETMzLUO9u3Hxctn/h9RApvy9j5XVLRMcHEyHDh0A0Gq1fPDBB+zcuZPz589z+PBhY9fDrRw+fJh33y1KbP379ycsLKzM7T799FOWLVtWBbWoWl9uPURObj49h86loFDP33kF9Bw6l/XvjaOhfdFX+ob29ahXtzZebkXv05Pt/o//a/RvTv6eiEFv4FmvLsyYMAAAg8HA5bTr2NrUIfzN54Cyu2WaNLAjOe268XXyles0crDletbf/Hjkd7yfLvoW1LF1U9o/2pjT55IkuVdCVnYu3R9vwVCf7gBcTstgwaoY6tvUwWAwMHOyP326twUgOyeP3PwCAGI/eg0ou1tGQeFaZjb1bYq6yZKvXKeRvS1LPv2Ov3Pz6TNiAfkFheTmFdBnxAIiw1/C0b5edVf9rnmQpx+osaNlatUq6mvMzs5m0KBBJCQk0LlzZwIDA0tsV3yhqLCwsMxyjUZzy18GI0aMYMeOHSWWyMjIqq5Kpe34NIgD66by4xchrF8yjtpWFvz4RYgxsQN06fAwtSzNifvxBFB0Me984hXaPdqY3t3bsHHbUZKvFCXq1Rv3MuDl902e18vtMT7fcoDCQj3Xb+Sw6buj6Nw6YKbVMn7m5xw8dg6AM+cu80d8Ck+2+7+qr7yKJV+5jv8r73MjOxeAJZ98x4BnnkSj0dCra2tWb/iR/IJCDAYDr8/7ijkrtpZ7PHNzM/r2aMvnUfsBOP1nIr+fT6bHEy2I+/h1dkeGsOPTKUS++xK1rCzY8ekUVSV24H9Nd1OLCt33LXdT4uPj0Wg0jB07FkVRmDJlCnq9HoD69evz559/0q1bN7Zv327cp0ePHmzZsoWAgAC+++478vLyyjy2jY0NNjY21VKPqnA5LYNnJ60wtuA3vP8Kby78mhnLtgCw7K0AGjnY0sjBlknD++L3yjK0Wg11H6rFZwtGl+hyGerTjaE+3Uocf9TAnpxPuELPoXPJL9Tzgt9TPPXkowBELhxD6LsbKSjUY2VpzqpZL5Tonxdl+/XMRV6f9xU7Pp1CC6cGjA/si9d/FmFQFLp0eJg5/+2nf3WkOzPej6bvCwsx6A20e7Qxb08cUOJYS8MCSh1/3huDeW3uV7gFzEWj0bBsWiA21rWrpW73gwf5DlWNch+Pgbr5IujNgoOD6dKlC/7+/uj1eoKCgjh16hQajQYXFxe2b9/Orl27OH78OMHBwVhZWdGnTx82bdrEzp07SUlJISgoiIyMDNq3b09cXBw///xzhWPq06cPMd/toHHjJnej2uIuyCvQ3+sQRAUlJSYyQPcMO3bsoEmT2/sZK/45XfRJFPYNyr9gnJaSxGsvDLij892P7uvkfj+S5F4zSXKvOaoyuS+uYHJ/VYXJvcZ3ywghRLnU2etiUo29oCqEEKZo/zu3jKmlMrKysvD29iYhIQGA/fv34+PjQ79+/Vi8eLFxuzNnzuDv74+7uztTp041DupISkoiICAADw8Pxo0bR3Z20RQTmZmZjBkzBk9PTwICAkhLSwMgPz+foKAgPD098fPz49y5cxWre6VqJYQQNUhVD5Y5duwYzz//PPHx8QDk5uYSGhrK8uXLiYmJ4eTJk+zevRuAoKAgpk2bxrZt21AUhfXr1wMwY8YMhg4dSlxcHO3bt2f58uUALFmyBGdnZ2JjYxk8eDCzZ88GYO3atdSuXZvY2FhCQ0MJCQmpUKyS3IUQ6lXF2X39+vVMnz4dBwcHAI4fP46TkxNNmzbF3NwcHx8f4uLiSExMJDc3l06dOgHg7+9PXFwcBQUFHDlyBHd39xLlALt27TLepOnt7c2ePXsoKChg165d9O/fH4DOnTuTnp5OUlKSyVilz10IoWIVmfWxaH1Zd5//czh0cWu6WGpqKvb29sbXDg4OpKSklCq3t7cnJSWFa9euYW1tjbm5eYnyfx7L3Nwca2tr0tPTyzxWcnIyjRqVf6FYkrsQQrUqM7dMQEDp+wTGjx/PhAkTbrmvwVB6Sg6NRnPL8uK/S56/7AAVRUGr1Zbap7jcFEnuQgjVqkxyj4yMxNHRscQ6UzcxOjo6Gi98AqSlpeHg4FCq/MqVKzg4OGBnZ8eNGzfQ6/WYmZkZt4eiVv+VK1dwdHSksLCQ7OxsbG1tadCgAampqTRr1qzEsUyRPnchhGpV5mEdjo6ONGnSpMRiKrl37NiR8+fPc+HCBfR6PVu3bsXV1ZXGjRtjZWXF0aNHAYiOjsbV1RULCwucnZ2JiYkBICoqCldXVwDc3NyIiooCICYmBmdnZywsLHBzczPOr/XTTz9hZWVlsksGpOUuhFCxuz3lr5WVFfPmzWPChAnk5eXh5uaGh4cHAOHh4YSFhZGVlUW7du0YPnw4ANOnTyc4OJgVK1bQsGFDFi1aBMCkSZMIDg5Gp9NRt25dwsPDAQgMDGTatGnodDosLS1ZsGBBhWKT5C6EUK27NeXvzVOidO/enS1btpTapnXr1mzYsKFUeePGjVm7dm2p8uJpzP/JysqK+fPnVzpGSe5CCPV6gCd0l+QuhFCtB3lWSEnuQgjV0lTgAdkqfVaHJHchhMqpNHmbIsldCKFa0i0jhBAqdLeHQt7PJLkLIVTrAR4sI8ldCKFiD3B2l+QuhFAtrQaTD+MwNZqmppLkLoRQrQe44S7JXQihYg9wdpfkLoRQLRkKKYQQalSBoZAqze2S3IUQ6iXj3IUQQoWkW0YIIVRIWu5CCKFCD/BgGUnuQgj10lCBlnu1RFL9JLkLIVTswW27S3IXQqiWtgIP65DpB4QQoqaRce5CCKE+MhRSCCHU6MHtcpfkLoRQrwc4t0tyF0Kol9zEJIQQKqTRaNCYyN6m1tdUktyFEKol3TJCCKFC0i0jhBAqJEMhhRBCjeQmJiGEUB+ZOEwIIVRIumWEEEKF5IKqEEKokAyFFEIINXqAs7skdyGEahXldlN97uokyV0IoVrysA4hhFAj6ZYRFaXX6wFISU6+x5GIysgvMNzrEEQFpaQU/WwV/6zdidSUFExl76Jt1EeSeyWlpaUBMHJ4wD2ORAh1S0tLw8nJ6bb2tba2pl69ehX+Oa1Xrx7W1ta3da77lUZRFOVeB1GT5ObmcvLkSezt7TEzM7vX4VSZ5ORkAgICiIyMxNHR8V6HIypArZ+ZXq8nLS2N9u3bU6tWrds+TkZGBllZWRXa1traGltb29s+1/1IWu6VVKtWLZydne91GHeNo6MjTZo0uddhiEpQ42d2uy32m9na2qouYVeG9l4HIIQQoupJchdCCBWS5C6EECokyV0AYGNjw/jx47GxsbnXoYgKks9MlEdGywghhApJy10IIVRIkrsQQqiQJPcaJiEhgVatWrFv374S5b179yYhIaHCx2nVqhW+vr4llmPHjlV1uJWO60GVkJBA+/btS30mly9frtLztGrVqkqPJ+5fchNTDWRhYcFbb73Fli1b7uiW6ejo6CqMStwpBwcH+UxElZHkXgM5ODjQo0cP5s+fz8yZM0utj4iIYMuWLZiZmfHUU08RFBRU4akS3n//fX799VcuX77MsGHDaNGiBYsXLyY3N5fMzExCQkLo27cvwcHBdOnSBX9/f6CoRXj27FkyMjIICgoiOTmZ5s2bk5eXV6V1f9AEBweTkZHBhQsXCAoKIi8vjzVr1pCbm0t+fj5z5szhiSeeIDAwkPHjx9O1a1cSEhIYPnw4O3fuJCEhgaCgIHJycujYseO9ro6oRpLca6jg4GB8fHzYt28fTz31lLF89+7d7Ny5k40bN2JhYcGECRP46quvCAgoPYGSr6+v8d9du3YlNDQUgPz8fGJiYgCYOHEis2bNonnz5hw4cIA5c+bQt2/fW8a1dOlS2rZty6pVqzhy5AixsbFVVWXVS01NLfGZ+Pj4AEW30UdERGAwGBg5ciQRERHY2dmxYcMGPvzwQyIiIm55zJkzZ+Lv78/gwYOJiopi3bp1d70e4v4gyb2Gsra2ZubMmcbumWIHDx5Ep9NRu3ZtAAYOHEhUVFSZyf1WXQAdOnQw/nvhwoX88MMPxMXFcezYMbKzs8uN6/Dhw7z77rsAdO7cmaZNm1a6bg+qsrplgoODjZ+HVqvlgw8+YOfOnZw/f57Dhw+j1ZZ/2ezmz6N///6EhYXdneDFfUcuqNZgLi4uxu6ZYgZD6XnLCwsLK3Xcm2fiGzp0KMePH6d9+/aMHTvWWK7RaCi+RaKgoKDMckBVM2feK8WfR3Z2NoMGDSIhIYHOnTsTGBhYYrvi9/2fn3dxuUajMfnLQKiHfNI1XHBwMHv37iU1NRWAbt268e2335Kbm0thYSEbN26kW7dut3XsjIwM4uPjmTRpEq6uruzYscP4AAVbW1v+/PNPALZv327cp3v37sbW5/Hjx7l48eKdVE/cJD4+Ho1Gw9ixY+natSvff/+98fOoX79+mZ9Hjx49jN/svvvuO7kG8gCR5F7DFXfPFLeen376aXr16sXAgQPR6XQ0atSIYcOG3daxbW1tGTRoEDqdDk9PT7Kzs8nNzSUnJ4fnn3+eQ4cO4ePjw88//4y9vT1Q1Ed/6dIldDodq1atkm6ZKtS6dWvatGmDp6cnOp2O+vXrk5SUBMB//vMfvvjiC/z8/MjNzTXuM23aNLZt20b//v3ZvXs3Dz300L0KX1QzmX5ACCFUSFruQgihQpLchRBChSS5CyGECklyF0IIFZLkLoQQKiTJXVSJhIQE2rRpU2JGw/79+7Nhw4Y7PvZLL73Epk2bgKIpEzIzM2+57Y0bNxg+fHilzxEXF1fqpiCAQ4cO4e3tbXL/Vq1akZ6eXqlzBgcH8/HHH1dqHyEqSqYfEFWmVq1aJW6fT0lJwdvbm/bt29O6desqOYepWROvX7/OiRMnquRcQtRkktzFXdOgQQOcnJyIj4/n9OnTbNiwgb///htra2vWrl3L119/zZdffonBYMDW1pa33nqL5s2bk5KSQnBwMKmpqTRq1IirV68aj9mqVSsOHDiAnZ0dK1euZPPmzZibm+Pk5MS8efMICQkhNzcXX19fNm3aRHx8PLNnzyYjIwO9Xk9gYCCDBg0C4L333uObb77B1tYWJycnk/U5f/4877zzDtnZ2aSlpdG6dWuWLFmClZUVAEuWLOHEiRMYDAYmT57M008/DXDLegpxVylCVIFLly4pnTp1KlH2888/K507d1aSkpKUjRs3Kp07d1Zu3LihKIqiHDp0SBk6dKiSk5OjKIqi/Pjjj4qHh4eiKIry8ssvK4sXL1YURVHi4+OVTp06KRs3blQURVFatmypXL16Vdm+fbvSr18/JSMjQ1EURZkzZ46yfPnyEnEUFBQoXl5eysmTJxVFUZTMzEzF09NT+eWXX5Tvv/9e8fLyUm7cuKEUFBQoY8aMUYYNG1aqXgcPHlR0Op2iKIoyb948JSoqSlEURcnPz1e8vb2VuLg4Y1wrV65UFEVRzp49q3Tp0kW5evVqufV88803lY8++ujO3nghbkFa7qLKFLeYAfR6PfXr12fhwoU0bNgQKGp1Fz9cZNeuXVy4cIEhQ4YY98/MzCQjI4P9+/fz5ptvAuDk5ETXrl1LnevAgQN4eHhQr149AEJCQgBKPPUpPj6eixcvGqcyLo7x9OnTnDt3jmeeecYYz8CBA1m7dm259QsKCmLfvn2sWrWK+Ph4UlNTycnJMa5//vnnAWjZsiXNmzfnl19+4ejRo7espxB3kyR3UWX+2ef+T3Xq1DH+22Aw4OvrS1BQkPF1amoq9erVKzWzpLl56f+mZmZmaDQa4+vMzMxSF1r1ej1169YtEdOVK1eoW7cuCxYsqPTsla+99hp6vR5PT0969erF5cuXSxzj5hkXDQYD5ubm5dZTiLtJRsuIe8LFxYVvv/3WOJvll19+yYgRIwDo2bOn8aESSUlJHDp0qNT+PXr04PvvvycrKwsoeoLUJ598grm5OXq9HkVRePjhh0v8wrl8+TLe3t6cPHkSV1dX4uLiyMzMxGAwVOjxdnv37uWVV17By8sLgGPHjhlnZQTYvHkzAKdOneLixYt07Nix3HoKcTdJy13cEy4uLowePZpRo0ah0WiwtrZm2bJlaDQapk+fTkhICJ6enjg6OpY50sbNzY0///zT2BXSokULZs6cSe3atenQoQM6nY7IyEiWL1/O7Nmz+eijjygsLGTSpEk8+eSTAJw9e5aBAwdiY2ND69atuXbtWrkxv/rqq7zyyivUqVMHa2trOnfuXGJK40uXLjFgwAA0Gg2LFi3C1ta23HoKcTfJrJBCCKFC0i0jhBAqJMldCCFUSJK7EEKokCR3IYRQIUnuQgihQpLchRBChSS5CyGECklyF0IIFfp/YvR6vkNpn+oAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.77\n",
      "Recall: \t\t 0.81\n",
      "Domain Specific return:  -72120 €\n"
     ]
    }
   ],
   "source": [
    "plot_confusion_mat(X_test, y_test, cls_rforest, \"Random Forest\")\n",
    "calc_domain_specific_metrics(y_test, predY_rforest)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2.3 Logistic Regression - Weight configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "cls_LR = pipeline.Pipeline([                           # Setting up Pipeline for LR\n",
    "    ('scale', StandardScaler()),\n",
    "    ('clf', LogisticRegression(C=1, class_weight={0:1.0, 1:1.0}, random_state = np.random.seed(1304)))\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "cls_LR = cls_LR.fit(X_train, y_train)                  # Train on training set\n",
    "predY_LR = cls_LR.predict(X_test)                      # Predict on testing set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.70\n",
      "Recall: \t\t 0.99\n",
      "Domain Specific return:  -140645 €\n"
     ]
    }
   ],
   "source": [
    "plot_confusion_mat(X_test, y_test, cls_LR, \"Logistic Regression\")\n",
    "calc_domain_specific_metrics(y_test, predY_LR)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.3 Optimal decision thresholds"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "My aim with threshold is to not just only build a general robust classifier, but a classifier which is optimized based on sensitivity towards predicting positives. The idea therefore is to the decision threshold to identify the best decision point. This basically means that I am trying to optimize domain specific costs while holding other parameters constant. \n",
    "\n",
    "In the following function, I will plot the consequences of different decision boundaries on the domain specific costs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "from tqdm import tqdm\n",
    "\n",
    "def compute_thresholds(clf):\n",
    "    returns = []\n",
    "    thresholds = []\n",
    "    \n",
    "    for threshold in tqdm(np.logspace(0, 1, num=100, base=0.1)):\n",
    "        predicted_proba = clf.predict_proba(X_test)\n",
    "        predicted = (predicted_proba [:,1] >= threshold).astype('int')\n",
    "        returns.append(calc_return_only(y_test, predicted))\n",
    "        thresholds.append(threshold)\n",
    "        \n",
    "    ax = sns.lineplot(x=thresholds, y=returns)\n",
    "    ax.set(xlabel='Threshold', ylabel='Expected return in €', title='Threshold vs. Expected Return')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3.1 Support Vector Machines - decision thresholds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 500/500 [14:32<00:00,  1.75s/it]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compute_thresholds(clf = cls_svm)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hence, setting a threshold for probabilities **> ~0.8** will lead to positive + stable returns!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.93\n",
      "Recall: \t\t 0.88\n",
      "Domain Specific return:  53880 €\n"
     ]
    }
   ],
   "source": [
    "threshold = 0.9\n",
    "predicted_proba = cls_svm.predict_proba(X_test)\n",
    "predicted = (predicted_proba [:,1] >= threshold).astype('int')\n",
    "\n",
    "calc_domain_specific_metrics(y_test, predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3.2 Random Forest - decision thresholds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 500/500 [06:48<00:00,  1.22it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compute_thresholds(clf = cls_rforest)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hence, setting a threshold for probabilities **> ~0.85** will lead to best returns! However negative!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.88\n",
      "Recall: \t\t 0.66\n",
      "Domain Specific return:  -14340 €\n"
     ]
    }
   ],
   "source": [
    "threshold = 0.85\n",
    "predicted_proba = cls_rforest.predict_proba(X_test)\n",
    "predicted = (predicted_proba [:,1] >= threshold).astype('int')\n",
    "\n",
    "calc_domain_specific_metrics(y_test, predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3.3 Logistic Regression - decision thresholds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 500/500 [04:08<00:00,  2.01it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compute_thresholds(clf = cls_LR)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Hence, setting a threshold for probabilities **~0.9** will lead to best returns!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.95\n",
      "Recall: \t\t 0.85\n",
      "Domain Specific return:  55415 €\n"
     ]
    }
   ],
   "source": [
    "threshold = 0.9\n",
    "predicted_proba = cls_LR.predict_proba(X_test)\n",
    "predicted = (predicted_proba [:,1] >= threshold).astype('int')\n",
    "\n",
    "calc_domain_specific_metrics(y_test, predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.4 Hyperparameter Optimization and Cross Validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.4.1 SVM - GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "estimator_svm = pipeline.Pipeline([                                 # Setting up pipeline for SVM\n",
    "    ('scale', StandardScaler()),\n",
    "    ('clf', SVC(kernel='linear',gamma='auto', probability=True, random_state = np.random.seed(1304)))\n",
    "])\n",
    "\n",
    "param_grid_svm =  {                                                 # Setting up hyperparameter grid\n",
    "    'clf__C' : np.logspace(0, 1, num=10, base=0.01),\n",
    "    'clf__class_weight': [{0:w, 1:1} for w in [1, 2, 5, 10]],\n",
    "    'clf__shrinking': [True, False],\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 80 candidates, totalling 800 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  42 tasks      | elapsed:    3.9s\n",
      "[Parallel(n_jobs=-1)]: Done 192 tasks      | elapsed:   14.2s\n",
      "[Parallel(n_jobs=-1)]: Done 442 tasks      | elapsed:   32.3s\n",
      "[Parallel(n_jobs=-1)]: Done 792 tasks      | elapsed:  1.1min\n",
      "[Parallel(n_jobs=-1)]: Done 800 out of 800 | elapsed:  1.1min finished\n"
     ]
    }
   ],
   "source": [
    "cls_svm_grid = GridSearchCV(estimator_svm, param_grid_svm, cv=10, n_jobs=-1, verbose=1, refit='precision_score')\n",
    "cls_svm_grid = cls_svm_grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.87\n",
      "Recall: \t\t 0.94\n",
      "Domain Specific return:  18705 €\n"
     ]
    }
   ],
   "source": [
    "predY_svm_grid = cls_svm_grid.predict(X_test)                   # Predict on test set\n",
    "plot_confusion_mat(X_test, y_test, cls_svm_grid, \"SVM Grid\")    # Plot confusion matrix\n",
    "calc_domain_specific_metrics(y_test, predY_svm_grid)            # compute domain return"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 100/100 [02:33<00:00,  1.53s/it]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compute_thresholds(clf = cls_svm_grid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.95\n",
      "Recall: \t\t 0.83\n",
      "Domain Specific return:  53335 €\n"
     ]
    }
   ],
   "source": [
    "threshold = 0.95\n",
    "predicted_proba = cls_svm_grid.predict_proba(X_test)\n",
    "predicted = (predicted_proba [:,1] >= threshold).astype('int')\n",
    "\n",
    "calc_domain_specific_metrics(y_test, predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.4.2 Random Forest - GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "estimator_rf = pipeline.Pipeline([\n",
    "    ('scale', StandardScaler()),\n",
    "    ('clf', RandomForestClassifier(random_state = np.random.seed(1304)))\n",
    "])\n",
    "\n",
    "param_grid_rf = {\n",
    "    'clf__n_estimators' : [5, 10, 50],\n",
    "    'clf__max_depth' : [None, 10, 20],\n",
    "    'clf__class_weight': [{0:w, 1:1} for w in [1, 2, 5, 10]],\n",
    "    'clf__ccp_alpha' : [0.0, 0.1, 0.2]\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 108 candidates, totalling 1080 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  42 tasks      | elapsed:    2.4s\n",
      "[Parallel(n_jobs=-1)]: Done 620 tasks      | elapsed:   23.0s\n",
      "[Parallel(n_jobs=-1)]: Done 1080 out of 1080 | elapsed:   41.9s finished\n"
     ]
    }
   ],
   "source": [
    "cls_rforest_grid = GridSearchCV(estimator_rf, param_grid_rf, cv=10, n_jobs=-1, verbose=1, refit='precision_score')\n",
    "cls_rforest_grid = cls_rforest_grid.fit(X_train, y_train)\n",
    "predY_rforest_grid = cls_rforest_grid.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.81\n",
      "Recall: \t\t 0.82\n",
      "Domain Specific return:  -41860 €\n"
     ]
    }
   ],
   "source": [
    "plot_confusion_mat(X_test, y_test, cls_rforest_grid, \"RF Grid\")\n",
    "calc_domain_specific_metrics(y_test, predY_rforest_grid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 100/100 [03:58<00:00,  2.38s/it]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compute_thresholds(clf = cls_rforest_grid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.94\n",
      "Recall: \t\t 0.58\n",
      "Domain Specific return:  -5770 €\n"
     ]
    }
   ],
   "source": [
    "threshold = 0.85\n",
    "predicted_proba = cls_rforest_grid.predict_proba(X_test)\n",
    "predicted = (predicted_proba [:,1] >= threshold).astype('int')\n",
    "\n",
    "calc_domain_specific_metrics(y_test, predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.4.3 Logistic Regression - GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "estimator_LR = pipeline.Pipeline([\n",
    "    ('scale', StandardScaler()),\n",
    "    ('clf', LogisticRegression(random_state = np.random.seed(1304)))\n",
    "])\n",
    "\n",
    "param_grid_LR =  {\n",
    "    'clf__C' : np.logspace(0, 1, num=10, base=0.01),\n",
    "    'clf__class_weight': [{0:w, 1:1} for w in [1, 2, 5, 10]],\n",
    "    'clf__penalty': ['l1', 'l2', 'elasticnet'],\n",
    "    'clf__solver' : ['newton-cg', 'lbfgs', 'liblinear', 'sag', 'saga']\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 600 candidates, totalling 6000 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done 943 tasks      | elapsed:    8.4s\n",
      "[Parallel(n_jobs=-1)]: Done 2605 tasks      | elapsed:   25.5s\n",
      "[Parallel(n_jobs=-1)]: Done 5355 tasks      | elapsed:   47.8s\n",
      "[Parallel(n_jobs=-1)]: Done 6000 out of 6000 | elapsed:   52.0s finished\n"
     ]
    }
   ],
   "source": [
    "cls_LR_grid = GridSearchCV(estimator_LR, param_grid_LR, cv=10, n_jobs=-1, verbose=1, refit='precision_score')\n",
    "cls_LR_grid = cls_LR_grid.fit(X_train, y_train)\n",
    "predY_LR_grid = cls_LR_grid.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.86\n",
      "Recall: \t\t 0.95\n",
      "Domain Specific return:  14635 €\n"
     ]
    }
   ],
   "source": [
    "plot_confusion_mat(X_test, y_test, cls_LR_grid, \"LR Grid\")\n",
    "calc_domain_specific_metrics(y_test, predY_LR_grid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 100/100 [00:46<00:00,  2.13it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compute_thresholds(clf = cls_LR_grid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: \t\t 0.96\n",
      "Recall: \t\t 0.82\n",
      "Domain Specific return:  54020 €\n"
     ]
    }
   ],
   "source": [
    "threshold = 0.8\n",
    "predicted_proba = cls_LR_grid.predict_proba(X_test)\n",
    "predicted = (predicted_proba [:,1] >= threshold).astype('int')\n",
    "\n",
    "calc_domain_specific_metrics(y_test, predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.4 Final Prediction using all data\n",
    "For the final prediction I will use the Support Vector Machine model with CV, weighted classes, prediction threshold and tuned hyperparameters. This model gives a stabel tradeoff between precision and recall and generates a positive overall domain-specific return!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, y_train = sm.fit_sample(df_train.drop(columns=['fraud']), df_train['fraud']) # Using all data to train\n",
    "X = df_test # For final prediciton"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 80 candidates, totalling 800 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  42 tasks      | elapsed:    2.6s\n",
      "[Parallel(n_jobs=-1)]: Done 192 tasks      | elapsed:   13.0s\n",
      "[Parallel(n_jobs=-1)]: Done 442 tasks      | elapsed:   29.6s\n",
      "[Parallel(n_jobs=-1)]: Done 792 tasks      | elapsed:  1.0min\n",
      "[Parallel(n_jobs=-1)]: Done 800 out of 800 | elapsed:  1.1min finished\n"
     ]
    }
   ],
   "source": [
    "cls_svm_grid = GridSearchCV(estimator_svm, param_grid_svm, cv=10, n_jobs=-1, verbose=1, refit='precision_score')\n",
    "cls_svm_grid = cls_svm_grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "predY_final = cls_svm_grid.predict(X)                   # Predict on test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pd.DataFrame(predY_final, columns = ['fraud']).to_csv('./data/Justin_Lange-prediction.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 6. Summery\n",
    "| Algorithm | Test in % | Train in % | CV | Hyperp. tuning |  Weight Config  | threshold | Precision | Recall | Domain-specific exp. return |\n",
    "|:---------:|:---------:|:----------:|:---------------:|:------------------:|:---------------:|:------------------:|:---------:|:------:|:---------------------------:|\n",
    "|     NB    |     80    |     20     |        -        |          -         |        -        |          -         |    .33    |   .94  |            - 929.970 €            |\n",
    "|    SVM    |     80    |     20     |        -        |          -         |  {0:1.0, 1:2.0} |          -         |    .59    |   1.0  |            - 267.760 €            |\n",
    "|     LR    |     80    |     20     |        -        |          -         |  {0:1.0, 1:1.0} |          -         |    .67    |   1.0  |            -140.645 €            |\n",
    "|     RF    |     80    |     20     |        -        |          -         | {0:1.0, 1:50.0} |          -         |    .85    |   .69  |             - 72.120 €            |\n",
    "|    SVM    |     80    |     20     |        -        |          -         |  {0:1.0, 1:2.0} |        >0.8        |    .88    |   .94  |              53.880 €             |\n",
    "|     LR    |     80    |     20     |        -        |          -         |  {0:1.0, 1:1.0} |        >0.9        |    .83    |   .62  |             55.415 €            |\n",
    "|     RF    |     80    |     20     |        -        |          -         | {0:1.0, 1:10.0} |        >0.65       |    1.0    |   .69  |             - 14.340 €             |\n",
    "|    SVM    |     80    |     20     |     10 Fold     |         Yes        |  {0:1.0, 1:2.0} |        >0.95        |    .82    |   .94  |             53.335 €             |\n",
    "|     LR    |     80    |     20     |     10 Fold     |         Yes        |  {0:1.0, 1:1.0} |        >0.9        |    .88    |   .96  |             54.020 €            |\n",
    "|     RF    |     80    |     20     |     10 Fold     |         Yes        | {0:1.0, 1:10.0} |        >0.65       |     1.    |   .44  |             - 5.770 €             |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Final Remarks\n",
    "The main goal of the retailer is to get more customers by a smoother shopping experience. This would eventually target more customers coming to the shop and therefore create higher revenue margins. \n",
    "This prediction model is however NOT a measurem about if this business model is worth it or not, but rather a question in what extend we should provide control towards possible theft. Therefore, the following remarks of this business setting can be concluded:\n",
    "#### 1.\tThere is more to lose then to win\n",
    "The prediction model required an extreme high precision rate with low False Positive rates. Implementing a prediction model which is not generalising properly could result in catastrophic effects in returning profits. A good generalised model would however only create small increasing profit margins. Therefore, it is necessary to question if we should provide such reactive approaches in general, but rather a proactive approach:\n",
    "\n",
    "#### 2.\tProactive vs. reactive\n",
    "Our model is a reactive approach based on the customers shopping pattern. Since there is much to lose in this reactive approach, it is reasonable to consider a proactive approach. An example of such proactive approach would be staff standing at the scan ensemble to monitor a bunch of scanners. This is a common approach in for common retailers too (IKEA) that provide self-service checkouts.\n",
    "\n",
    "Furthermore, even if the retailer implements a reactive model which generates negative returns, it might still be useful if they would consider the costs of staff savings. Since a reactive approach requires staff which is constantly monitoring, it could be that a model which is generating negative return might generate overall savings if staff could therefore be reduced."
   ]
  }
 ],
 "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}