Ex_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 371,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "%matplotlib inline \n",
    "%config IPCompleter.greedy=True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Reading data from file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 372,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({-1: 338997, 1: 80149})\n"
     ]
    }
   ],
   "source": [
    "from collections import Counter\n",
    "\n",
    "data1 = pd.read_csv(\"20151219.txt\", sep = \"\\t\", header = None)\n",
    "data1.columns = [\"conn len\", \"service\", \"src bytes\", \"dst bytes\", \"conn numb\", \"Same srv rate\", \"Serror rate\", \"Srv serror rate\", \"Dst host count\", \"Dst host srv count\", \"Dst host same src port rate\", \"Dst host serror rate\", \"Dst host srv serror rate\", \"Conn state\", \"IDS detection\", \"Malware detection\", \"Ashula detection\", \"attack_flag\", \"src IP\", \"src port\", \"dst IP\", \"dst port\", \"start time\", \"proto\"]\n",
    "\n",
    "data2 = pd.read_csv(\"20151220.txt\", sep = \"\\t\", header = None)\n",
    "data2.columns = [\"conn len\", \"service\", \"src bytes\", \"dst bytes\", \"conn numb\", \"Same srv rate\", \"Serror rate\", \"Srv serror rate\", \"Dst host count\", \"Dst host srv count\", \"Dst host same src port rate\", \"Dst host serror rate\", \"Dst host srv serror rate\", \"Conn state\", \"IDS detection\", \"Malware detection\", \"Ashula detection\", \"attack_flag\", \"src IP\", \"src port\", \"dst IP\", \"dst port\", \"start time\", \"proto\"]\n",
    "\n",
    "data = data1.append(data2, ignore_index=True)\n",
    "\n",
    "print(Counter(data[\"attack_flag\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Transform to binary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 373,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({1: 338997, 0: 80149})\n"
     ]
    }
   ],
   "source": [
    "data.loc[:, \"attack_flag\"] = data.loc[:,\"attack_flag\"].replace(-2, -1)\n",
    "data.loc[:, \"attack_flag\"] = data.loc[:, \"attack_flag\"] * -1\n",
    "data.loc[:, \"attack_flag\"] = data.loc[:,\"attack_flag\"].replace(-1, 0)\n",
    "\n",
    "print(Counter(data[\"attack_flag\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Drop additional features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 374,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(419146, 24)\n",
      "(419146, 21)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1    338997\n",
       "0     80149\n",
       "Name: attack_flag, dtype: int64"
      ]
     },
     "execution_count": 374,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# slice dataset, drop additional features\n",
    "print(data.shape)\n",
    "data = data.drop(['start time', 'dst IP', 'src IP'], axis=1)\n",
    "print(data.shape)\n",
    "\n",
    "# drop software markers\n",
    "data = data.drop([\"IDS detection\", \"Malware detection\", \"Ashula detection\"], axis=1)\n",
    "data[\"attack_flag\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 375,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 419146 entries, 0 to 419145\n",
      "Data columns (total 18 columns):\n",
      "conn len                       419146 non-null float64\n",
      "service                        419146 non-null object\n",
      "src bytes                      419146 non-null int64\n",
      "dst bytes                      419146 non-null int64\n",
      "conn numb                      419146 non-null int64\n",
      "Same srv rate                  419146 non-null float64\n",
      "Serror rate                    419146 non-null float64\n",
      "Srv serror rate                419146 non-null float64\n",
      "Dst host count                 419146 non-null int64\n",
      "Dst host srv count             419146 non-null int64\n",
      "Dst host same src port rate    419146 non-null float64\n",
      "Dst host serror rate           419146 non-null float64\n",
      "Dst host srv serror rate       419146 non-null float64\n",
      "Conn state                     419146 non-null object\n",
      "attack_flag                    419146 non-null int64\n",
      "src port                       419146 non-null int64\n",
      "dst port                       419146 non-null int64\n",
      "proto                          419146 non-null object\n",
      "dtypes: float64(7), int64(8), object(3)\n",
      "memory usage: 57.6+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 376,
   "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>conn len</th>\n",
       "      <th>service</th>\n",
       "      <th>src bytes</th>\n",
       "      <th>dst bytes</th>\n",
       "      <th>conn numb</th>\n",
       "      <th>Same srv rate</th>\n",
       "      <th>Serror rate</th>\n",
       "      <th>Srv serror rate</th>\n",
       "      <th>Dst host count</th>\n",
       "      <th>Dst host srv count</th>\n",
       "      <th>Dst host same src port rate</th>\n",
       "      <th>Dst host serror rate</th>\n",
       "      <th>Dst host srv serror rate</th>\n",
       "      <th>Conn state</th>\n",
       "      <th>attack_flag</th>\n",
       "      <th>src port</th>\n",
       "      <th>dst port</th>\n",
       "      <th>proto</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>419136</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>S0</td>\n",
       "      <td>1</td>\n",
       "      <td>54695</td>\n",
       "      <td>8081</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419137</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>S0</td>\n",
       "      <td>1</td>\n",
       "      <td>44997</td>\n",
       "      <td>21320</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419138</th>\n",
       "      <td>0.485847</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>RSTR</td>\n",
       "      <td>1</td>\n",
       "      <td>65024</td>\n",
       "      <td>3389</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419139</th>\n",
       "      <td>3.253905</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.50</td>\n",
       "      <td>1</td>\n",
       "      <td>39</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>S0</td>\n",
       "      <td>1</td>\n",
       "      <td>18399</td>\n",
       "      <td>25</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419140</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>S0</td>\n",
       "      <td>1</td>\n",
       "      <td>33863</td>\n",
       "      <td>20012</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419141</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>OTH</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>icmp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419142</th>\n",
       "      <td>3.135446</td>\n",
       "      <td>rdp</td>\n",
       "      <td>1726</td>\n",
       "      <td>5591</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>5</td>\n",
       "      <td>72</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>S1</td>\n",
       "      <td>1</td>\n",
       "      <td>62518</td>\n",
       "      <td>3389</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419143</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>S0</td>\n",
       "      <td>1</td>\n",
       "      <td>35137</td>\n",
       "      <td>123</td>\n",
       "      <td>udp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419144</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>S0</td>\n",
       "      <td>1</td>\n",
       "      <td>42613</td>\n",
       "      <td>81</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419145</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>S0</td>\n",
       "      <td>1</td>\n",
       "      <td>42346</td>\n",
       "      <td>21320</td>\n",
       "      <td>tcp</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        conn len service  src bytes  dst bytes  conn numb  Same srv rate  \\\n",
       "419136  0.000000   other          0          0          0            0.0   \n",
       "419137  0.000000   other          0          0          0            0.0   \n",
       "419138  0.485847   other          0          0          0            0.0   \n",
       "419139  3.253905   other          0          0          0            0.0   \n",
       "419140  0.000000   other          0          0          0            0.0   \n",
       "419141  0.000000   other          0          0          0            0.0   \n",
       "419142  3.135446     rdp       1726       5591          1            0.0   \n",
       "419143  0.000000   other          0          0          0            0.0   \n",
       "419144  0.000000   other          0          0          0            0.0   \n",
       "419145  0.000000   other          0          0          0            0.0   \n",
       "\n",
       "        Serror rate  Srv serror rate  Dst host count  Dst host srv count  \\\n",
       "419136          0.0             1.00               0                   0   \n",
       "419137          0.0             0.00               1                   1   \n",
       "419138          0.0             1.00               5                   4   \n",
       "419139          0.0             0.50               1                  39   \n",
       "419140          0.0             0.67               4                   4   \n",
       "419141          0.0             0.75               0                   0   \n",
       "419142          0.0             0.00               5                  72   \n",
       "419143          0.0             0.60               0                   2   \n",
       "419144          0.0             0.67               0                   0   \n",
       "419145          0.0             1.00               0                   0   \n",
       "\n",
       "        Dst host same src port rate  Dst host serror rate  \\\n",
       "419136                          0.0                   0.0   \n",
       "419137                          0.0                   1.0   \n",
       "419138                          0.0                   0.0   \n",
       "419139                          1.0                   1.0   \n",
       "419140                          1.0                   1.0   \n",
       "419141                          0.0                   0.0   \n",
       "419142                          0.0                   0.0   \n",
       "419143                          0.0                   0.0   \n",
       "419144                          0.0                   0.0   \n",
       "419145                          0.0                   0.0   \n",
       "\n",
       "        Dst host srv serror rate Conn state  attack_flag  src port  dst port  \\\n",
       "419136                       0.0         S0            1     54695      8081   \n",
       "419137                       1.0         S0            1     44997     21320   \n",
       "419138                       0.0       RSTR            1     65024      3389   \n",
       "419139                       1.0         S0            1     18399        25   \n",
       "419140                       1.0         S0            1     33863     20012   \n",
       "419141                       0.0        OTH            1         3         3   \n",
       "419142                       0.0         S1            1     62518      3389   \n",
       "419143                       1.0         S0            1     35137       123   \n",
       "419144                       0.0         S0            1     42613        81   \n",
       "419145                       0.0         S0            1     42346     21320   \n",
       "\n",
       "       proto  \n",
       "419136   tcp  \n",
       "419137   tcp  \n",
       "419138   tcp  \n",
       "419139   tcp  \n",
       "419140   tcp  \n",
       "419141  icmp  \n",
       "419142   tcp  \n",
       "419143   udp  \n",
       "419144   tcp  \n",
       "419145   tcp  "
      ]
     },
     "execution_count": 376,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.tail(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 377,
   "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>conn len</th>\n",
       "      <th>src bytes</th>\n",
       "      <th>dst bytes</th>\n",
       "      <th>conn numb</th>\n",
       "      <th>Same srv rate</th>\n",
       "      <th>Serror rate</th>\n",
       "      <th>Srv serror rate</th>\n",
       "      <th>Dst host count</th>\n",
       "      <th>Dst host srv count</th>\n",
       "      <th>Dst host same src port rate</th>\n",
       "      <th>Dst host serror rate</th>\n",
       "      <th>Dst host srv serror rate</th>\n",
       "      <th>attack_flag</th>\n",
       "      <th>src port</th>\n",
       "      <th>dst port</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>419146.000000</td>\n",
       "      <td>4.191460e+05</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "      <td>419146.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.897800</td>\n",
       "      <td>2.168004e+04</td>\n",
       "      <td>364.750896</td>\n",
       "      <td>2.522923</td>\n",
       "      <td>0.357259</td>\n",
       "      <td>0.064550</td>\n",
       "      <td>0.548421</td>\n",
       "      <td>25.437657</td>\n",
       "      <td>28.919419</td>\n",
       "      <td>0.073814</td>\n",
       "      <td>0.068824</td>\n",
       "      <td>0.153320</td>\n",
       "      <td>0.808780</td>\n",
       "      <td>34004.366686</td>\n",
       "      <td>3584.727816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>4.749450</td>\n",
       "      <td>4.708272e+06</td>\n",
       "      <td>2285.497265</td>\n",
       "      <td>5.451557</td>\n",
       "      <td>0.477454</td>\n",
       "      <td>0.244823</td>\n",
       "      <td>0.462899</td>\n",
       "      <td>41.519257</td>\n",
       "      <td>42.252740</td>\n",
       "      <td>0.259441</td>\n",
       "      <td>0.252677</td>\n",
       "      <td>0.357215</td>\n",
       "      <td>0.393262</td>\n",
       "      <td>19713.691673</td>\n",
       "      <td>10928.964687</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>12375.000000</td>\n",
       "      <td>23.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.000091</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.750000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>38884.000000</td>\n",
       "      <td>53.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.721284</td>\n",
       "      <td>4.500000e+01</td>\n",
       "      <td>77.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>73.000000</td>\n",
       "      <td>92.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>50359.000000</td>\n",
       "      <td>1433.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1061.910079</td>\n",
       "      <td>1.696585e+09</td>\n",
       "      <td>157723.000000</td>\n",
       "      <td>84.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>65535.000000</td>\n",
       "      <td>65507.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            conn len     src bytes      dst bytes      conn numb  \\\n",
       "count  419146.000000  4.191460e+05  419146.000000  419146.000000   \n",
       "mean        0.897800  2.168004e+04     364.750896       2.522923   \n",
       "std         4.749450  4.708272e+06    2285.497265       5.451557   \n",
       "min         0.000000  0.000000e+00       0.000000       0.000000   \n",
       "25%         0.000000  0.000000e+00       0.000000       0.000000   \n",
       "50%         0.000091  0.000000e+00       0.000000       0.000000   \n",
       "75%         0.721284  4.500000e+01      77.000000       3.000000   \n",
       "max      1061.910079  1.696585e+09  157723.000000      84.000000   \n",
       "\n",
       "       Same srv rate    Serror rate  Srv serror rate  Dst host count  \\\n",
       "count  419146.000000  419146.000000    419146.000000   419146.000000   \n",
       "mean        0.357259       0.064550         0.548421       25.437657   \n",
       "std         0.477454       0.244823         0.462899       41.519257   \n",
       "min         0.000000       0.000000         0.000000        0.000000   \n",
       "25%         0.000000       0.000000         0.000000        0.000000   \n",
       "50%         0.000000       0.000000         0.750000        0.000000   \n",
       "75%         1.000000       0.000000         1.000000       73.000000   \n",
       "max         1.000000       1.000000         1.000000      100.000000   \n",
       "\n",
       "       Dst host srv count  Dst host same src port rate  Dst host serror rate  \\\n",
       "count       419146.000000                419146.000000         419146.000000   \n",
       "mean            28.919419                     0.073814              0.068824   \n",
       "std             42.252740                     0.259441              0.252677   \n",
       "min              0.000000                     0.000000              0.000000   \n",
       "25%              0.000000                     0.000000              0.000000   \n",
       "50%              1.000000                     0.000000              0.000000   \n",
       "75%             92.000000                     0.000000              0.000000   \n",
       "max            100.000000                     1.000000              1.000000   \n",
       "\n",
       "       Dst host srv serror rate    attack_flag       src port       dst port  \n",
       "count             419146.000000  419146.000000  419146.000000  419146.000000  \n",
       "mean                   0.153320       0.808780   34004.366686    3584.727816  \n",
       "std                    0.357215       0.393262   19713.691673   10928.964687  \n",
       "min                    0.000000       0.000000       0.000000       0.000000  \n",
       "25%                    0.000000       1.000000   12375.000000      23.000000  \n",
       "50%                    0.000000       1.000000   38884.000000      53.000000  \n",
       "75%                    0.000000       1.000000   50359.000000    1433.000000  \n",
       "max                    1.000000       1.000000   65535.000000   65507.000000  "
      ]
     },
     "execution_count": 377,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 378,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(419146, 18)"
      ]
     },
     "execution_count": 378,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# removing NaNs\n",
    "data = data.dropna()\n",
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 379,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "S0        247950\n",
       "SF        104018\n",
       "RSTO       26466\n",
       "OTH        19608\n",
       "REJ        15659\n",
       "RSTRH       1853\n",
       "RSTOS0      1514\n",
       "SHR         1195\n",
       "RSTR         778\n",
       "S1           103\n",
       "SH             2\n",
       "Name: Conn state, dtype: int64"
      ]
     },
     "execution_count": 379,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[\"Conn state\"].value_counts(dropna=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 380,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "other      259775\n",
       "dns        102883\n",
       "ssh         29769\n",
       "sip         20915\n",
       "rdp          3640\n",
       "snmp         1096\n",
       "smtp          999\n",
       "dhcp           43\n",
       "http           21\n",
       "ssl             3\n",
       "dns,sip         2\n",
       "Name: service, dtype: int64"
      ]
     },
     "execution_count": 380,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[\"service\"].value_counts(dropna=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 381,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tcp     256581\n",
       "udp     153868\n",
       "icmp      8697\n",
       "Name: proto, dtype: int64"
      ]
     },
     "execution_count": 381,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[\"proto\"].value_counts(dropna=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# a bit of feature engineering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## port numbers classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 382,
   "metadata": {},
   "outputs": [],
   "source": [
    "t1 = 1023  # well-known port numbers\n",
    "t2 = 49151 # registered ports\n",
    "t3 = 65535 # client ports\n",
    "\n",
    "def wk(data_row):\n",
    "    if (data_row[\"src port\"] <= t1): \n",
    "        value = 1\n",
    "    elif ((data_row[\"src port\"] > t1) and (data_row[\"src port\"] <= t2)):\n",
    "        value = 0\n",
    "    elif ((data_row[\"src port\"] > t2) and (data_row[\"src port\"] <= t3)):\n",
    "        value = 0\n",
    "    return value\n",
    "\n",
    "def reg(data_row):\n",
    "    if (data_row[\"src port\"] <= t1): \n",
    "        value = 0\n",
    "    elif ((data_row[\"src port\"] > t1) and (data_row[\"src port\"] <= t2)):\n",
    "        value = 1\n",
    "    elif ((data_row[\"src port\"] > t2) and (data_row[\"src port\"] <= t3)):\n",
    "        value = 0\n",
    "    return value\n",
    "\n",
    "def cli(data_row):\n",
    "    if (data_row[\"src port\"] <= t1): \n",
    "        value = 0\n",
    "    elif ((data_row[\"src port\"] > t1) and (data_row[\"src port\"] <= t2)):\n",
    "        value = 0\n",
    "    elif ((data_row[\"src port\"] > t2) and (data_row[\"src port\"] <= t3)):\n",
    "        value = 1\n",
    "    return value\n",
    "\n",
    "data[\"well-known src pool\"] = data.apply(wk, axis=1)\n",
    "data[\"registered src pool\"] = data.apply(reg, axis=1)\n",
    "data[\"cli src pool\"] = data.apply(cli, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 383,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({0: 406741, 1: 12405})\n",
      "Counter({1: 291278, 0: 127868})\n",
      "Counter({0: 303683, 1: 115463})\n"
     ]
    }
   ],
   "source": [
    "print(Counter(data[\"well-known src pool\"]))\n",
    "print(Counter(data[\"registered src pool\"]))\n",
    "print(Counter(data[\"cli src pool\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 384,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 419146 entries, 0 to 419145\n",
      "Data columns (total 21 columns):\n",
      "conn len                       419146 non-null float64\n",
      "service                        419146 non-null object\n",
      "src bytes                      419146 non-null int64\n",
      "dst bytes                      419146 non-null int64\n",
      "conn numb                      419146 non-null int64\n",
      "Same srv rate                  419146 non-null float64\n",
      "Serror rate                    419146 non-null float64\n",
      "Srv serror rate                419146 non-null float64\n",
      "Dst host count                 419146 non-null int64\n",
      "Dst host srv count             419146 non-null int64\n",
      "Dst host same src port rate    419146 non-null float64\n",
      "Dst host serror rate           419146 non-null float64\n",
      "Dst host srv serror rate       419146 non-null float64\n",
      "Conn state                     419146 non-null object\n",
      "attack_flag                    419146 non-null int64\n",
      "src port                       419146 non-null int64\n",
      "dst port                       419146 non-null int64\n",
      "proto                          419146 non-null object\n",
      "well-known src pool            419146 non-null int64\n",
      "registered src pool            419146 non-null int64\n",
      "cli src pool                   419146 non-null int64\n",
      "dtypes: float64(7), int64(11), object(3)\n",
      "memory usage: 70.4+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 385,
   "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>src port</th>\n",
       "      <th>dst port</th>\n",
       "      <th>well-known src pool</th>\n",
       "      <th>registered src pool</th>\n",
       "      <th>cli src pool</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>40061</td>\n",
       "      <td>111</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>53488</td>\n",
       "      <td>80</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>35051</td>\n",
       "      <td>23</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1408</td>\n",
       "      <td>445</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>55244</td>\n",
       "      <td>995</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>44544</td>\n",
       "      <td>995</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>50678</td>\n",
       "      <td>5900</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>35051</td>\n",
       "      <td>23</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>50678</td>\n",
       "      <td>5900</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>50678</td>\n",
       "      <td>5900</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>59625</td>\n",
       "      <td>21</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>42626</td>\n",
       "      <td>111</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2194</td>\n",
       "      <td>445</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>7432</td>\n",
       "      <td>22</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>7206</td>\n",
       "      <td>80</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>55133</td>\n",
       "      <td>53</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>60169</td>\n",
       "      <td>80</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>63030</td>\n",
       "      <td>2222</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>31905</td>\n",
       "      <td>445</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>59625</td>\n",
       "      <td>21</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>36939</td>\n",
       "      <td>111</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>7432</td>\n",
       "      <td>22</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>25385</td>\n",
       "      <td>445</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>36877</td>\n",
       "      <td>2332</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>7206</td>\n",
       "      <td>80</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>56346</td>\n",
       "      <td>8888</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>37854</td>\n",
       "      <td>445</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>56248</td>\n",
       "      <td>445</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>38273</td>\n",
       "      <td>1080</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>3026</td>\n",
       "      <td>445</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    src port  dst port  well-known src pool  registered src pool  cli src pool\n",
       "0      40061       111                    0                    1             0\n",
       "1      53488        80                    0                    0             1\n",
       "2      35051        23                    0                    1             0\n",
       "3       1408       445                    0                    1             0\n",
       "4      55244       995                    0                    0             1\n",
       "5      44544       995                    0                    1             0\n",
       "6      50678      5900                    0                    0             1\n",
       "7      35051        23                    0                    1             0\n",
       "8      50678      5900                    0                    0             1\n",
       "9      50678      5900                    0                    0             1\n",
       "10     59625        21                    0                    0             1\n",
       "11     42626       111                    0                    1             0\n",
       "12      2194       445                    0                    1             0\n",
       "13      7432        22                    0                    1             0\n",
       "14      7206        80                    0                    1             0\n",
       "15     55133        53                    0                    0             1\n",
       "16     60169        80                    0                    0             1\n",
       "17     63030      2222                    0                    0             1\n",
       "18     31905       445                    0                    1             0\n",
       "19     59625        21                    0                    0             1\n",
       "20     36939       111                    0                    1             0\n",
       "21      7432        22                    0                    1             0\n",
       "22     25385       445                    0                    1             0\n",
       "23     36877      2332                    0                    1             0\n",
       "24      7206        80                    0                    1             0\n",
       "25     56346      8888                    0                    0             1\n",
       "26     37854       445                    0                    1             0\n",
       "27     56248       445                    0                    0             1\n",
       "28     38273      1080                    0                    1             0\n",
       "29      3026       445                    0                    1             0"
      ]
     },
     "execution_count": 385,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# data.loc[:, [\"src port\", \"dst port\", \"well-known src pool\", \"registered src pool\", \"cli src pool\", \"well-known dst pool\", \"registered dst pool\", \"cli dst pool\"]].head(30)\n",
    "data.loc[:, [\"src port\", \"dst port\", \"well-known src pool\", \"registered src pool\", \"cli src pool\"]].head(30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Change port number info to categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 386,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "56410\n",
      "4462\n"
     ]
    }
   ],
   "source": [
    "data[\"src port\"] = data[\"src port\"].apply(str)\n",
    "data[\"dst port\"] = data[\"dst port\"].apply(str)\n",
    "print(data['src port'].nunique())\n",
    "print(data['dst port'].nunique())\n",
    "\n",
    "thresh1 = 230\n",
    "thresh2 = 230\n",
    "counts1 = data['src port'].value_counts()\n",
    "counts2 = data['dst port'].value_counts()\n",
    "\n",
    "repl1 = counts1[counts1 <= thresh1].index\n",
    "repl2 = counts2[counts2 <= thresh2].index\n",
    "\n",
    "#hot one encoding for src port numbers\n",
    "data = pd.concat([data, pd.get_dummies(data[\"src port\"].replace(repl1, 'other_src'), prefix=\"is_src\")], axis=1)\n",
    "\n",
    "#hot one encoding for dst port numbers\n",
    "data = pd.concat([data, pd.get_dummies(data[\"dst port\"].replace(repl2, 'other_dst'), prefix=\"is_dst\")], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 387,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(419146, 267)"
      ]
     },
     "execution_count": 387,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Hot one encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 388,
   "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>conn len</th>\n",
       "      <th>service</th>\n",
       "      <th>src bytes</th>\n",
       "      <th>dst bytes</th>\n",
       "      <th>conn numb</th>\n",
       "      <th>Same srv rate</th>\n",
       "      <th>Serror rate</th>\n",
       "      <th>Srv serror rate</th>\n",
       "      <th>Dst host count</th>\n",
       "      <th>Dst host srv count</th>\n",
       "      <th>...</th>\n",
       "      <th>state_REJ</th>\n",
       "      <th>state_RSTO</th>\n",
       "      <th>state_RSTOS0</th>\n",
       "      <th>state_RSTR</th>\n",
       "      <th>state_RSTRH</th>\n",
       "      <th>state_S0</th>\n",
       "      <th>state_S1</th>\n",
       "      <th>state_SF</th>\n",
       "      <th>state_SH</th>\n",
       "      <th>state_SHR</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>419136</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419137</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419138</th>\n",
       "      <td>0.485847</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419139</th>\n",
       "      <td>3.253905</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.50</td>\n",
       "      <td>1</td>\n",
       "      <td>39</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419140</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419141</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419142</th>\n",
       "      <td>3.135446</td>\n",
       "      <td>rdp</td>\n",
       "      <td>1726</td>\n",
       "      <td>5591</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>5</td>\n",
       "      <td>72</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419143</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419144</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>419145</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 292 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        conn len service  src bytes  dst bytes  conn numb  Same srv rate  \\\n",
       "419136  0.000000   other          0          0          0            0.0   \n",
       "419137  0.000000   other          0          0          0            0.0   \n",
       "419138  0.485847   other          0          0          0            0.0   \n",
       "419139  3.253905   other          0          0          0            0.0   \n",
       "419140  0.000000   other          0          0          0            0.0   \n",
       "419141  0.000000   other          0          0          0            0.0   \n",
       "419142  3.135446     rdp       1726       5591          1            0.0   \n",
       "419143  0.000000   other          0          0          0            0.0   \n",
       "419144  0.000000   other          0          0          0            0.0   \n",
       "419145  0.000000   other          0          0          0            0.0   \n",
       "\n",
       "        Serror rate  Srv serror rate  Dst host count  Dst host srv count  \\\n",
       "419136          0.0             1.00               0                   0   \n",
       "419137          0.0             0.00               1                   1   \n",
       "419138          0.0             1.00               5                   4   \n",
       "419139          0.0             0.50               1                  39   \n",
       "419140          0.0             0.67               4                   4   \n",
       "419141          0.0             0.75               0                   0   \n",
       "419142          0.0             0.00               5                  72   \n",
       "419143          0.0             0.60               0                   2   \n",
       "419144          0.0             0.67               0                   0   \n",
       "419145          0.0             1.00               0                   0   \n",
       "\n",
       "          ...      state_REJ  state_RSTO  state_RSTOS0 state_RSTR  \\\n",
       "419136    ...              0           0             0          0   \n",
       "419137    ...              0           0             0          0   \n",
       "419138    ...              0           0             0          1   \n",
       "419139    ...              0           0             0          0   \n",
       "419140    ...              0           0             0          0   \n",
       "419141    ...              0           0             0          0   \n",
       "419142    ...              0           0             0          0   \n",
       "419143    ...              0           0             0          0   \n",
       "419144    ...              0           0             0          0   \n",
       "419145    ...              0           0             0          0   \n",
       "\n",
       "        state_RSTRH state_S0 state_S1 state_SF  state_SH  state_SHR  \n",
       "419136            0        1        0        0         0          0  \n",
       "419137            0        1        0        0         0          0  \n",
       "419138            0        0        0        0         0          0  \n",
       "419139            0        1        0        0         0          0  \n",
       "419140            0        1        0        0         0          0  \n",
       "419141            0        0        0        0         0          0  \n",
       "419142            0        0        1        0         0          0  \n",
       "419143            0        1        0        0         0          0  \n",
       "419144            0        1        0        0         0          0  \n",
       "419145            0        1        0        0         0          0  \n",
       "\n",
       "[10 rows x 292 columns]"
      ]
     },
     "execution_count": 388,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### hot one encoding ###\n",
    "\n",
    "# hot one encoding for service\n",
    "data = pd.concat([data, pd.get_dummies(data[\"service\"], prefix=\"is\")], axis=1)\n",
    "data.tail(10)\n",
    "\n",
    "#hot one encoding for proto - commented due to reduced column number\n",
    "data = pd.concat([data, pd.get_dummies(data[\"proto\"], prefix=\"is\")], axis=1)\n",
    "data.tail(10)\n",
    "\n",
    "# hot one encoding for conn state\n",
    "data = pd.concat([data, pd.get_dummies(data[\"Conn state\"], prefix=\"state\")], axis=1)\n",
    "data.tail(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 389,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 419146 entries, 0 to 419145\n",
      "Columns: 292 entries, conn len to state_SHR\n",
      "dtypes: float64(7), int64(9), object(5), uint8(271)\n",
      "memory usage: 178.7+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot function definition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 390,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plotting method definition\n",
    "\n",
    "def plot_stats(stat, name, width, rot):\n",
    "    fig, ax = plt.subplots(figsize=(width, 4))    \n",
    "    ax.bar(range(len(stat)), list(stat.values()), align='center')\n",
    "    plt.xticks(range(len(stat)), list(stat.keys()), rotation=rot)\n",
    "    for a,b in enumerate(stat.items()):\n",
    "        ax.text(a, b[1] + max(stat.values())/60 , str(round(b[1],5)), horizontalalignment ='center', verticalalignment='baseline')\n",
    "    ax.spines['right'].set_visible(False)\n",
    "    ax.spines['top'].set_visible(False)\n",
    "    plt.title(name + \" of tested algorithms\")\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 391,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.rename(columns={'Dst host count': 'Dst_host_count', \"Dst host srv count\" : \"Dst_host_srv_count\",\n",
    "                    \"Dst host same src port rate\" : \"Dst_host_same_src_port_rate\",\n",
    "                    \"Dst host serror rate\" : \"Dst_host_serror_rate\",\n",
    "                    \"Dst host srv serror rate\" : \"Dst_host_srv_serror_rate\",\n",
    "                    \"Same srv rate\" : \"Same_srv_rate\",\n",
    "                    \"Srv serror rate\" : \"Srv_serror_rate\",\n",
    "                    \"conn len\" : \"conn_len\",\n",
    "                    \"src bytes\" : \"src_bytes\",\n",
    "                    \"dst bytes\" : \"dst_bytes\",\n",
    "                    \"conn numb\" : \"conn_numb\",\n",
    "                    \"Serror rate\" : \"Serror_rate\",\n",
    "                    \"cli src pool\" : \"cli_src_pool\",\n",
    "                    \"registered src pool\" : \"registered_src_pool\",\n",
    "                    \"well-known src pool\" : \"well_known_src_pool\",\n",
    "                    \"dst port\" : \"dst_port\",\n",
    "                    \"src port\" : \"src_port\",\n",
    "                     \"Conn state\" : \"Conn_state\" \n",
    "                    }, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 392,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({1: 54763, 0: 8109})\n",
      "Counter({1: 59796, 0: 3076})\n",
      "Counter({1: 224438, 0: 68964})\n"
     ]
    }
   ],
   "source": [
    "train, validate, test = np.split(data, [int(.7*len(data)), int(.85*len(data))])\n",
    "\n",
    "train_ex = train\n",
    "validate_ex = validate\n",
    "test_ex = test\n",
    "\n",
    "train = train.drop([\"service\", \"Conn_state\", \"proto\", \"src_port\", \"dst_port\"], axis=1)\n",
    "validate = validate.drop([\"service\", \"Conn_state\", \"proto\", \"src_port\", \"dst_port\"], axis=1)\n",
    "test = test.drop([\"service\", \"Conn_state\", \"proto\", \"src_port\", \"dst_port\"], axis=1)\n",
    "\n",
    "train = train.sample(frac=1)\n",
    "validate = validate.sample(frac=1)\n",
    "test = test.sample(frac=1)\n",
    "\n",
    "y_train = train.loc[:, \"attack_flag\"].values.ravel()\n",
    "X_train = train.drop([\"attack_flag\"], axis=1).astype(np.float32)\n",
    "y_validate = validate.loc[:, \"attack_flag\"].values.ravel()\n",
    "X_validate = validate.drop([\"attack_flag\"], axis=1).astype(np.float32)\n",
    "y_test = test.loc[:, \"attack_flag\"].values.ravel()\n",
    "X_test = test.drop([\"attack_flag\"], axis=1).astype(np.float32)\n",
    "\n",
    "print(Counter(y_test))\n",
    "print(Counter(y_validate))\n",
    "print(Counter(y_train))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Algoritms performance comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({1: 54763, 0: 8109})\n"
     ]
    }
   ],
   "source": [
    "from collections import Counter\n",
    "\n",
    "print(Counter(y_test))\n",
    "\n",
    "Acc = {}\n",
    "F1S = {}\n",
    "Prec = {}\n",
    "Rec = {}\n",
    "FPR = {}\n",
    "FNR = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: baseline     Accuracy: 0.87102, F1-score: 0.93107, Precision: 0.87102, Recall: 1.00000, FPR: 1.00000, FNR: 0.00000\n",
      "TN:   0; FP: 8109; FN:    0; TP: 54763\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Metrics of baseline method\n",
    "from sklearn import metrics\n",
    "\n",
    "name = \"baseline\"\n",
    "\n",
    "if(y_test.mean() > (1-y_test.mean())):\n",
    "    y_pred = np.ones(len(y_test)).ravel() \n",
    "else:\n",
    "    y_pred = np.zeros(len(y_test)).ravel() \n",
    "    \n",
    "Acc[name] = metrics.accuracy_score(y_test, y_pred)\n",
    "F1S[name] = metrics.f1_score(y_test, y_pred)\n",
    "Prec[name] = metrics.precision_score(y_test, y_pred)\n",
    "Rec[name] = metrics.recall_score(y_test, y_pred)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(y_test, y_pred).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: LogisticReg  Accuracy: 0.92776, F1-score: 0.95724, Precision: 0.98793, Recall: 0.92840, FPR: 0.07658, FNR: 0.07160\n",
      "TN: 7488; FP: 621; FN: 3921; TP: 50842\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn import metrics\n",
    "\n",
    "clf = LogisticRegression(solver = 'newton-cg')\n",
    "\n",
    "model = clf.fit(X_train, y_train)\n",
    "\n",
    "y_pred = model.predict(X_test)\n",
    "\n",
    "name = \"LogisticReg\"\n",
    "\n",
    "Acc[name] = metrics.accuracy_score(y_test, y_pred)\n",
    "F1S[name] = metrics.f1_score(y_test, y_pred)\n",
    "Prec[name] = metrics.precision_score(y_test, y_pred)\n",
    "Rec[name] = metrics.recall_score(y_test, y_pred)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(y_test, y_pred).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: AdaBoost     Accuracy: 0.98082, F1-score: 0.98901, Precision: 0.98689, Recall: 0.99114, FPR: 0.08891, FNR: 0.00886\n",
      "TN: 7388; FP: 721; FN:  485; TP: 54278\n",
      "\n",
      "Algorithm: DecisionTree Accuracy: 0.97463, F1-score: 0.98555, Precision: 0.97777, Recall: 0.99346, FPR: 0.15255, FNR: 0.00654\n",
      "TN: 6872; FP: 1237; FN:  358; TP: 54405\n",
      "\n",
      "Algorithm: NaiveBayes   Accuracy: 0.86727, F1-score: 0.91772, Precision: 0.99739, Recall: 0.84984, FPR: 0.01505, FNR: 0.15016\n",
      "TN: 7987; FP: 122; FN: 8223; TP: 46540\n",
      "\n",
      "Algorithm: RandomTree   Accuracy: 0.97118, F1-score: 0.98363, Precision: 0.97351, Recall: 0.99396, FPR: 0.18264, FNR: 0.00604\n",
      "TN: 6628; FP: 1481; FN:  331; TP: 54432\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/preprocessing/data.py:625: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  return self.partial_fit(X, y)\n",
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/base.py:465: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  return self.fit(X, y, **fit_params).transform(X)\n",
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/pipeline.py:331: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  Xt = transform.transform(Xt)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: RandomForest Accuracy: 0.98176, F1-score: 0.98960, Precision: 0.98251, Recall: 0.99680, FPR: 0.11987, FNR: 0.00320\n",
      "TN: 7137; FP: 972; FN:  175; TP: 54588\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# # zmiana n_estimators daje poprawe\n",
    "# # kfold tylko dla AdaBoostClassifier(DecisionTreeClassifier(max_depth = 11), n_estimators=num_trees, random_state = 0)\n",
    "# # learning rate [1 - default] najlepsze wyniki\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.ensemble import BaggingClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "\n",
    "seed = 12\n",
    "num_trees = 40\n",
    "\n",
    "classifiers = {\n",
    "               \"AdaBoost\":AdaBoostClassifier(DecisionTreeClassifier(max_depth = 13), n_estimators=num_trees, random_state = 0),\n",
    "               \"DecisionTree\":DecisionTreeClassifier(random_state=0, max_depth = 13),\n",
    "               \"NaiveBayes\":GaussianNB(),\n",
    "               \"RandomTree\":BaggingClassifier(DecisionTreeClassifier(random_state=0, criterion = \"entropy\", max_depth = 13)),\n",
    "               \"RandomForest\":make_pipeline(StandardScaler(),RandomForestClassifier(n_estimators=167, max_depth=38, random_state=0)),\n",
    "}\n",
    "\n",
    "\n",
    "for name, clf in classifiers.items():\n",
    "    \n",
    "    model = clf.fit(X_train, y_train)\n",
    "    \n",
    "    y_pred = model.predict(X_test)\n",
    "    \n",
    "    Acc[name] = metrics.accuracy_score(y_test, y_pred)\n",
    "    F1S[name] = metrics.f1_score(y_test, y_pred)\n",
    "    Prec[name] = metrics.precision_score(y_test, y_pred)\n",
    "    Rec[name] = metrics.recall_score(y_test, y_pred)\n",
    "    tn, fp, fn, tp = metrics.confusion_matrix(y_test, y_pred).ravel()\n",
    "    FPR[name] = fp/(fp+tn)\n",
    "    FNR[name] = fn/(tp+fn)\n",
    "\n",
    "    print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "    print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAE/CAYAAAB8YAsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzs3Xm8l3P+//HHS4skshTSieQklbYpKcZgshUia81kDz9jGcaIYSxj+VpC1jEYo6QpYSIj2WUbk5CQLYkSadFCUzl5/f54Xef4nNOpTvU5Ptfnc5732+3c6nNd17k+7+t8ltf13l5vc3dEREQkPTbIdQFERESkPAVnERGRlFFwFhERSRkFZxERkZRRcBYREUkZBWcREZGUUXAWSSEz29rMXjKzxWZ2Y67Lszpm9qKZDVjH33UzK852mZJzP2lmx69m/xAzu6o6nltkfSk4SyolX/jfmtmGuS5LjpwKzAU2dffzKu7MVmAxs+ZJgKy9vudKG3fv6e5DAczsBDN7JddlEqkqBWdJHTNrDuwJOND7Z37utASp7YEprixBa82Cvtskr+kNLGl0HPA6MAQo1yxpZhuZ2Y1m9rmZLTSzV8xso2TfL83sNTNbYGYzzOyEZHu5ZteKtaik5niGmX0CfJJsuyU5xyIze9PM9sw4vpaZXWRmnybNzm+aWTMzu6NiE7SZPW5m51R2kWa2u5m9kVzHG2a2e7K99LoHmtl3ZrZvhd87Ffhtxv7Hk+3bmtkjZjbHzD4zs7MzfqermU1Mrme2md2U7Hop+XdBcq7uyfEnmdkHSevFU2a2fca59jOzD5Ny3w5YZdeX8bz/SV6Tr8zsdjOru4pjt0z+XouSv8dVFV6nSv9eyb4XzexqM3sVWAK0KH3dzaw18Dege3KNCzKednMzeyJ5Hf9rZjtmnNPN7Hdm9kmy/0oz2zG5nkVmNqr0WsyskZn9O7nO+Wb2sm4QZL24u370k6ofYCrwO6Az8AOwdca+O4AXgaZALWB3YENgO2Ax0A+oA2wJdEx+50VgQMY5TgBeyXjswDPAFsBGybb+yTlqA+cBXwP1kn3nA+8CrYjA1CE5tiswC9ggOa4RESi2ruQatwC+BY5NnqNf8njLZP8Q4KrV/I3K7SdutN8ELgXqAi2AacAByf7/AMcm/28AdEv+3zy5/toZ5zoseQ1aJ2X7M/BaxjUtAo5M/s7nAiWZf98K5ewMdEvO0xz4ADinwt++OPn/yOSnPtAGmFH6OlXh7/Ui8AXQNtlfJ/N1r/iaZ/wN5yevW21gODCyQtnGAJsm510GPJf8bRsCU4Djk2OvIW4A6iQ/ewKW68+SfvL3R3d2kipm9kuiSXeUu78JfAr8Jtm3AXAS8Ht3/9LdV7j7a+6+jKhJPuvuI9z9B3ef5+6T1uKpr3H3+e7+PwB3fyA5R4m730jcALRKjh0A/NndP/LwTnLsBGAh0CM5ri/worvPruT5DgI+cfdhyXOMAD4EDlmLMmfaFWjs7le4+3J3nwbck5QB4ian2Mwauft37v76as51WvL3+MDdS4D/AzomtedeRHP7w+7+A3AzceNSKXd/091fT65xOnAXsFfF48ysFnAEcJm7L3H3KcDQjEOq8vca4u7vJ/t/WM31ZfqXu09IrnM40LHC/uvcfZG7vw+8Bzzt7tPcfSHwJNApOe4HoAmwffL+e9nd1SUh60zBWdLmeOILcG7y+J/81LTdCKhHBOyKmq1ie1XNyHxgZuclzboLk2bQhsnzr+m5hhK1bpJ/h63iuG2Bzyts+5xoEVgX2wPbJs2qC5IyXwRsnew/GdgJ+DBpEj54Dee6JeM884kWgqZJucv+VkkAmlHpWQAz2ylp7v3azBYRgb5RJYc2JmqvmefK/H9V/l6rLMdqZN5YLCFaFTJl3lj9r5LHpccPIlobnjazaWZ24TqURaSMgrOkRtJ3fDSwV/Jl/jXRbNrBzDoQo5eXAjtW8uszVrEd4HuiqbTUNpUcU1bLSfqXL0jKsrm7b0bUiEv7Vlf3XA8AhyblbQ08uorjZhFBMNN2wJerOH6V5c0o02fuvlnGzybu3gvA3T9x937AVsB1wMNmtnEl5yk912kVzrWRu78GfEXcnAAx+CrzcSXuJGq4Ld19U+KGobI+6jlE83hRxrbM81bl77W6mmq11mLdfbG7n+fuLYja/B/MrMeafk9kVRScJU0OA1YQ/Y0dk5/WwMvAce7+I/AP4KZk8FMtM+tuMd1qOLCvmR1tZrWTwUWlTZSTgMPNrL7FnNqT11COTYhAMQeobWaXEv2Opf4OXGlmLS20N7MtAdx9JvAGUWN+pLSZvBJjgZ3M7DdJeY9JrvvfVfxbzSb6PktNABaZ2QUWg+ZqmdkuZrYrgJn1N7PGyd+wdEDUiuQaf6xwrr8BfzKztsnvNjSzo5J9TwBtzexwi5HtZ1P5zU6pTYg+6u/MbGfg9MoOcvcVwL+Ay5PXaWdiYGCpbPy9ilY1GG19mdnBZlac3KwsIv62K6rjuaRmUHCWNDkeuM/dv3D3r0t/gNuB3ybB4I/EYKw3iObW64gBWF8Q/aHnJdsnEQO1AAYDy4kv6KFEIF+dp4j+xI+JptOllG8yvQkYBTxNfBHfC2yUsX8o0I5VN2nj7vOAg5PyzgMGAgdnNOevyb1Am6Tp+dEkuB1C3NB8RrQy/J1ojgc4EHjfzL4DbgH6uvtSd18CXA28mpyrm7uPJv6uI5Om6PeAnkm55wJHAdcm5W4JvLqacv6RGDOwmOgDf3A1x56ZlPdr4m83ghiElY2/1/PA+8DXZlbV31kbLYFnge+IwXd/dfcXq+F5pIYwjVkQyS4z+xXRvN08qanKOjCz64Bt3H2VWb5ECpVqziJZZGZ1gN8Df1dgXjtmtnPSRWBm1pXofhid63KJ5IKCs0iWJMkuFhBTam7OcXHy0SZEv/P3RLfBjcBjOS2RSI6oWVtERCRlVHMWERFJmZwF5wMPPNCJuYf60c86/YwbN85btWrlxcXFfu211660//PPP/cePXp4+/btfe+99/aZM2eW7Rs4cKC3bdvWW7du7WeffbYnyTT8zTff9Hbt2nlxcXG57ZMmTfJu3bp5x44dvUuXLj5hwgQHfPjw4d6+fXtv376977777v7OO+/k/O+iH/2k+ac6PrcXX3yxN2vWzBs0aFDuXOeee6537NjRO3bs6DvttJNvttlmZfu++OIL33///b1169bepk0bnz59+s9x/VW3tvk+s/XTuXNnF1lXJSUl3qJFC//000992bJl3r59e3///ffLHXPkkUf6kCFD3N39ueee8/79+7u7+6uvvuq77767l5SUeElJiXfr1s1feOEFd3ffdddd/bXXXvMff/zRDzzwQB87dqy7u++3335l/3/iiSd8r732KjvX/Pnz3d197Nix3rVr1+q+dJG8VV2f2//85z8+a9Ys33jjjVf53LfeequfeOKJZY/32msvf/rpp93dffHixf79999n81JXRbm1pbBNmDCB4uJiWrRoQd26denbty+PPVZ+7NCUKVPo0SOSNO2zzz5l+82MpUuXsnz5cpYtW8YPP/zA1ltvzVdffcWiRYvo3r07ZsZxxx3Ho48+WvY7ixYtAmDhwoVsu+22AOy+++5svvnmAHTr1o2ZM2dWy/WOGzeOVq1aUVxczLXXXrvS/s8//5wePXrQvn179t5777JyvPDCC3Ts2LHsp169emXXVOqss86iQYPyWStHjRpFmzZtaNu2Lb/5zW/KnqNz58507NiRtm3b8re//a1arrUmqo7X9/bbb6e4uBgzY+7cn6Z2f/jhh3Tv3p0NN9yQG264odzzDB48mLZt27LLLrvQr18/li5dmtXrrI7PLcRnr0mTJqt97hEjRtCvX7+y5ygpKWG//fYDoEGDBtSvX391v/7zW5tIns0f1ZxlfTz00EN+8sknlz2+//77/Ywzzih3TL9+/fzmm292d/dHHnnEAZ87d667u5933nnesGFD33TTTf2iiy5yd/c33njDe/ToUfb7L730kh900EHu7j5lyhRv1qyZFxUV+bbbbuvTp09fqUyDBg0qV6ZsWZ/aRqZ58+b55ptvXq6G8MYbb3j//v3L1Tg+/vhj79ixY1mLwOzZs93dfdmyZb506VJ3j5rG9ttv719++WV2L7YGqq7X96233vLPPvvMt99+e58zZ07ZcbNnz/YJEyb4RRdd5IMGDSrbPnPmTG/evLkvWbLE3d2POuoov++++7J6rdXxuc20qprz9OnTfZtttvGSkhJ3dx89erQfdNBB3qdPH+/YsaP/8Y9/LNtXzbJXczazf5jZN2b23ir2m5ndamZTzWyymf0i2zcQIhV5JbMMInPiT2644QbGjx9Pp06dGD9+PE2bNqV27dpMnTqVDz74gJkzZ/Lll1/y/PPP89JLL632nHfeeSeDBw9mxowZDB48mJNPLp8B9IUXXuDee+/luuuuy+JVhvWpbWR6+OGH6dmzZ1kNYcWKFZx//vlcf/315Y675557OOOMM8paBLbaaisA6taty4YbbgjAsmXL+PFHTePOhup6fTt16kTz5s1XOm6rrbZi1113pU6dOivtKykp4X//+x8lJSUsWbKkrIUoW6rjc1sVI0eO5Mgjj6RWrVpAXOfLL7/MDTfcwBtvvMG0adMYMmTIel9fNlWlWXsIkfpvVXoSqetaAqcSie5FqlVRUREzZvyUUXPmzJkrfZFsu+22/Otf/+Ltt9/m6quvBqBhw4aMHj2abt260aBBAxo0aEDPnj15/fXXKSoqKtcsnXnOoUOHcvjhhwNw1FFHMWHChLLjJk+ezIABA3jsscfYcssts36tX375Jc2a/bQGRFFREV9+WX59jA4dOvDII48AMHr0aBYvXsy8efPKHTNy5MiyZj2IZs/evXuv1Bz48ccf8/HHH7PHHnvQrVs3xo0bV7ZvxowZtG/fnmbNmnHBBRdk/cu7Jqqu13dtNW3alD/+8Y9st912NGnShIYNG7L//vuv8/kqUx2f26qo+LcpKiqiU6dOtGjRgtq1a3PYYYfx1ltvZeEKs2eNwdndXyJyFa/KocD9SZX9dWAzM1t947/Ietp111355JNP+Oyzz1i+fDkjR46kd+/e5Y6ZO3duWe3ummuu4aSTTgJgu+22Y/z48ZSUlPDDDz8wfvx4WrduTZMmTdhkk014/fXXcXfuv/9+Dj30UCC+MMaPHw/A888/T8uWLQH44osvOPzwwxk2bBg77bRTtVzr+tQ2Sn311Ve8++67HHDAAQDMmjWLhx56iLPOOmulc5eUlPDJJ5/w4osvMmLECAYMGMCCBbFWRrNmzZg8eTJTp05l6NChzJ5d2VLVsjaq4/VdF99++y2PPfYYn332GbNmzeL777/ngQceWOfzVaY6Prdr8tFHH/Htt9/SvXv3cuX49ttvmTNnDhCf6TZt2mTrMrMiGwPCmlJ+UYCZrGJNWjM71cwmmtnE0j+KyLqoXbs2t99+OwcccACtW7fm6KOPpm3btlx66aWMGTMGgBdffJFWrVqx0047MXv2bC6++GIAjjzySHbccUfatWtHhw4d6NChA4cccggQzdcDBgyguLiYHXfckZ49ewLR1HveeefRoUMHLrroIu6++24ArrjiCubNm8fvfvc7OnbsSJcuXbJ+retT2yg1atQo+vTpU9aU+fbbbzN16lSKi4tp3rw5S5Ysobi4uOz5Dj30UOrUqcMOO+xAq1at+OSTT1Z6vrZt2/Lyyy9n/Xprmup4fdfFs88+yw477EDjxo2pU6cOhx9+OK+99to6n68y1fW5HThwIEVFRSxZsoSioiIuv/zysuccMWIEffv2LXfDU6tWLW644QZ69OhBu3btcHdOOeWUrF7reqtKxzTQHHhvFfueAH6Z8fg5oPOazqkBYSJV88MPP/gOO+zg06ZNKxsw9N5775U7Zs6cOb5ixQp3d7/ooov8kksuKbd/t9128+eff36Vz5E5kObJJ5/04447ruy8RUVFPnfuXJ8xY0bZYKH58+d7y5YtffLkyVm5xpqsul/figPCSl122WXlBoS9/vrr3qZNG//+++/9xx9/9OOOO85vvfXW9b08Ka/KA8KyEZzvAvplPP4IaLKmcyo4i1TdE0884S1btvQWLVr4VVdd5e7ul1xyiT/22GPuHqNgi4uLvWXLln7yySeXjap2d//ss8982223Lftyr0xmcP7xxx/93HPP9datW/suu+ziI0aMcHf3p59+2tu1a+ft27f3du3a+V133VUdl1ojVcfre8stt3jTpk29Vq1a3qRJk7JR0l999ZU3bdrUN9lkE2/YsKE3bdrUFy5c6O7ul156qbdq1crbtm3r/fv3L/c8khVVDs5Vyq1tZs2Bf7v7LpXsO4hYh7UXsBtwq7t3XdM5u3Tp4hMnTlzjc4uIiBQIW/MhofaaDjCzEcDeQCMzmwlcBtQBcPe/AWOJwDwVWAKcuPbllWwYN24cv//971mxYgUDBgzgwgsvLLf/888/56STTmLOnDlsscUWPPDAAxQVFTFp0iROP/10Fi1aRK1atbj44os55phjANhzzz1ZvHgxAN988w1du3bl0UcfZdCgQQwfPhyIAUQffPBB2XkXLFjAgAEDeO+99zAz/vGPf5QbjLEuml/4xHr9fi5Nv/agXBdBJCfy+XMLuf3s5mxVKtWcs2vFihXstNNOPPPMMxQVFbHrrrsyYsSIciMQjzrqKA4++GCOP/54nn/+ee677z6GDRvGxx9/jJnRsmVLZs2aRefOnfnggw/YbLPNyj3HEUccwaGHHspxxx1Xbvvjjz/O4MGDef755wE4/vjj2XPPPRkwYADLly9nyZIlK51rbeXzh3xtP+D5fK2gmxH5id7LK8lezVnyQ2YiA6AskUFmcJ4yZQqDBw8GIpHBYYcdBlBuCtC2227LVlttxZw5c8oF1MWLF5cF9Ioy0+ItWrSIl156qWxCf926dalbt252L1YkjylgSVUot3aByFYigwkTJrB8+XJ23HHHcttHjx5Njx492HTTTcttX7JkCePGjeOII44AYNq0aTRu3JgTTzyRTp06MWDAAL7//vusXaeISE2g4FwgKuueWJdEBsceeyz33XcfG2xQ/q2RWTvO9Pjjj7PHHnuwxRZbANH//NZbb3H66afz9ttvs/HGG1eayF9ERFZNzdoFYm0SGQB89913PPLII2WJDBYtWsRBBx3EVVddRbdu3cr93rx585gwYQKjR49e6XkrS4tXVFTEbrvtBkTiAAVnEZG1o5pzgViftHjLly+nT58+HHfccRx11FErnfuhhx7i4IMPpl69euW2L1y4kPHjx5eluATYZpttaNasGR999BEAzz33XOrS4omIpJ2Cc4FYn7R4o0aNKhvEVbou7KRJk8rOvaqE+qNHj2b//fdn4403Lrf9tttu47e//S3t27dn0qRJXHTRRdV45SIihUdTqSQv5PMIV02lkkw16fWtSddaRVWeSqWas4iISMooOIuIiKSMRmvnKTUXiYgUroKuOY8bN45WrVpRXFxc6XSeL774gn322YdOnTrRvn17xo4dC8Dw4cPLBkZ17NiRDTbYoGyA1MUXX0yzZs1o0KBBuXMtW7aMY445huLiYnbbbTemT58OwDPPPEPnzp1p164dnTt3LktxKSIisioFG5xXrFjBGWecwZNPPsmUKVMYMWIEU6ZMKXfMVVddxdFHH83bb7/NyJEj+d3vfgfAb3/7WyZNmsSkSZMYNmwYzZs3p2PHjgAccsghTJgwYaXnu/fee9l8882ZOnUq5557LhdccAEAjRo14vHHH+fdd99l6NChHHvssdV85SL5b11vrAEmT55M9+7dadu2Le3atWPp0qUsXry43A13o0aNOOeccwC46aabaNOmDe3bt6dHjx58/vnnALzwwgvlfqdevXo8+uijP88fQGq8gm3WrkquaTNj0aJFQMzZrZi0A1bOjFUxQUepxx57jMsvvxyIxBtnnnkm7k6nTp3Kjmnbti1Lly5l2bJlbLjhhut9jSKFqPTGOnMRl969e5f77JbeWJ9++ulMmTKFXr16MX36dEpKSujfvz/Dhg2jQ4cOzJs3jzp16lCvXr1y0wM7d+7M4YcfDkCnTp2YOHEi9evX584772TgwIE8+OCD7LPPPmW/M3/+fIqLi9l///1/3j+G1FgFW3OuSq7pyy+/vGzZxF69enHbbbetdJ4HH3yw0jm+q3u+2rVr07Bhw5XyVj/yyCN06tRJgVlkNTJvrOvWrVt2Y51pVTfWTz/9NO3bt6dDhw4AbLnlltSqVavc737yySd888037LnnnkAsAlO/fn0gbr5nzpy5UpkefvhhevbsWXacSHUr2OBclVzTI0aM4IQTTmDmzJmMHTuWY489tiyDFsB///tf6tevzy677LLez/f+++9zwQUXcNddd63NZYjUOOtzY126/OkBBxzAL37xC66//vqVzj9ixAiOOeaYlb4PILqnevbsudL2VSXiEakuBRucq5Jr+t577+Xoo48GoHv37ixdupS5c+eW7V+bD2Tm85WUlLBw4cKyxSBmzpxJnz59uP/++1da7UlEylufG+uSkhJeeeUVhg8fziuvvMLo0aN57rnnyv3uqj7XDzzwABMnTuT8888vt/2rr77i3Xff5YADDsjC1YlUTcEG56rkmt5uu+3KPrgffPABS5cupXHjxgD8+OOPPPTQQ/Tt27dKz9e7d2+GDh0KRBPYr3/9a8yMBQsWcNBBB3HNNdewxx57ZPEKRQrT+txYFxUVsddee9GoUSPq169Pr169eOutt8p+75133qGkpITOnTuXO9+zzz7L1VdfzZgxY1bqdho1ahR9+vShTp062b5UkVUq2OBclVzTN954I/fccw8dOnSgX79+DBkypOwO/aWXXqKoqKhsQFmpgQMHUlRUxJIlSygqKiobBHbyySczb948iouLuemmm8pGmN5+++1MnTqVK6+8smzU5zfffPPz/SFE8sz63FgfcMABTJ48mSVLllBSUsL48ePLDSSrbOnTt99+m9NOO40xY8aw1VZbrVSeVS2XKlKdlFs7T9W0JCT5fL016VohOwlmxo4dyznnnMOKFSs46aSTuPjii7n00kvp0qULvXv3ZsqUKZxyyil89913mBnXX3992UjqBx54gGuuuQYzo1evXuX6nVu0aMHYsWPZeeedy7btu+++vPvuuzRp0gSIwF96Az99+nT22GMPZsyYsdIa5+uqJr2+Nelaq6jKubULdiqViOSvXr160atXr3LbrrjiirL/t2nThldffbXS3+3fvz/9+/evdN+0adNW2vbss8+ushzNmzdfaTCayM+hYJu1RURE8lXB1JzVfCIiIoWiYIKziOQn3ViLrEzN2iIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMlUKzmZ2oJl9ZGZTzezCSvZvZ2YvmNnbZjbZzHpVdh4RERFZszUGZzOrBdwB9ATaAP3MrE2Fw/4MjHL3TkBf4K/ZLqiIiEhNUZWac1dgqrtPc/flwEjg0ArHOLBp8v+GwKzsFVFERKRmqUpwbgrMyHg8M9mW6XKgv5nNBMYCZ1V2IjM71cwmmtnEOXPmrENxRURECl9VgrNVss0rPO4HDHH3IqAXMMzMVjq3u9/t7l3cvUvjxo3XvrQiIiI1QFWC80ygWcbjIlZutj4ZGAXg7v8B6gGNslFAERGRmqYqwfkNoKWZ7WBmdYkBX2MqHPMF0APAzFoTwVnt1iIiIutgjcHZ3UuAM4GngA+IUdnvm9kVZtY7Oew84BQzewcYAZzg7hWbvkVERKQKalflIHcfSwz0ytx2acb/pwB7ZLdoIiIiNZMyhImIiKSMgrOIiEjKKDiLiIikjIKziIhIyig4i4iIpIyCs4iISMooOIuIiKSMgrOIiEjKKDiLiIikjIKziIhIyig4i4iIpIyCs4iISMooOIuIiKSMgrOIiEjKKDiLiIikjIKziIhIyig4i4iIpIyCs4iISMooOIuIiKSMgrOIiEjKKDiLiIikjIKziIhIyig4i4iIpIyCs4iISMooOIuIiKSMgrOIiEjKKDiLiIikjIKziIhIyig4i4iIpIyCs4iISMooOIuIiKSMgrOIiEjKKDiLiIikTJWCs5kdaGYfmdlUM7twFcccbWZTzOx9M/tndospIiJSc9Re0wFmVgu4A9gPmAm8YWZj3H1KxjEtgT8Be7j7t2a2VXUVWEREpNBVpebcFZjq7tPcfTkwEji0wjGnAHe4+7cA7v5NdospIiJSc1QlODcFZmQ8nplsy7QTsJOZvWpmr5vZgZWdyMxONbOJZjZxzpw561ZiERGRAleV4GyVbPMKj2sDLYG9gX7A381ss5V+yf1ud+/i7l0aN268tmUVERGpEaoSnGcCzTIeFwGzKjnmMXf/wd0/Az4igrWIiIispaoE5zeAlma2g5nVBfoCYyoc8yiwD4CZNSKauadls6AiIiI1xRqDs7uXAGcCTwEfAKPc/X0zu8LMeieHPQXMM7MpwAvA+e4+r7oKLSIiUsjWOJUKwN3HAmMrbLs04/8O/CH5ERERkfWgDGEiIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKVOl4GxmB5rZR2Y21cwuXM1xR5qZm1mX7BVRRESkZlljcDazWsAdQE+gDdDPzNpUctwmwNnAf7NdSBERkZqkKjXnrsBUd5/m7suBkcChlRx3JXA9sDSL5RMREalxqhKcmwIzMh7PTLaVMbNOQDN3//fqTmRmp5rZRDObOGfOnLUurIiISE1QleBslWzzsp1mGwBlnCo/AAAgAElEQVSDgfPWdCJ3v9vdu7h7l8aNG1e9lCIiIjVIVYLzTKBZxuMiYFbG402AXYAXzWw60A0Yo0FhIiIi66YqwfkNoKWZ7WBmdYG+wJjSne6+0N0buXtzd28OvA70dveJ1VJiERGRArfG4OzuJcCZwFPAB8Aod3/fzK4ws97VXUAREZGapnZVDnL3scDYCtsuXcWxe69/sURERGouZQgTERFJGQVnERGRlFFwFhERSRkFZxERkZRRcBYREUkZBWcREZGUUXAWERFJGQVnERGRlFFwFhERSRkFZxERkZRRcBYREUkZBWcREZGUUXAWERFJGQVnERGRlFFwFhERSRkFZxERkZRRcBYREUkZBWcREZGUUXAWERFJGQVnERGRlFFwFhERSRkFZxERkZRRcBYREUkZBWcREZGUUXAWERFJGQVnERGRlFFwFhERSRkFZxERkZRRcBYREUkZBWcREZGUUXAWERFJGQVnERGRlFFwFhERSZkqBWczO9DMPjKzqWZ2YSX7/2BmU8xsspk9Z2bbZ7+oIiIiNcMag7OZ1QLuAHoCbYB+ZtamwmFvA13cvT3wMHB9tgsqIiJSU1Sl5twVmOru09x9OTASODTzAHd/wd2XJA9fB4qyW0wREZGaoyrBuSkwI+PxzGTbqpwMPFnZDjM71cwmmtnEOXPmVL2UIiIiNUhVgrNVss0rPdCsP9AFGFTZfne/2927uHuXxo0bV72UIiIiNUjtKhwzE2iW8bgImFXxIDPbF7gY2Mvdl2WneCIiIjVPVWrObwAtzWwHM6sL9AXGZB5gZp2Au4De7v5N9ospIiJSc6wxOLt7CXAm8BTwATDK3d83syvMrHdy2CCgAfCQmU0yszGrOJ2IiIisQVWatXH3scDYCtsuzfj/vlkul4iISI2lDGEiIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjIKziIiIimj4CwiIpIyCs4iIiIpo+AsIiKSMgrOIiIiKaPgLCIikjJVCs5mdqCZfWRmU83swkr2b2hmDyb7/2tmzbNdUBERkZpijcHZzGoBdwA9gTZAPzNrU+Gwk4Fv3b0YGAxcl+2CioiI1BRVqTl3Baa6+zR3Xw6MBA6tcMyhwNDk/w8DPczMsldMERGRmsPcffUHmB0JHOjuA5LHxwK7ufuZGce8lxwzM3n8aXLM3ArnOhU4NXnYCvgoWxfyM2gEzF3jUYVD11u4atK1gq63kOXbtc519wOrcmDtKhxTWQ24YkSvyjG4+93A3VV4ztQxs4nu3iXX5fi56HoLV026VtD1FrJCvtaqNGvPBJplPC4CZq3qGDOrDTQE5mejgCIiIjVNVYLzG0BLM9vBzOoCfYExFY4ZAxyf/P9I4HlfU3u5iIiIVGqNzdruXmJmZwJPAbWAf7j7+2Z2BTDR3ccA9wLDzGwqUWPuW52FzpG8bI5fD7rewlWTrhV0vYWsYK91jQPCRERE5OelDGEiIiIpo+AsIiKSMgrOIiIiKaPgLLIGynYnIj83BecsqMlf3mZWE95DpXP4TzCzjrkuTC7UlPd4spZAjVL6GTazTcys2ZqOLxRmtmHG/9slOTpSI1WFyUdmtoG7/5i80B2ABcBid/8qx0X7Wbj7jwBm9gdgkbv/PcdFyioz2wE41MyaAEcDu+a4SDlRmrfAzHYBvidmekzLbamyK/ksrzCzOsBFwDXAD4WcsyHj+6sYuBmoa2YOnFADvsO6mNkWwL7Aj+5+rplZWl5vBef1VBqcgPuA74BOwCNmdl1aXuTqYma1ki+zbsQCKS3NrB0wqDTPer5z98/M7DPgUuAloLGZrXD3b3NctJ9Nxhf4AOAQYCGwwszOc/eCyQSY8Vn+K/FlvTypTZXksFjVKuOaBwG3AhsCl7n7V2ZWN1nsqOAkrQUbAAOAbsD/g3I3oTkP0jWhSbLaJYuD1HH3U4EGwH/d3ZPaVsFKAvPWwN+BfwK3AIuAv5rZ/jktXBZkNOVOBo5J/j0P6GVm9czsbDOrUhL7fJYE5m2Bc4DDgY2AL919vpk1S1tz4PpIbjRbu/tpyY3mcDN7ysxa57ps1cHCDkQK5s+Ai4Gzkt1/MrODc1a4alJ6s+nuLxPJtV4BOprZSWa2eXLYnmZWL3elVHDOlo2A583sOmCcu79gZjsBV5vZpjkuW3XrBrySZIr7JzCEWPTkj2Z2upnVysf+yqRVwM2sF3CBuz/j7pcCTwMHAzcBFwJv57KcP6NtgH8BewGN3P3PyfZzgeKclSr7FgHfmtn/AQOBccA7wImF2B/t4TNgMTACuN/d/5MEqaOAqTktYDXI6Io7HXgOOJlYL6IzcKqZPQSc6u5Lc1dKBedseYX4wu4B/CXZdjkw290X5apQ1Skj4H4OHGJmvd29xN0/Jb7Q3gJ+AWyc6+ahdZG0CjQkWgOGQFl/65vEsqfDgH3dfXbOCvnz+hDYFngQuBYgaebe1d0/zGXB1kfFG0d3n0K8tj8Cl7j7fcn/l7v7ihwUMesyB3Emg8AaAg8AdYFtzew84GEiVXPevraVyRj8dgTQH1jg7gvc/S4iFehyYtzQebkrZVD6znWQ0Qe3NbCZu39kZn2Bg4jlMw1o4u6/zmlBq0FGP3N94H/EuIV9iYA1k+TuG+gJXAfc7e7jclXe9ZG8pr9291OTL6yDge7AUe7+eG5LV70yXudGxJq5C4iaxcFEv+QSohbdz93fy11J10/GZ/k0YlT+HsBF7v6fZP/ZwDHuvkfyOOd9kdliZrcBPwDHJT/vEn2v3xBBa2gOi1dtkgWcxhNdVYuBY4nr/kOavqsUnNdSxod5T+AGoA7RV3MMsZxmG+JOe6K759Mi4GuU+cVkZiOJ624L3Ek0C/2RGCz0HPAREag7u3teDagpvU4za0E01W9DXOPdwH5AC3e/NpdlrE4VXuexwBwicN1AfHGXBux33f2dnBU0S8ysLTCU+Az/E3jU3a9J9h0BfOLuk0tvWHJY1KxJ+pLPdvf9zext4OZCDcYVJSO0hxJjSHYgvqu+IQY7nuDu3+SweGUUnNdCaWBO/v8QMMzdx5jZP4laxEnu/lROC/kzMLMbiRuQp4hm3/2BWRlf6JsRIz+HuPvzOSvoWsqoLTYANiFWYZsH7Oju75nZNsC/gQvd/dlclvXnYGbXEjdbQ4DXgV+4+zwzq+PuP+S0cFlkZpcCE4EVwEB372FmmxDL3w5NbsYLpsYMYLHS4DtEa8hu7t7PzJoTNchrgYUFdr1l393J4x2BS4BH3P3xZPDfXcCeabkBK5hRltWtQmA+jvjbTQdw99+Y2THAk2Z2vrvfmLuSVp+kf64u0Wx/AfFmHuLuX5rZr8xsK+ARYkrZNe7+Qe5Ku3aSL9/SD+WdRM1wITGnd3Cy/WrguUIOzBmtBkYEq9eIG7Ark8Dch5gyNyifv7wrBNvXiGbdNkCvZNtJQPekz5l8vtaKzKwr0QV1CdDA3XdPdv0B2MDdF+SscNUg46a7OdG6txz4zN1PSPa3J1r5fp+WwAwaELY2TjOz7ZL/1wbqA30tySzj7g8CDYlpRQUp+YL6AfiWGMXZxN0HJbuvBTZKRn+W5FNghnLzGy8ENgaOAK4i+uHOSJrCLnf3C3JXyuqV3IB6MkCoPfAi0ZRdy39KLnM+8EW+B6vkOndPRmW/ANQjuqW2tpgaeQIxraigsuAl4yhOdPdHiff2zmbWP5lp0pkYfV9QKtx0v0OMl+gNZVnClgN3pO2mWzXnKkjmu30CzDKzU4HhwKvEiL6TiRrzJHefnY/ThlbHYn7rMcB2wGPu/qKZDSb6IBskI3bbA9PdfVgOi5ot3wOPu/sSM/uQ+CDfADR397dyW7TqldHsdx0xCv9vxPvczew+Ysrgx+4+MkdFzLYviKx+rdz9aDO7CLgHeBn4P3f/tGJzaD4ys2buPiNp2WoDPAng7ueZ2QRgT6L/dXghdVdkMrPdgGnufo+ZTQTOSHb1AZ5197/mrnSVU5/zGphZU6Cxu08ys52JtH7fAvcSAwn+H9HnOsjdX8xZQauJmf0bmE18kfUmRrI+ZZG4oDsxQv1l4AF3/y7fvswq9iVaJBW5m7jOB5JtzwC3uPu/c1TMamdmLdx9mpl1B64ADnH3pckX+i+JPvh5wJNpavpbFxUGvJ1D9KUflzyu7+5LclrALDKzvYm0s2cR31XnA48T08QWJMcUVH96qYzm7E2BLYHrgS2Ah939zqSZezTxXk9dRkMF5zUws/7Em3p4xgt6NNAaeJ6YH7grMTo7b4JSVZjZ+cAe7n5Y8vhYovb0NIC7T88cwZpvgTmTRQaoFURNeQfgdKAp8B5Q7O4H5LB41crM/gLUdfc/JS1DvyPmqV/h7tNzWrhqYGa/IvpcF7v7HDO7HVhKNGOX5PvNRymLHOHjgYvd/YVk20FEcpVXiJvQGfn6mV2dCmOEHiTe08cTWe4uBZ4gvrvHufvgVZ4ohxSc18AiU86uRB/UYmJQ0EzgMOBAYq7vQHdflqsyVodkxPLjwMdEX+tXFnNBbyCaOn9BjHAd6e73566k6y7jzvoQ4gtrDtH/NJMYBHYQ0Z0x0d0X566k1ScJVIOBvZKWj82ALsSUsY2Af7v707ksY7ZkDHYbROSCn0LUpF4lBoQd7QW0mIeZ/R04FNgm84bDzIqI8RRbAzemra81m5KpcO3c/XIz2wj4NXET9gmRJGpgTgu4GgrOq1Ch6cuI2tTxxKCJ+9z9ETPrBGzh7s/lsKjVxiI3+GVAE6Kf/RLiRuTJpLmzpbu/mssyri+LZCpvAr8lmu63Bs4EXi1t1i5USc3qfeJaT8ycImWxNOavgN2IAD0ih0VdL5U12yY3n/WIm+yWRLDeEuhdCK0FyeyRc4gupx5EEP5nhWMGEv2wD+egiNXOIoXyh8BYdz+4wr7Uz1lXcF6FjLvsVkTawreJea+/JuY/ziVyLn+Xw2JWi4pvXDM7GTiRGCx1INEEuixjf972WSX9UX8Hjkv6WOsCfYnAdBqxOlFeXtuaJDWrhsSc9UXEakzve7ISkUUGvL2AFz0liRnWhf2UOOgEImnO/4DXPMkGZcnqS2b2Z+Atdx+bw+Kut+Q9/Shwvru/mUx/u4AY9HW25zhndHWqZAzJrkRimS+Aw0pbwPLhO6tgpghkU8aUkl8RCRiOAz4gvshGE9OGViSPC0rypi3tQ77TzLq6+71Es/4iIrPOLy1jEYC0v8krMrNuyU0XHrnPVwCPmVn7JDDNJm7IPN+uraqSmnEzdz/K3Y8B5hMDZo5JunJw99nuPirPA7Mlgbk9UZN8jpjL/DszuyYZyVy6LOLOQCGMLVhODPh6M/kuG01UKGoBz5nZfrktXvXJaO3sbmZ7EMmRWhItRB+bWb/M49JMNefVsEhdeCExjehMdz/QzDZ3928t5jbnVVrKtWEx73E7j8xBmYMrLiNWJTpr9WdIJzO7nlhzuy1wrbvfajGP9f8BpxBzPzsA5xV4X1xj4jtqbsa2Q4DfE838Q4EPC2WwkJndSmR3W06MRr+CGD8yh0hM8RGxCtO/8vlzvbrvpeSG+nhiQNQh7v7uz1q4apYxhuR3xM3ICiLF8BfufpWZHUX0N/8iH97XCs6VSL6sNyW+qCYSgycOd/fPzOwu4G13/1suy1idLBJuDAHO9ZjruaG7LzOzbdz969K+yXwbnZ1MK7ne3bsmNefriFGb3wBfA18Rg/+meJ4lUVkbFW62yqXitFjo4kYiyc5vPE/nvWY0ZZc2WXclWr8eAC7zmBp5NzDZ3W/PbWmzzyK5ykJiEOvj7j4jY9+27j4rZ4WrRsl310vE3O3viYGrxwMvuPuofPrOUrN2BvspE1AdjzmAGwA3AzckgXkXYv3iR3JVxupiZgdZ5JvF3ecDnwKHlQbm5LDBSTP3D8lxefEmh7JawzVEtwTA9sQ87Z2JlZZ+C3zl7o/UhMBsZg3N7GbgDotEMgC4+1x3P57or8zLwJy42sxGAecntckJSX/jHKCPxdS5rsTSiJhZwSRkSkYo70v0rW8P/MnMyprrCy0wm9mOZtYmeVibuMlu5O7L3f114D/AockgwLyh4JxImkR+tEhdeGcyqOJuouZ8jJndS2RMutrd5+SyrNWkITA/oz/qUSL5xBFmdqTF9JP67j4hZyVcP41Ixg0k/U7XEK0h5xFjCNoS04cKWsYN1a3Ea/4UscD8w8l7v/S46TkoXlYkA7+6EZ/XXsDRZtbEYuGSq4hkOtcQKRu/Tm5Y8rYpG8puPkkGNG5GtHrdSuSM/hA4yswuyBwrUggsZhzcS7R0koyPeAW4POOGZBlQz92/y6sKhZq1yzOzx4hm68uTx7WA3YnBFEvyODitUsbI9NIEI98AA4BWRI1yI2Ixi0vcfXE+NQ1lSr64ziBqFQac4u5fJvueJb6sR6/mFAXBzH5JJNI5P+muqA8MIvKJn+ApWtN2bVlk9HsNOMLdJ1qsrLUXUWOeD9zj7q+aWSMvkCVdM0ceW6wY1wX41n9KHrQVceO5wN2fyF1Js8/M/gYsdfdzkppxEfE6H0nchNVLfk5198m5K+naU3DOYGa/IFbfOcjMehGDhPYEjirUwUFm1tgjS1Id4sv5aWKAzEFEZqF/J8fVdveSfA3MmSyWhzub6I/7J3Hdxe5+VE4LVs3MzIj1x4cB7YhBUaNLuy3M7DfE2sVv5K6U68cio9/xRL95CdEt1ZO4wTwU2N7dz8xdCbMv4+b6dGKq57XEtLi5xNSpgsgRXpGZ7UusjNfDI1vhw8AEd7/eYsnPDYBdgM89hek510TBmXJv7pbAGGIpyFnEiNWtgL2J0buFlgWsDrE0XiciVeXDniy4bmY9iZGN3xKjWH/wlE/ar0zGCM6uRED6mlib+H9Eis59gB2BXxVod0Vlcz+NaBn5DXFz8i93n5er8mVTUns6AuhI5Lwfl3RdYLGq3CgiE9gXuStl9iXXdg8x4PG5ZNsgYs7+5R7TIQtK0gV3GjE9bjugq7v3SPalfh7zmtTo4FzZC2iRO7snMMLdFyTNJnPd/c85KGK1SwbBvUT0uR7t7s9k7CsiRuxen6vyrY+MwU+bE8sCPgwcTkwVesTdx5nZnsByd/9vLstanTJuPo8EtiH6JO8lMmL9H9EM+Bd3/yyHxcyqZMBX6brrk4mbkOuIBDpn57Js1cHMfk3M496UWEv9qWT7wUBDdx+ey/JVF4vFiE4hkgaNBG7PaAnK6wCt4BxfWqcRo3a/Bp4FJiW1rUuA/dz9VzktaDXIqFHWJpLC/0BkAfuvJ3OYzewPRD/sMsuDdHerYrGwwSRi8NPTRLPuAcQd9y3uvjCHxatWGa9zd+AOoubYgEhbeRHxfh9EzPmeseozpVvGjVhnYj3uFcQo3VOBFkTryHbuvmvm8TkrcDWwyOj2G2KsyPvA/YX43s54T2f2tfchWkzeA8YB7+X9IL+aGpwzAvMewO1EspG/Ek2eE4i7sAZEc26hNYFlvqmvB27zWO91JyJBwS5ENrAn3P26HBZ1nZnZ7kRiiQXE4KexxMjVG9z9eTN7ihiJ/+d8vruuKjMbQqThHJI8/jUxpqJfvt50lcoIzLsQNeS/ArcB3TyyZO1DNO/e6TG/OW9vNEtlBKj9icFeexCtINOIikYvYhDcn/P9WlfFYi7390Sl6kFiDYCziMVMBrn7Ozks3nqrsVOpMr6QTyX6Hjci5vY+kWy7ilhSrqACc8Kg7M29fRKY6xKB7HRigYtHSgNz0keZN5LA8ydiIfUmHos2fEckZaiXHFYb+GehBmYza2CRdKXUZKLrAgB3fx6oS/S557WMGvCNwB+AGcBzSWDeBvgvMTBqUnJjmtfBKrkZWZEMerqZWLr2PuJmq6+7/wv4B1FzzutrXRX7aS73YmL53utJppARLWR5n/2sYCber40KfRH38FNN+QR3n2WRxvAtz+OcwquT1DK2J5p2e1vkWT6f6I98yDOyn+Vj819SM65L9DnuZjFHfQLwEDDMzD4kVmJ6P5flrGb7Al9apOnchGjO/pdFYo5zgR2IQTRv5a6I2ZHxHh1P1KL+TnTVQAx4XOLuN0N+5FRek+Tza0TynCnu/iSAmY0BHjezT0sHdhaSjNaCzLncr9pPK6idQSzvWhAZ32pkcE6asw8EDiGab3sQAylOtJjvWkzUIAtKcre5xN2fdPfPzWw8MUimIZHWcAXQz8xGAgs95FVgtp9yC+9ItBDsRlzf4+4+LLnmZu4+KZfl/Bn822Pq21VEH+xQ4n1+GfAk0Sd5rUc2uLyUcZPdApiabH6HmA75lpl1IG7QDsxVGbMt6YYa7LG++itEkqCuRB/rbDP7K/H3KCgVWjyuIZnLTdxkTzKzWcA8onWsINSoPmeLzFAL3X2smb0M/IsIztsTqfx2IBKl/90LcI1Ti3ncnxLzPccCjYn+5TfdfZqZnQ+0dfcTclfK9WeRN/shd2+fPO5P9EVNIul3zGX5fi7JYL/WRDKGrYk+9mFEprfvc1m2bDKzh4ikG6cmgxjPI25AWhJrrw/JxxagisxsY+AmYk35q919tJn9ieiueI4Y1HkekVymoPIyZIwRqjFzuWtacD6WGPj1OdFs/Wczq0c0751CNP2964W93ukGxKIWuxAZv55Itvch5jX/2t0X5fMb3czaEh/cc939rWRbX2J08qke+XYLUsbgqP5AC3e/IplK1ovIdLeYuHF5M6cFzSKLxQ4uJ6Y//ieZRrUT8JG7f5jTwmVJhUGc1wLHEDdalxE3X3uRTBtz97/nrKDVyGrYXO4aFZwBzKwF8BdisFBPd3852X4DUNvdz8ll+apDxhf2xqU1JjM7GhhI9MWeTaxfvJm7T8630azJl/POwPzSL+OkRrGUyBj0ajJdbkd3H5jDov4szKyYqDke5u7vJ4+/AJoRfbF3uPu0XJYxG6z86loDiPdxv4pjCSqMMclLGTXHS4g88XOJpU0bA6cVyk3I6lgNm8td44IzlNUeLyTS/D1GDAb7CzHNZnwuy5ZtFe64RxAf6F4e6e62A64kVmfq4Xk6z9Vi7enzgeFEM+41RJPmDkQfVCciN3rfQh3kZ8nSiMn/LwG+c/fBZnYekR+9GZE1a0E+N2kn/cjvEv3orxGtQJ8AzwD9iCUC/+SxAlVBSUaePwPs5e7zLfL+X0fUHO9y9ysL4UZkdayGzOWGGjqVyt1/dPf/IwaL9CQyZP2z0AJzonTaVF9i0Mx84B0zO8fdv/BYHvDKfA3MiVHAq0TWqyuBNsSqRPsSU+RmAGcWcGCuBfzGYuESgJeB35vZROKmZG9ihbVd8zwwb0mk5DTiugYQr++vgCnJv6cSr30hmkesMNUVIGndGkpM/5ySbCuYwGw/rbS1v5kNMrPXiPXWS5MJdQAutAJbaatUjaw5Z7LIxdvH3YfluizVJWnKfwH4pcec5o7EgLDXgePc/bvkuHzuZz6D+GLenWjqG08s7NCCGF9QcGtwl7JIs7oz8DbR9ziWGEexvbs/k7z+TxDv87xt/kym0JSOEfkLkarxhWTfjsSc7SJPVpQrRMlN9jXEzda1RNKg+u5+QU4LlmUZXXGbEPPUzyNWnDqU+DxfapFo6Fsv0PXXa3xwrgnMbFvig3x8RhP3EcQHeymRP/vTHBYxK8xsIJFIpQ+RhvTyjKlVBalCt0UfYl7vc8R63DOI1rFxxFSym3NW0PVU4TpbEa0iHYlFah5w988zj8u3cROVyQhQexOr431K3HRuTAx4/IFoQTjW3WfnrKDVxMyMaLI/wt2PTLZtDTxOjJsouLncmRSca4CkxvEgkYr06GTbqcQHuynwTCE06ZvZrsQI1gbuXpRsK+g+uFJmtpe7j7dYyON4Yorgk8T0sSaeZ2vZZqr4GprZJh7rivcguqU2JN7DY3JWyCyrUHN8hWjl+pFYTe0Zd3/SzDYkWrKX57Ks2Wbl53I3AwYTGcDec/clZnYCsIO7X5bLcla3GtnnXOgy+mp2NLNuwAHu3geYb2afmNltwAVEf1ULoH3uSps9HusQHwaMN7NuSe2pYANzMrCxtMZ8oZk1SmYfDCRG8w4gEu3kfSpDADP7i8UiJnea2fHJdJrBwGxinEHByOhe+gMxN/804FYid3ZPM7sC2KIAA/PGJEmDzKxPMhbmTWIk/jHJFMHfE+MqCppqzgUmo1mvLvEGfpMYwdwQOIhYn7oekeawPXCRu3fPVXmzLWkKuwLYwN0vznV5qpuZ1ScGww3wyCVdx91/SPb1AuYkNy15KaMGuR8xp/cUYoT298Sa65e4+2dmtpG7/6+QWkqSsQQvEQtY/Mpjdbj6RNrOFsB1hXKtoLncFSk4Fygzu4aYt3x68vgsornzYHf/OhkIdwzwsrt/nMOiVgsz29TdF+W6HNUtGex1O5EX/puM7b8l+pnz9m9gZvU8SQhkkYdgOJEN63CiJvUwMef1kEIYMwGVNuE3J1oHtiGmiL2YbC/LWVAoMioWNXYudyY1axeuKST5hpM5sLcRox53BEhGaN9fiIEZIJ+D0trwSCbyMXBs0j+HmR0AnECsxJWXzGwz4E9JHzrEilOfEjWoa9x9JtEX+3+FGJjNrK2Z7ePu05MuqfuAm8zs1uTwJTkraDVJAvM2xBKvf3H3K4kKxBvAs0nQzrtV8tZVjVz4otBk3HFm3nXPAK4ws8XufndSU+5GrMwEQGnzp+QP+2llnk2IBVq+JOZ8HgpcYJEtbSfggnydFpdoTsxh7pPcdDzm7t+b2QxicdOTzsYAAAUlSURBVJaWxJznSyG/pwFWlDTpNiFWVPuYqDHfbTFvvT8U1nzmCjLnco9L3utDiZXVCm4u9+qoWbsAmFl9d1+S/L9sCkkyenkokUt8OTDN3c/NXUllfVSoWT1JTCM6gGji/ZSY/7sZ8LkXQP5wiwx2+xMZ3r4imrEXErXo+cTKW+MKLDB3A25x993M7AHiJmV7YJC737raXy4QVkPmcq+JgnMBMLMXiQU7zkoe1yFuMEuSx4cBr7v718njgvkyq4nM7GygpbufZWafE+lY3y8dFJXr8q2PpI+1CzGYsa5H0pwDiL7HzYDR+TzAbU2SkfdLiEGbA9z9EDO7HPgTsI+7v5bL8mVbTZ/LvTrqcy4M+wIbmNlbZrZD0lxdWsMqTXlY1kelwJy/LJaB3Ah4PmnuuysJzPsBNyaj9PPZP4l0rPcQ08NeI5ZIPICYLvWQxdKvBcHMNjKzbSxW0sLdRxPdFMX8tEb1V8ClBRyYNwFuITKA7UlkA9vR3fclpkYeXNMCM6jmXFDM7JdEZqgr3f2WZNvLxIjsi3JaOFkvySCY+km/657An4mkMgcn+8cBTyQD//JWMkf/L8SUqT2JGtTmwIFEspG9gRM9WQo035nZcKJ/+Rsi/epgd19usfb6hUTSke5Ad3efl7uSVh8zuwyY7e5/S25SehDjJhYQc7y/ymkBc0TBucAk8yAfIdbt/ZDIp/3rZF/BzAGtKZIBXqcQzbotiAF99xJLP3YhBv41AWq5+xG5Kme2mdkewGjgYne/J9mWuURk3nfNmNmZROAdSAzW7Alc6O5zk/3diKli3xVarblUTZvLvTYUnAtU0vQ3HNjJ3adaAeQaronM7AngPSIN51zgNmKJxN8SzbzNiC6L8cn0uIKRcaNpwNGFND3OzLYiEgH18J8W77gXqE/UousBL8L/b+9+QauMwjiO/x5EQYPbRASTyWA1iehwK2qQxYWJgkHEMMaaxWEwiVGLoAbR5B+wKCzJLCKrCkO0K2JSEORnOOeOl1vU+d57zn3v9xNfbnhuet5z3uePVm1/LhXnIIxzL/e/IDl3WO5v/kliHk35ZHXc9nzf8/tKFcyzXb3qbMoDVW4qzVMe6YK3poiYlvRI0nXbtyNiQ2mpwyelCu3dSsM3RvqGoKm/l1vSvsbLyUVJlySt2V4c95s+kjNQocbJasZpocU2SdsbE7NWJa3Yfl0yzmHpTQvrwnV2U74deKK06vSu7aXG3IId7t7s7N5/2+zlVhqicyUXNh6WdNb2ctFAK0C1NlAhp1GcJyQ9jIgl279yctqVf/Jeqbp1LPReSrqUmCXJ9nfbpySdk7QQEWd6p8WuJWZpcwrYEaWXzvOS3kraI+lFRCzaXicxJyRnoFK2X0k6KOlkRLyMiEmnlXmTkmYkfSwbIdpi+5nSgppbEXGtdDwDtl/S1YiYkzRh+5hSkeONiDhaNrR6ML4TqFie/HY6IuYlbeRCvzlJj22/KRsd2mT7S0QcUGod64yI2Km0FW/K9jvbT3Nr4LI63sv9P/jmDIyIiNirdA34w/ah0vEAf4Ne7q0hOQMjJJ84Jmx/Kx0L8Cf0cm8dyRkA0Lpx7uVuAwVhAIDWNToOHkTE5fx4Wun78gelfeOzSmsi0YeTMwBgYMatl7stJGcAwMDl1bV3JF2w/bx0PLUjOQMAhiJ3HKxLumd7pXQ8NSM5AwCGJnccTNn+WjqWmpGcAQCoDNXaAABUhuQMAEBlSM4AAFSG5AwAQGVIzgAAVOY3IucP1dKH2q0AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11f2f43400>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1219c1c240>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11fa8fb780>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11f8bb99e8>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1276b09518>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f12190ab3c8>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_stats(Acc, \"Accuracy\", 8, 55)\n",
    "plot_stats(F1S, \"F1-score\", 8, 55)\n",
    "plot_stats(Prec, \"Precision\", 8, 55)\n",
    "plot_stats(Rec, \"Recall\", 8, 55)\n",
    "plot_stats(FPR, \"False Positive Rate\", 8, 55)\n",
    "plot_stats(FNR, \"False Negative Rate\", 8, 55)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random Features for Large-Scale Kernel Machines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(293402, 286)"
      ]
     },
     "execution_count": 238,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.values.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_input_fn = tf.estimator.inputs.numpy_input_fn(    \n",
    "    x={\"features\": X_train.values},    \n",
    "    y=y_train,    \n",
    "    batch_size=180,    \n",
    "    num_epochs=None,    \n",
    "    shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(62872, 286)"
      ]
     },
     "execution_count": 240,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_validate.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [],
   "source": [
    "eval_input_fn = tf.estimator.inputs.numpy_input_fn(\n",
    "    x={\"features\": X_validate.values},\n",
    "    y=y_validate,\n",
    "    batch_size=X_validate.shape[0],\n",
    "    num_epochs=1,\n",
    "    shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(62872, 286)"
      ]
     },
     "execution_count": 242,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_input_fn = tf.estimator.inputs.numpy_input_fn(\n",
    "    x={\"features\": X_test.values},\n",
    "    y=y_test,\n",
    "    batch_size=X_test.shape[0],\n",
    "    num_epochs=1,\n",
    "    shuffle=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using explicit kernel mappings with the linear model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 393,
   "metadata": {},
   "outputs": [],
   "source": [
    "data.rename(columns={'Dst host count': 'Dst_host_count', \"Dst host srv count\" : \"Dst_host_srv_count\",\n",
    "                    \"Dst host same src port rate\" : \"Dst_host_same_src_port_rate\",\n",
    "                    \"Dst host serror rate\" : \"Dst_host_serror_rate\",\n",
    "                    \"Dst host srv serror rate\" : \"Dst_host_srv_serror_rate\",\n",
    "                    \"Same srv rate\" : \"Same_srv_rate\",\n",
    "                    \"Srv serror rate\" : \"Srv_serror_rate\",\n",
    "                    \"conn len\" : \"conn_len\",\n",
    "                    \"src bytes\" : \"src_bytes\",\n",
    "                    \"dst bytes\" : \"dst_bytes\",\n",
    "                    \"conn numb\" : \"conn_numb\",\n",
    "                    \"Serror rate\" : \"Serror_rate\",\n",
    "                    \"cli src pool\" : \"cli_src_pool\",\n",
    "                    \"registered src pool\" : \"registered_src_pool\",\n",
    "                    \"well-known src pool\" : \"well_known_src_pool\", \n",
    "                    \"dst port\" : \"dst_port\",\n",
    "                    \"src port\" : \"src_port\",\n",
    "                    \"Conn state\" : \"Conn_state\" \n",
    "                    }, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "metadata": {},
   "outputs": [],
   "source": [
    "feat_column = tf.contrib.layers.real_valued_column('features', dimension=len(X_train.columns.values))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {},
   "outputs": [],
   "source": [
    "kernel_mapper = tf.contrib.kernel_methods.RandomFourierFeatureMapper(input_dim=len(X_train.columns.values), output_dim=3000, stddev=5, name='rffm')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "metadata": {},
   "outputs": [],
   "source": [
    "kernel_mappers = {feat_column: [kernel_mapper]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "metadata": {},
   "outputs": [],
   "source": [
    "! rm -r kernel_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = tf.train.FtrlOptimizer(learning_rate=30, l2_regularization_strength=0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using default config.\n",
      "INFO:tensorflow:Using config: {'_task_type': None, '_task_id': 0, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f11f2f43a20>, '_master': '', '_num_ps_replicas': 0, '_num_worker_replicas': 0, '_environment': 'local', '_is_chief': True, '_evaluation_master': '', '_train_distribute': None, '_eval_distribute': None, '_device_fn': None, '_tf_config': gpu_options {\n",
      "  per_process_gpu_memory_fraction: 1.0\n",
      "}\n",
      ", '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_secs': 600, '_log_step_count_steps': 100, '_protocol': None, '_session_config': None, '_save_checkpoints_steps': None, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_model_dir': 'kernel_train'}\n"
     ]
    }
   ],
   "source": [
    "estimator_kernel = tf.contrib.kernel_methods.KernelLinearClassifier(\n",
    "    n_classes=2,\n",
    "    optimizer=optimizer,\n",
    "    kernel_mappers=kernel_mappers, \n",
    "    model_dir=\"kernel_train\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Casting <dtype: 'int64'> labels to bool.\n",
      "WARNING:tensorflow:Casting <dtype: 'int64'> labels to bool.\n",
      "WARNING:tensorflow:Trapezoidal rule is known to produce incorrect PR-AUCs; please switch to \"careful_interpolation\" instead.\n",
      "WARNING:tensorflow:Trapezoidal rule is known to produce incorrect PR-AUCs; please switch to \"careful_interpolation\" instead.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "INFO:tensorflow:Saving checkpoints for 0 into kernel_train/model.ckpt.\n",
      "INFO:tensorflow:loss = 0.6931474, step = 1\n",
      "INFO:tensorflow:global_step/sec: 165.542\n",
      "INFO:tensorflow:loss = 0.07498501, step = 101 (0.605 sec)\n",
      "INFO:tensorflow:global_step/sec: 177.754\n",
      "INFO:tensorflow:loss = 0.043834057, step = 201 (0.562 sec)\n",
      "INFO:tensorflow:global_step/sec: 178.638\n",
      "INFO:tensorflow:loss = 0.07361633, step = 301 (0.560 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.867\n",
      "INFO:tensorflow:loss = 0.039425515, step = 401 (0.568 sec)\n",
      "INFO:tensorflow:global_step/sec: 178.733\n",
      "INFO:tensorflow:loss = 0.06621093, step = 501 (0.560 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.053\n",
      "INFO:tensorflow:loss = 0.06463543, step = 601 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 181.203\n",
      "INFO:tensorflow:loss = 0.050641526, step = 701 (0.552 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.969\n",
      "INFO:tensorflow:loss = 0.020946557, step = 801 (0.568 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.241\n",
      "INFO:tensorflow:loss = 0.0144449035, step = 901 (0.558 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.158\n",
      "INFO:tensorflow:loss = 0.030149026, step = 1001 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.089\n",
      "INFO:tensorflow:loss = 0.03739189, step = 1101 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 173.467\n",
      "INFO:tensorflow:loss = 0.009334033, step = 1201 (0.576 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.701\n",
      "INFO:tensorflow:loss = 0.03874248, step = 1301 (0.557 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.442\n",
      "INFO:tensorflow:loss = 0.073412456, step = 1401 (0.554 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.013\n",
      "INFO:tensorflow:loss = 0.036982127, step = 1501 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.327\n",
      "INFO:tensorflow:loss = 0.05070571, step = 1601 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.631\n",
      "INFO:tensorflow:loss = 0.040981885, step = 1701 (0.566 sec)\n",
      "INFO:tensorflow:global_step/sec: 173.934\n",
      "INFO:tensorflow:loss = 0.02187066, step = 1801 (0.575 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.739\n",
      "INFO:tensorflow:loss = 0.007973402, step = 1901 (0.569 sec)\n",
      "INFO:tensorflow:global_step/sec: 178.388\n",
      "INFO:tensorflow:loss = 0.040816944, step = 2001 (0.561 sec)\n",
      "INFO:tensorflow:global_step/sec: 178.162\n",
      "INFO:tensorflow:loss = 0.025049793, step = 2101 (0.561 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.013\n",
      "INFO:tensorflow:loss = 0.010873211, step = 2201 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 177.598\n",
      "INFO:tensorflow:loss = 0.034152605, step = 2301 (0.563 sec)\n",
      "INFO:tensorflow:global_step/sec: 173.519\n",
      "INFO:tensorflow:loss = 0.03329205, step = 2401 (0.576 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.12\n",
      "INFO:tensorflow:loss = 0.06420751, step = 2501 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.732\n",
      "INFO:tensorflow:loss = 0.07139455, step = 2601 (0.566 sec)\n",
      "INFO:tensorflow:global_step/sec: 178.589\n",
      "INFO:tensorflow:loss = 0.014136229, step = 2701 (0.560 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.78\n",
      "INFO:tensorflow:loss = 0.044546895, step = 2801 (0.566 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.813\n",
      "INFO:tensorflow:loss = 0.02749118, step = 2901 (0.569 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.08\n",
      "INFO:tensorflow:loss = 0.0499808, step = 3001 (0.568 sec)\n",
      "INFO:tensorflow:global_step/sec: 173.651\n",
      "INFO:tensorflow:loss = 0.008715525, step = 3101 (0.576 sec)\n",
      "INFO:tensorflow:global_step/sec: 167.041\n",
      "INFO:tensorflow:loss = 0.015999658, step = 3201 (0.598 sec)\n",
      "INFO:tensorflow:global_step/sec: 173.205\n",
      "INFO:tensorflow:loss = 0.07184455, step = 3301 (0.578 sec)\n",
      "INFO:tensorflow:global_step/sec: 173.067\n",
      "INFO:tensorflow:loss = 0.056297906, step = 3401 (0.578 sec)\n",
      "INFO:tensorflow:global_step/sec: 169.669\n",
      "INFO:tensorflow:loss = 0.028835505, step = 3501 (0.589 sec)\n",
      "INFO:tensorflow:global_step/sec: 174.673\n",
      "INFO:tensorflow:loss = 0.042850547, step = 3601 (0.573 sec)\n",
      "INFO:tensorflow:global_step/sec: 174.055\n",
      "INFO:tensorflow:loss = 0.0148154255, step = 3701 (0.574 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.085\n",
      "INFO:tensorflow:loss = 0.048828226, step = 3801 (0.568 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.342\n",
      "INFO:tensorflow:loss = 0.064221874, step = 3901 (0.571 sec)\n",
      "INFO:tensorflow:global_step/sec: 174.051\n",
      "INFO:tensorflow:loss = 0.005873191, step = 4001 (0.575 sec)\n",
      "INFO:tensorflow:global_step/sec: 174.98\n",
      "INFO:tensorflow:loss = 0.029102368, step = 4101 (0.571 sec)\n",
      "INFO:tensorflow:global_step/sec: 173.497\n",
      "INFO:tensorflow:loss = 0.053066917, step = 4201 (0.577 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.268\n",
      "INFO:tensorflow:loss = 0.0060510905, step = 4301 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 172.296\n",
      "INFO:tensorflow:loss = 0.008735152, step = 4401 (0.581 sec)\n",
      "INFO:tensorflow:global_step/sec: 178.095\n",
      "INFO:tensorflow:loss = 0.007455939, step = 4501 (0.561 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.662\n",
      "INFO:tensorflow:loss = 0.0098478785, step = 4601 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.582\n",
      "INFO:tensorflow:loss = 0.04682129, step = 4701 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 177.462\n",
      "INFO:tensorflow:loss = 0.00805171, step = 4801 (0.563 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.562\n",
      "INFO:tensorflow:loss = 0.0097333845, step = 4901 (0.570 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.133\n",
      "INFO:tensorflow:loss = 0.0441731, step = 5001 (0.568 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.381\n",
      "INFO:tensorflow:loss = 0.04507794, step = 5101 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.098\n",
      "INFO:tensorflow:loss = 0.014277299, step = 5201 (0.571 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.558\n",
      "INFO:tensorflow:loss = 0.040422767, step = 5301 (0.566 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.494\n",
      "INFO:tensorflow:loss = 0.011191956, step = 5401 (0.570 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.462\n",
      "INFO:tensorflow:loss = 0.028478477, step = 5501 (0.554 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.023\n",
      "INFO:tensorflow:loss = 0.012146677, step = 5601 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 181.134\n",
      "INFO:tensorflow:loss = 0.0052959644, step = 5701 (0.552 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.384\n",
      "INFO:tensorflow:loss = 0.037021972, step = 5801 (0.570 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.633\n",
      "INFO:tensorflow:loss = 0.0060655433, step = 5901 (0.557 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.421\n",
      "INFO:tensorflow:loss = 0.036980387, step = 6001 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.783\n",
      "INFO:tensorflow:loss = 0.010244255, step = 6101 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.909\n",
      "INFO:tensorflow:loss = 0.03127115, step = 6201 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 181.444\n",
      "INFO:tensorflow:loss = 0.011337989, step = 6301 (0.551 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.939\n",
      "INFO:tensorflow:loss = 0.015598843, step = 6401 (0.553 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.292\n",
      "INFO:tensorflow:loss = 0.014756387, step = 6501 (0.570 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.902\n",
      "INFO:tensorflow:loss = 0.013507209, step = 6601 (0.568 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.311\n",
      "INFO:tensorflow:loss = 0.056300767, step = 6701 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.872\n",
      "INFO:tensorflow:loss = 0.012783189, step = 6801 (0.569 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.028\n",
      "INFO:tensorflow:loss = 0.01050051, step = 6901 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.908\n",
      "INFO:tensorflow:loss = 0.0403987, step = 7001 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.224\n",
      "INFO:tensorflow:loss = 0.06971167, step = 7101 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 177.173\n",
      "INFO:tensorflow:loss = 0.043725844, step = 7201 (0.564 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.939\n",
      "INFO:tensorflow:loss = 0.060182028, step = 7301 (0.553 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.647\n",
      "INFO:tensorflow:loss = 0.012706224, step = 7401 (0.554 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.841\n",
      "INFO:tensorflow:loss = 0.004273366, step = 7501 (0.556 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.916\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:loss = 0.0073607224, step = 7601 (0.553 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.236\n",
      "INFO:tensorflow:loss = 0.011813828, step = 7701 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.369\n",
      "INFO:tensorflow:loss = 0.03877332, step = 7801 (0.570 sec)\n",
      "INFO:tensorflow:global_step/sec: 174.692\n",
      "INFO:tensorflow:loss = 0.009772851, step = 7901 (0.572 sec)\n",
      "INFO:tensorflow:global_step/sec: 177.619\n",
      "INFO:tensorflow:loss = 0.051316462, step = 8001 (0.563 sec)\n",
      "INFO:tensorflow:global_step/sec: 179.268\n",
      "INFO:tensorflow:loss = 0.0200385, step = 8101 (0.558 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.512\n",
      "INFO:tensorflow:loss = 0.0062377797, step = 8201 (0.570 sec)\n",
      "INFO:tensorflow:global_step/sec: 174.719\n",
      "INFO:tensorflow:loss = 0.05605266, step = 8301 (0.572 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.483\n",
      "INFO:tensorflow:loss = 0.0067392513, step = 8401 (0.554 sec)\n",
      "INFO:tensorflow:global_step/sec: 177.931\n",
      "INFO:tensorflow:loss = 0.020801803, step = 8501 (0.562 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.169\n",
      "INFO:tensorflow:loss = 0.05096438, step = 8601 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.661\n",
      "INFO:tensorflow:loss = 0.031513862, step = 8701 (0.566 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.209\n",
      "INFO:tensorflow:loss = 0.005324862, step = 8801 (0.555 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.776\n",
      "INFO:tensorflow:loss = 0.0049007414, step = 8901 (0.553 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.498\n",
      "INFO:tensorflow:loss = 0.008497779, step = 9001 (0.566 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.443\n",
      "INFO:tensorflow:loss = 0.0064228713, step = 9101 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.525\n",
      "INFO:tensorflow:loss = 0.0034521369, step = 9201 (0.554 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.69\n",
      "INFO:tensorflow:loss = 0.005513321, step = 9301 (0.553 sec)\n",
      "INFO:tensorflow:global_step/sec: 180.877\n",
      "INFO:tensorflow:loss = 0.006491818, step = 9401 (0.553 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.501\n",
      "INFO:tensorflow:loss = 0.0049367417, step = 9501 (0.566 sec)\n",
      "INFO:tensorflow:global_step/sec: 176.532\n",
      "INFO:tensorflow:loss = 0.0068382933, step = 9601 (0.567 sec)\n",
      "INFO:tensorflow:global_step/sec: 175.34\n",
      "INFO:tensorflow:loss = 0.032444417, step = 9701 (0.571 sec)\n",
      "INFO:tensorflow:global_step/sec: 174.678\n",
      "INFO:tensorflow:loss = 0.04734677, step = 9801 (0.572 sec)\n",
      "INFO:tensorflow:global_step/sec: 178.933\n",
      "INFO:tensorflow:loss = 0.003876846, step = 9901 (0.559 sec)\n",
      "INFO:tensorflow:Saving checkpoints for 10000 into kernel_train/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.009976976.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "KernelLinearClassifier(params={'head': <tensorflow.contrib.learn.python.learn.estimators.head._BinaryLogisticHead object at 0x7f11f2f436a0>, 'feature_columns': {_RealValuedColumn(column_name='features_MAPPED', dimension=3000, default_value=None, dtype=tf.float32, normalizer=None)}, 'optimizer': <tensorflow.python.training.ftrl.FtrlOptimizer object at 0x7f11f2f43c18>, 'kernel_mappers': {_RealValuedColumn(column_name='features', dimension=286, default_value=None, dtype=tf.float32, normalizer=None): [<tensorflow.contrib.kernel_methods.python.mappers.random_fourier_features.RandomFourierFeatureMapper object at 0x7f11f2f43e10>]}})"
      ]
     },
     "execution_count": 251,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator_kernel.fit(input_fn=train_input_fn, steps=10000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Casting <dtype: 'int64'> labels to bool.\n",
      "WARNING:tensorflow:Casting <dtype: 'int64'> labels to bool.\n",
      "WARNING:tensorflow:Trapezoidal rule is known to produce incorrect PR-AUCs; please switch to \"careful_interpolation\" instead.\n",
      "WARNING:tensorflow:Trapezoidal rule is known to produce incorrect PR-AUCs; please switch to \"careful_interpolation\" instead.\n",
      "INFO:tensorflow:Starting evaluation at 2019-06-12-13:15:50\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from kernel_train/model.ckpt-10000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "INFO:tensorflow:Evaluation [1/1]\n",
      "INFO:tensorflow:Finished evaluation at 2019-06-12-13:15:52\n",
      "INFO:tensorflow:Saving dict for global step 10000: accuracy = 0.9449198, accuracy/baseline_label_mean = 0.9510752, accuracy/threshold_0.500000_mean = 0.9449198, auc = 0.86609143, auc_precision_recall = 0.9906777, global_step = 10000, labels/actual_label_mean = 0.9510752, labels/prediction_mean = 0.899043, loss = 0.17453441, precision/positive_threshold_0.500000_mean = 0.9713822, recall/positive_threshold_0.500000_mean = 0.97068363\n"
     ]
    }
   ],
   "source": [
    "eval_metrics = estimator_kernel.evaluate(input_fn=eval_input_fn, steps=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy: 0.9449198\n",
      "accuracy/baseline_label_mean: 0.9510752\n",
      "accuracy/threshold_0.500000_mean: 0.9449198\n",
      "auc: 0.86609143\n",
      "auc_precision_recall: 0.9906777\n",
      "global_step: 10000\n",
      "labels/actual_label_mean: 0.9510752\n",
      "labels/prediction_mean: 0.899043\n",
      "loss: 0.17453441\n",
      "precision/positive_threshold_0.500000_mean: 0.9713822\n",
      "recall/positive_threshold_0.500000_mean: 0.97068363\n"
     ]
    }
   ],
   "source": [
    "for key,value in sorted(eval_metrics.items()):\n",
    "  print('%s: %s' % (key, value))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 254,
   "metadata": {},
   "outputs": [],
   "source": [
    "correctness = []\n",
    "class_id = []\n",
    "pred_iterator = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from kernel_train/model.ckpt-10000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n"
     ]
    }
   ],
   "source": [
    "pred_iterator = estimator_kernel.predict(input_fn = test_input_fn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {},
   "outputs": [],
   "source": [
    "for pred_dict in pred_iterator:\n",
    "    class_id.append(pred_dict['classes'])\n",
    "    #print(pred_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: RFFM         Accuracy: 0.95974, F1-score: 0.97702, Precision: 0.97151, Recall: 0.98260, FPR: 0.19460, FNR: 0.01740\n",
      "TN: 6531; FP: 1578; FN:  953; TP: 53810\n",
      "\n"
     ]
    }
   ],
   "source": [
    "name = \"RFFM\"\n",
    "\n",
    "Acc[name] = metrics.accuracy_score(y_test, class_id)\n",
    "F1S[name] = metrics.f1_score(y_test, class_id)\n",
    "Prec[name] = metrics.precision_score(y_test, class_id)\n",
    "Rec[name] = metrics.recall_score(y_test, class_id)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(y_test, class_id).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Neural network part"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 394,
   "metadata": {},
   "outputs": [],
   "source": [
    "import collections\n",
    "import io\n",
    "import math\n",
    "import tensorflow as tf\n",
    "from IPython import display"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 395,
   "metadata": {},
   "outputs": [],
   "source": [
    "deep_numeric_feat = [\"Same_srv_rate\", \"Serror_rate\", \"Srv_serror_rate\", \"Dst_host_count\", \"Dst_host_srv_count\", \"Dst_host_same_src_port_rate\", \"Dst_host_serror_rate\", \"Dst_host_srv_serror_rate\", \"well_known_src_pool\", \"registered_src_pool\", \"cli_src_pool\"]\n",
    "wide_numeric_feat = [\"conn_len\",  \"src_bytes\", \"dst_bytes\", \"conn_numb\"]\n",
    "categorical_feat = [\"service\", \"Conn_state\", \"src_port\", \"dst_port\", \"proto\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 396,
   "metadata": {},
   "outputs": [],
   "source": [
    "deep_numeric_feat_tf = [tf.feature_column.numeric_column(k) for k in deep_numeric_feat]\n",
    "wide_numeric_feat_tf = [tf.feature_column.numeric_column(k) for k in wide_numeric_feat]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 397,
   "metadata": {},
   "outputs": [],
   "source": [
    "src_port_mod = [\"src_\" + s for s in data[\"src_port\"].unique().tolist()]\n",
    "src_port = tf.feature_column.categorical_column_with_vocabulary_list(\n",
    "    'src_port', src_port_mod)\n",
    "\n",
    "dst_port_mod = [\"dst_\" + s for s in data[\"dst_port\"].unique().tolist()]\n",
    "dst_port = tf.feature_column.categorical_column_with_vocabulary_list(\n",
    "    'dst_port', dst_port_mod)\n",
    "\n",
    "service = tf.feature_column.categorical_column_with_vocabulary_list(\n",
    "    'service', data[\"service\"].unique().tolist())\n",
    "\n",
    "Conn_state = tf.feature_column.categorical_column_with_vocabulary_list(\n",
    "    'Conn_state', data[\"Conn_state\"].unique().tolist())\n",
    "\n",
    "proto = tf.feature_column.categorical_column_with_vocabulary_list(\n",
    "    'proto', data[\"proto\"].unique().tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 398,
   "metadata": {},
   "outputs": [],
   "source": [
    "deep_columns = [\n",
    "    tf.feature_column.indicator_column(service),\n",
    "    tf.feature_column.indicator_column(Conn_state),\n",
    "    tf.feature_column.indicator_column(proto),\n",
    "    tf.feature_column.embedding_column(src_port, dimension=10),\n",
    "    tf.feature_column.embedding_column(dst_port, dimension=6),\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 399,
   "metadata": {},
   "outputs": [],
   "source": [
    "crossed_columns = [\n",
    "      tf.feature_column.crossed_column(\n",
    "          ['src_port', 'service'], hash_bucket_size=5000),\n",
    "      tf.feature_column.crossed_column(\n",
    "          ['dst_port', 'service'], hash_bucket_size=10000),\n",
    "      tf.feature_column.crossed_column(\n",
    "          ['conn_len', 'src_bytes'], hash_bucket_size=10000),\n",
    "      tf.feature_column.crossed_column(\n",
    "          ['conn_len', 'dst_bytes'], hash_bucket_size=10000),\n",
    "      tf.feature_column.crossed_column(\n",
    "          ['dst_port', 'conn_numb'], hash_bucket_size=10000),  \n",
    "      tf.feature_column.crossed_column(\n",
    "          ['service', 'proto'], hash_bucket_size=1000),\n",
    "    \n",
    "  ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 400,
   "metadata": {},
   "outputs": [],
   "source": [
    "! rm -r NN_model2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 401,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using default config.\n",
      "INFO:tensorflow:Using config: {'_model_dir': '/home/matz/Desktop/ml_ids/NN_model2', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true\n",
      "graph_options {\n",
      "  rewrite_options {\n",
      "    meta_optimizer_iterations: ONE\n",
      "  }\n",
      "}\n",
      ", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f1213e14550>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n"
     ]
    }
   ],
   "source": [
    "NN_model = tf.estimator.DNNLinearCombinedClassifier(\n",
    "    model_dir = \"/home/matz/Desktop/ml_ids/NN_model2\",\n",
    "    #linear_optimizer = tf.train.FtrlOptimizer(learning_rate=3, l2_regularization_strength=0.001),\n",
    "    linear_feature_columns = deep_columns + deep_numeric_feat_tf + wide_numeric_feat_tf + crossed_columns,\n",
    "    dnn_activation_fn=tf.nn.selu,\n",
    "    dnn_feature_columns = deep_columns + deep_numeric_feat_tf + wide_numeric_feat_tf,\n",
    "    dnn_hidden_units = [2048, 1024]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 402,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_set= deep_numeric_feat + wide_numeric_feat + categorical_feat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 403,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_input_fn(data_set, num_epochs=None, n_batch = 256, shuffle=True):\n",
    "    return tf.estimator.inputs.pandas_input_fn(\n",
    "       x=pd.DataFrame({k: data_set[k].values for k in feature_set}),\n",
    "       y = pd.Series(data_set[\"attack_flag\"].values),\n",
    "       batch_size=n_batch,   \n",
    "       num_epochs=num_epochs,\n",
    "       shuffle=shuffle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 404,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "INFO:tensorflow:Saving checkpoints for 0 into /home/matz/Desktop/ml_ids/NN_model2/model.ckpt.\n",
      "INFO:tensorflow:loss = 417.935, step = 1\n",
      "INFO:tensorflow:global_step/sec: 36.7497\n",
      "INFO:tensorflow:loss = 36.65182, step = 101 (2.722 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.4087\n",
      "INFO:tensorflow:loss = 13.233688, step = 201 (2.358 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0988\n",
      "INFO:tensorflow:loss = 8.13615, step = 301 (2.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.359\n",
      "INFO:tensorflow:loss = 9.335029, step = 401 (2.361 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.4342\n",
      "INFO:tensorflow:loss = 19.275803, step = 501 (2.357 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.11\n",
      "INFO:tensorflow:loss = 4.1052847, step = 601 (2.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3129\n",
      "INFO:tensorflow:loss = 36.31238, step = 701 (2.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2234\n",
      "INFO:tensorflow:loss = 1.034519, step = 801 (2.368 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2551\n",
      "INFO:tensorflow:loss = 24.100914, step = 901 (2.367 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9376\n",
      "INFO:tensorflow:loss = 2.4097943, step = 1001 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3496\n",
      "INFO:tensorflow:loss = 5.6520424, step = 1101 (2.361 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3306\n",
      "INFO:tensorflow:loss = 5.2566485, step = 1201 (2.362 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.5869\n",
      "INFO:tensorflow:loss = 1.107127, step = 1301 (2.348 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3853\n",
      "INFO:tensorflow:loss = 10.8564625, step = 1401 (2.360 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1251\n",
      "INFO:tensorflow:loss = 19.275928, step = 1501 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9159\n",
      "INFO:tensorflow:loss = 2.957772, step = 1601 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1733\n",
      "INFO:tensorflow:loss = 2.3794937, step = 1701 (2.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9226\n",
      "INFO:tensorflow:loss = 0.3541596, step = 1801 (2.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3257\n",
      "INFO:tensorflow:loss = 2.9389505, step = 1901 (2.363 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.302\n",
      "INFO:tensorflow:loss = 0.22970867, step = 2001 (2.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8535\n",
      "INFO:tensorflow:loss = 17.941664, step = 2101 (2.389 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.4498\n",
      "INFO:tensorflow:loss = 6.0638285, step = 2201 (2.356 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1395\n",
      "INFO:tensorflow:loss = 3.0732617, step = 2301 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.294\n",
      "INFO:tensorflow:loss = 11.459339, step = 2401 (2.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.23\n",
      "INFO:tensorflow:loss = 9.1954365, step = 2501 (2.368 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.4066\n",
      "INFO:tensorflow:loss = 0.40195197, step = 2601 (2.358 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2642\n",
      "INFO:tensorflow:loss = 136.18733, step = 2701 (2.366 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9612\n",
      "INFO:tensorflow:loss = 1.6101183, step = 2801 (2.383 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9384\n",
      "INFO:tensorflow:loss = 17.393057, step = 2901 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.143\n",
      "INFO:tensorflow:loss = 2.007148, step = 3001 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2964\n",
      "INFO:tensorflow:loss = 1.1991265, step = 3101 (2.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.5303\n",
      "INFO:tensorflow:loss = 100.2668, step = 3201 (2.408 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.2306\n",
      "INFO:tensorflow:loss = 0.2599319, step = 3301 (2.426 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8979\n",
      "INFO:tensorflow:loss = 2.9840772, step = 3401 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7402\n",
      "INFO:tensorflow:loss = 1.2520416, step = 3501 (2.396 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3695\n",
      "INFO:tensorflow:loss = 0.5391582, step = 3601 (2.360 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.03\n",
      "INFO:tensorflow:loss = 8.255706, step = 3701 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.812\n",
      "INFO:tensorflow:loss = 2.2155426, step = 3801 (2.392 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1166\n",
      "INFO:tensorflow:loss = 12.680943, step = 3901 (2.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.913\n",
      "INFO:tensorflow:loss = 6.0180497, step = 4001 (2.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3249\n",
      "INFO:tensorflow:loss = 19.76437, step = 4101 (2.363 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1734\n",
      "INFO:tensorflow:loss = 23.831236, step = 4201 (2.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0441\n",
      "INFO:tensorflow:loss = 8.113055, step = 4301 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0936\n",
      "INFO:tensorflow:loss = 116.63454, step = 4401 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9393\n",
      "INFO:tensorflow:loss = 5.9367766, step = 4501 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7686\n",
      "INFO:tensorflow:loss = 15.758875, step = 4601 (2.394 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0944\n",
      "INFO:tensorflow:loss = 1.4958669, step = 4701 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2107\n",
      "INFO:tensorflow:loss = 2.085995, step = 4801 (2.369 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7929\n",
      "INFO:tensorflow:loss = 0.42308205, step = 4901 (2.393 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9789\n",
      "INFO:tensorflow:loss = 17.21388, step = 5001 (2.382 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7959\n",
      "INFO:tensorflow:loss = 2.4415348, step = 5101 (2.393 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.146\n",
      "INFO:tensorflow:loss = 0.77566844, step = 5201 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.102\n",
      "INFO:tensorflow:loss = 8.183012, step = 5301 (2.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0355\n",
      "INFO:tensorflow:loss = 3.3914747, step = 5401 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1298\n",
      "INFO:tensorflow:loss = 1.7627302, step = 5501 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3089\n",
      "INFO:tensorflow:loss = 2.4296463, step = 5601 (2.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9411\n",
      "INFO:tensorflow:loss = 28.647947, step = 5701 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1604\n",
      "INFO:tensorflow:loss = 8.35509, step = 5801 (2.372 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2644\n",
      "INFO:tensorflow:loss = 1.2587596, step = 5901 (2.366 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0996\n",
      "INFO:tensorflow:loss = 6.596719, step = 6001 (2.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9771\n",
      "INFO:tensorflow:loss = 2.1338527, step = 6101 (2.383 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1291\n",
      "INFO:tensorflow:loss = 6.7376857, step = 6201 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2147\n",
      "INFO:tensorflow:loss = 224.78186, step = 6301 (2.369 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2185\n",
      "INFO:tensorflow:loss = 7.303179, step = 6401 (2.369 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0122\n",
      "INFO:tensorflow:loss = 0.70205486, step = 6501 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1219\n",
      "INFO:tensorflow:loss = 0.14763011, step = 6601 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0873\n",
      "INFO:tensorflow:loss = 0.24737513, step = 6701 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0078\n",
      "INFO:tensorflow:loss = 9.049537, step = 6801 (2.381 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1844\n",
      "INFO:tensorflow:loss = 7.8229938, step = 6901 (2.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9013\n",
      "INFO:tensorflow:loss = 2.5547814, step = 7001 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9023\n",
      "INFO:tensorflow:loss = 4.3079376, step = 7101 (2.387 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3432\n",
      "INFO:tensorflow:loss = 1.2732174, step = 7201 (2.361 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8799\n",
      "INFO:tensorflow:loss = 8.232129, step = 7301 (2.388 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2426\n",
      "INFO:tensorflow:loss = 9.309405, step = 7401 (2.367 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0836\n",
      "INFO:tensorflow:loss = 2.082933, step = 7501 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9157\n",
      "INFO:tensorflow:loss = 2.9346066, step = 7601 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0032\n",
      "INFO:tensorflow:loss = 6.496351, step = 7701 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2646\n",
      "INFO:tensorflow:loss = 2.0246139, step = 7801 (2.366 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.981\n",
      "INFO:tensorflow:loss = 19.42863, step = 7901 (2.382 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0027\n",
      "INFO:tensorflow:loss = 0.49340218, step = 8001 (2.381 sec)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:global_step/sec: 42.0225\n",
      "INFO:tensorflow:loss = 0.31525812, step = 8101 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2274\n",
      "INFO:tensorflow:loss = 7.9066157, step = 8201 (2.368 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.2271\n",
      "INFO:tensorflow:loss = 0.23836967, step = 8301 (2.426 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.5897\n",
      "INFO:tensorflow:loss = 0.49869123, step = 8401 (2.404 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8827\n",
      "INFO:tensorflow:loss = 0.1534415, step = 8501 (2.388 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3657\n",
      "INFO:tensorflow:loss = 17.510006, step = 8601 (2.360 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3699\n",
      "INFO:tensorflow:loss = 11.830587, step = 8701 (2.360 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9626\n",
      "INFO:tensorflow:loss = 3.5563664, step = 8801 (2.383 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0497\n",
      "INFO:tensorflow:loss = 25.99097, step = 8901 (2.378 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0447\n",
      "INFO:tensorflow:loss = 8.719852, step = 9001 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3332\n",
      "INFO:tensorflow:loss = 10.541747, step = 9101 (2.362 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0219\n",
      "INFO:tensorflow:loss = 5.3030972, step = 9201 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1334\n",
      "INFO:tensorflow:loss = 24.916344, step = 9301 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1254\n",
      "INFO:tensorflow:loss = 8.032485, step = 9401 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.117\n",
      "INFO:tensorflow:loss = 501.30066, step = 9501 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0108\n",
      "INFO:tensorflow:loss = 0.75639194, step = 9601 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9316\n",
      "INFO:tensorflow:loss = 2.2181804, step = 9701 (2.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9978\n",
      "INFO:tensorflow:loss = 0.32343695, step = 9801 (2.381 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9769\n",
      "INFO:tensorflow:loss = 0.2806941, step = 9901 (2.382 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3683\n",
      "INFO:tensorflow:loss = 0.6308079, step = 10001 (2.360 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9085\n",
      "INFO:tensorflow:loss = 1.0723577, step = 10101 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2015\n",
      "INFO:tensorflow:loss = 20.83251, step = 10201 (2.370 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7185\n",
      "INFO:tensorflow:loss = 2.0731223, step = 10301 (2.397 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2462\n",
      "INFO:tensorflow:loss = 15.7834015, step = 10401 (2.367 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0793\n",
      "INFO:tensorflow:loss = 1.3248708, step = 10501 (2.377 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1618\n",
      "INFO:tensorflow:loss = 0.98743385, step = 10601 (2.372 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.144\n",
      "INFO:tensorflow:loss = 3.4507816, step = 10701 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0981\n",
      "INFO:tensorflow:loss = 1.3276042, step = 10801 (2.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9247\n",
      "INFO:tensorflow:loss = 144.02792, step = 10901 (2.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0698\n",
      "INFO:tensorflow:loss = 2.0469053, step = 11001 (2.377 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0174\n",
      "INFO:tensorflow:loss = 5.5172334, step = 11101 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3304\n",
      "INFO:tensorflow:loss = 55.368427, step = 11201 (2.362 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0319\n",
      "INFO:tensorflow:loss = 2.2389326, step = 11301 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9409\n",
      "INFO:tensorflow:loss = 2.2595475, step = 11401 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.128\n",
      "INFO:tensorflow:loss = 1.080132, step = 11501 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0212\n",
      "INFO:tensorflow:loss = 1.2426063, step = 11601 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0623\n",
      "INFO:tensorflow:loss = 2.729012, step = 11701 (2.377 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8612\n",
      "INFO:tensorflow:loss = 14.466118, step = 11801 (2.389 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9277\n",
      "INFO:tensorflow:loss = 2.3381016, step = 11901 (2.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1659\n",
      "INFO:tensorflow:loss = 13.652709, step = 12001 (2.372 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1862\n",
      "INFO:tensorflow:loss = 1.0646812, step = 12101 (2.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.841\n",
      "INFO:tensorflow:loss = 1.817629, step = 12201 (2.390 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2302\n",
      "INFO:tensorflow:loss = 7.0296774, step = 12301 (2.368 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9075\n",
      "INFO:tensorflow:loss = 8.412181, step = 12401 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2265\n",
      "INFO:tensorflow:loss = 0.9583534, step = 12501 (2.369 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7123\n",
      "INFO:tensorflow:loss = 2.7632327, step = 12601 (2.397 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.088\n",
      "INFO:tensorflow:loss = 7.431494, step = 12701 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0141\n",
      "INFO:tensorflow:loss = 16.593155, step = 12801 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9549\n",
      "INFO:tensorflow:loss = 6.935566, step = 12901 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0593\n",
      "INFO:tensorflow:loss = 0.22779465, step = 13001 (2.378 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1807\n",
      "INFO:tensorflow:loss = 6.7988415, step = 13101 (2.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2456\n",
      "INFO:tensorflow:loss = 7.3724174, step = 13201 (2.367 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.5162\n",
      "INFO:tensorflow:loss = 1.7276429, step = 13301 (2.409 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.3506\n",
      "INFO:tensorflow:loss = 1.27988, step = 13401 (2.418 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8568\n",
      "INFO:tensorflow:loss = 7.1525555, step = 13501 (2.389 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0797\n",
      "INFO:tensorflow:loss = 3.1055584, step = 13601 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1009\n",
      "INFO:tensorflow:loss = 1.9483917, step = 13701 (2.375 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8601\n",
      "INFO:tensorflow:loss = 11.891381, step = 13801 (2.389 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0841\n",
      "INFO:tensorflow:loss = 2.390061, step = 13901 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3631\n",
      "INFO:tensorflow:loss = 2.5774837, step = 14001 (2.361 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1331\n",
      "INFO:tensorflow:loss = 2.0336661, step = 14101 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.004\n",
      "INFO:tensorflow:loss = 2.653989, step = 14201 (2.381 sec)\n",
      "INFO:tensorflow:global_step/sec: 40.9881\n",
      "INFO:tensorflow:loss = 5.9537134, step = 14301 (2.440 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1897\n",
      "INFO:tensorflow:loss = 5.952035, step = 14401 (2.370 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.914\n",
      "INFO:tensorflow:loss = 11.291847, step = 14501 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1483\n",
      "INFO:tensorflow:loss = 0.32101107, step = 14601 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0163\n",
      "INFO:tensorflow:loss = 0.68732226, step = 14701 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7684\n",
      "INFO:tensorflow:loss = 0.10343123, step = 14801 (2.394 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9739\n",
      "INFO:tensorflow:loss = 0.5694943, step = 14901 (2.383 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.272\n",
      "INFO:tensorflow:loss = 1.2893771, step = 15001 (2.366 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8464\n",
      "INFO:tensorflow:loss = 1.3207314, step = 15101 (2.390 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2008\n",
      "INFO:tensorflow:loss = 1.4529448, step = 15201 (2.370 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0343\n",
      "INFO:tensorflow:loss = 5.8342075, step = 15301 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8934\n",
      "INFO:tensorflow:loss = 3.5330145, step = 15401 (2.387 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.008\n",
      "INFO:tensorflow:loss = 14.094214, step = 15501 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1768\n",
      "INFO:tensorflow:loss = 18.210781, step = 15601 (2.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9588\n",
      "INFO:tensorflow:loss = 5.262491, step = 15701 (2.383 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0881\n",
      "INFO:tensorflow:loss = 9.563666, step = 15801 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.003\n",
      "INFO:tensorflow:loss = 2.917776, step = 15901 (2.381 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0808\n",
      "INFO:tensorflow:loss = 114.67098, step = 16001 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0715\n",
      "INFO:tensorflow:loss = 0.60599905, step = 16101 (2.377 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.94\n",
      "INFO:tensorflow:loss = 0.14974444, step = 16201 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8714\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:loss = 0.8221551, step = 16301 (2.388 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0093\n",
      "INFO:tensorflow:loss = 10.563806, step = 16401 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9606\n",
      "INFO:tensorflow:loss = 6.4465065, step = 16501 (2.383 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0792\n",
      "INFO:tensorflow:loss = 9.259839, step = 16601 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2279\n",
      "INFO:tensorflow:loss = 8.061024, step = 16701 (2.368 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9195\n",
      "INFO:tensorflow:loss = 6.570778, step = 16801 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9746\n",
      "INFO:tensorflow:loss = 2.1221085, step = 16901 (2.382 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1183\n",
      "INFO:tensorflow:loss = 4.7548738, step = 17001 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0168\n",
      "INFO:tensorflow:loss = 1.3440753, step = 17101 (2.380 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7801\n",
      "INFO:tensorflow:loss = 28.736416, step = 17201 (2.394 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.3509\n",
      "INFO:tensorflow:loss = 1.761888, step = 17301 (2.361 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0285\n",
      "INFO:tensorflow:loss = 5.7579727, step = 17401 (2.379 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9162\n",
      "INFO:tensorflow:loss = 11.8863735, step = 17501 (2.386 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.227\n",
      "INFO:tensorflow:loss = 7.9489985, step = 17601 (2.368 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.901\n",
      "INFO:tensorflow:loss = 778.6219, step = 17701 (2.387 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.2974\n",
      "INFO:tensorflow:loss = 2.026344, step = 17801 (2.364 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8355\n",
      "INFO:tensorflow:loss = 0.57399446, step = 17901 (2.390 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9606\n",
      "INFO:tensorflow:loss = 1.9613636, step = 18001 (2.383 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.6987\n",
      "INFO:tensorflow:loss = 0.16039217, step = 18101 (2.398 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1427\n",
      "INFO:tensorflow:loss = 1.8692287, step = 18201 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7274\n",
      "INFO:tensorflow:loss = 0.8320848, step = 18301 (2.397 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.0089\n",
      "INFO:tensorflow:loss = 8.425418, step = 18401 (2.438 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8221\n",
      "INFO:tensorflow:loss = 10.427245, step = 18501 (2.391 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7389\n",
      "INFO:tensorflow:loss = 1.0156484, step = 18601 (2.396 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1459\n",
      "INFO:tensorflow:loss = 7.880296, step = 18701 (2.373 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0856\n",
      "INFO:tensorflow:loss = 3.2274604, step = 18801 (2.376 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1739\n",
      "INFO:tensorflow:loss = 0.72424376, step = 18901 (2.371 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0631\n",
      "INFO:tensorflow:loss = 0.599891, step = 19001 (2.377 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0102\n",
      "INFO:tensorflow:loss = 1.9402868, step = 19101 (2.381 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.7263\n",
      "INFO:tensorflow:loss = 3.282601, step = 19201 (2.396 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9861\n",
      "INFO:tensorflow:loss = 22.853209, step = 19301 (2.382 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.1345\n",
      "INFO:tensorflow:loss = 1.7597648, step = 19401 (2.374 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8119\n",
      "INFO:tensorflow:loss = 6.1305556, step = 19501 (2.392 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.942\n",
      "INFO:tensorflow:loss = 0.7697608, step = 19601 (2.384 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.9267\n",
      "INFO:tensorflow:loss = 0.64054537, step = 19701 (2.385 sec)\n",
      "INFO:tensorflow:global_step/sec: 41.8476\n",
      "INFO:tensorflow:loss = 0.1707711, step = 19801 (2.389 sec)\n",
      "INFO:tensorflow:global_step/sec: 42.0215\n",
      "INFO:tensorflow:loss = 0.34628212, step = 19901 (2.380 sec)\n",
      "INFO:tensorflow:Saving checkpoints for 20000 into /home/matz/Desktop/ml_ids/NN_model2/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 1.6636817.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.estimator.canned.dnn_linear_combined.DNNLinearCombinedClassifier at 0x7f1213e14080>"
      ]
     },
     "execution_count": 404,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "NN_model.train(input_fn=get_input_fn(train_ex, n_batch=180), steps=20000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 405,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n",
      "WARNING:tensorflow:Trapezoidal rule is known to produce incorrect PR-AUCs; please switch to \"careful_interpolation\" instead.\n",
      "WARNING:tensorflow:Trapezoidal rule is known to produce incorrect PR-AUCs; please switch to \"careful_interpolation\" instead.\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Starting evaluation at 2019-06-12-16:39:54\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from /home/matz/Desktop/ml_ids/NN_model2/model.ckpt-20000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n",
      "INFO:tensorflow:Evaluation [1/1]\n",
      "INFO:tensorflow:Finished evaluation at 2019-06-12-16:39:59\n",
      "INFO:tensorflow:Saving dict for global step 20000: accuracy = 0.6855198, accuracy_baseline = 0.9510752, auc = 0.70547944, auc_precision_recall = 0.9801119, average_loss = 0.65614486, global_step = 20000, label/mean = 0.9510752, loss = 41253.14, precision = 0.96783245, prediction/mean = 0.7577215, recall = 0.692354\n",
      "INFO:tensorflow:Saving 'checkpoint_path' summary for global step 20000: /home/matz/Desktop/ml_ids/NN_model2/model.ckpt-20000\n"
     ]
    }
   ],
   "source": [
    "eval_metrics = NN_model.evaluate(input_fn=get_input_fn(validate_ex, n_batch=validate_ex.shape[0], shuffle=False), steps=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 406,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy: 0.6855198\n",
      "accuracy_baseline: 0.9510752\n",
      "auc: 0.70547944\n",
      "auc_precision_recall: 0.9801119\n",
      "average_loss: 0.65614486\n",
      "global_step: 20000\n",
      "label/mean: 0.9510752\n",
      "loss: 41253.14\n",
      "precision: 0.96783245\n",
      "prediction/mean: 0.7577215\n",
      "recall: 0.692354\n"
     ]
    }
   ],
   "source": [
    "for key,value in sorted(eval_metrics.items()):\n",
    "  print('%s: %s' % (key, value))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 407,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from /home/matz/Desktop/ml_ids/NN_model2/model.ckpt-20000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n"
     ]
    }
   ],
   "source": [
    "results = []\n",
    "results = list(NN_model.predict(input_fn=get_input_fn(test_ex, num_epochs=1, n_batch = test_ex.shape[0], shuffle=False)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 408,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = []\n",
    "for el in results:\n",
    "    y_pred.append(el[\"class_ids\"])\n",
    "\n",
    "y_pred_NN = [int(a) for a in y_pred]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 409,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: Wide&Deep    Accuracy: 0.95222, F1-score: 0.97225, Precision: 0.98379, Recall: 0.96098, FPR: 0.10692, FNR: 0.03902\n",
      "TN: 7242; FP: 867; FN: 2137; TP: 52626\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn import metrics\n",
    "name = \"Wide&Deep\"\n",
    "\n",
    "Acc[name] = metrics.accuracy_score(test_ex[\"attack_flag\"], y_pred_NN)\n",
    "F1S[name] = metrics.f1_score(test_ex[\"attack_flag\"], y_pred_NN)\n",
    "Prec[name] = metrics.precision_score(test_ex[\"attack_flag\"], y_pred_NN)\n",
    "Rec[name] = metrics.recall_score(test_ex[\"attack_flag\"], y_pred_NN)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(test_ex[\"attack_flag\"], y_pred_NN).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 410,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f126e40c438>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11f8097400>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f126e40cd30>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAE/CAYAAAB8YAsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzs3XecVOX1x/HPkaIhKDZUBBURRIpgQUSNoKKIRLFgFNRYE00iGntvGBVQo8afmFgSu2LvglGj2GKLIlEsAQTBCiqIoNLO74/z7DIsW2Zxd+9d+L5fr32xc+fuzJnhzpx7n+c8z2PujoiIiOTHSlkHICIiIktSchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZ5EimNlOZjat4PZkM9t1GR/rIjObYWaf11yENc/MDjezF5fxb282s4tqOqb02Aeb2T8ruX+J/yuR+kjJWeqdlBi/N7PvzOzzlAiaZh1XMcxsA+BkoKO7r1fO/TWWWMzsOTP7TU08Vp64+x3u3qfktpm5mbXNMiaRmqbkLPXVXu7eFNgC2BI4M+N4irUR8JW7f5l1IPWRmTXMOgaRuqDkLPWau38OPEkkaQDMbGUzu9zMPjazL8zsb2b2s4L79zazsWb2rZlNNLO+afsRZvaemc02s0lmdsyyxGRmzczsVjObbmZTzOwcM1spNYM/BayfrvpvLvN3PwdGFdz/nZmtn/72jBTrV2Z2j5mtmf5mFTO7PW2faWavm9m6ZnYxsCNwTXqca9L+m5nZU2b2tZl9YGYHFDz/Wmb2SHpfXgM2qeJ13ptaLmaZ2fNm1qmSfU8zs8/M7FMz+03h1W5F71e673Aze8nMrjSzr4ELCpvbzez59BRvp9d5YMFznmxmX6bnPaJg+81mdq2ZjUp/85KZrWdmV5nZN2b2vpltWbD/6Wb2STouPjCz3pW9LyI1QclZ6jUzawXsAUwo2Dwc2JRI2G2BlsB5af/uwK3AqcDqQE9gcvq7L4E9gdWAI4ArzWyrZQjr/4BmQBugF3AocIS7P51i/dTdm7r74YV/5O5zytzf1N0/BY4H9kmPtT7wDTAi/dlh6bk2ANYCfgd87+5nAy8Ag9PjDE7J/yngTmAdYBBwbUFSHQH8ALQAjkw/lRkFtEuP9SZwR3k7pZOfk4Bdif+PXsW8XwX3bwtMSs9zceEfunvP9GvX9DrvTrfXS4/ZEjgKGGFmaxT86QHAOcDawI/Av9NrWBu4D7gixd4eGAxs4+6rAruz+HgRqT3urh/91Ksf4svxO2A24MAzwOrpPgPmAJsU7L8d8FH6/TrgyiKf5yHgj+n3nYBpZWLYtZy/aUB82Xcs2HYM8Fx5j1PO3y91P/Ae0LvgdgtgPtCQSKAvA13KeazngN8U3D4QeKHMPtcB56e45wObFdx3CfBike/V6un/olm6fTNwUfr9H8DQgn3bpn3bFvF+HQ58XOa5Di+Mq+SxyryH3wMNC7Z9CfQoiO2GgvuOA94ruL05MLMg1i+JE4tGWR/7+llxfnTlLPXVPh5XMjsBmxFXPADNgSbAf1Iz70xgdNoOcYU5sbwHNLM9zOyV1OQ7E+hX8LjFWhtoDEwp2DaFuIJbVhsBDxa8nveAhcC6wG1Es/7I1GR8qZk1quRxti15nPRYBxNXmc2JZD+1TNzlMrMGZjYsNbV/y+KryfLer/XLPG7h78W8X4X7F+srd19QcHsuUFg0+EXB79+Xc7spgLtPAE4ALgC+NLORZrb+MsQjUi1KzlKvufsY4kro8rRpBvHl2sndV08/zTyKxyC+6JfqSzWzlYH70+Os6+6rA08QV+LVMYO4At2oYNuGwCdF/n15y8RNBfYoeD2ru/sq7v6Ju8939yHu3hHYnmiWP7SCx5oKjCnzOE3d/ffAdGABcfJSGHdFDgL2Jq4omwGt0/by3q/PgFYFtwufo5j3K9Ol89z9Tnf/BRGjE90mIrVKyVmWB1cBu5nZFu6+CLiB6C9eB8DMWprZ7mnfvwNHmFnvVGjV0sw2I67eViYlKTPbA+iz9FNVzt0XAvcAF5vZqma2EdHfenuRD/EFsJaZNSvY9rf0eBul19PczPZOv+9sZpubWQPgWyLRLSx4rDYFj/MYsKmZ/drMGqWfbcysQ4r7AaLgqomZdST6syuyKtEc/RXRUnFJJfveQ7znHcysCan/H2rk/SrvddYYM2tvZrukk7cfiBO/hVX8mchPpuQs9Z67TyeKvM5Nm04nCsReSU2uTwPt076vkYq9gFnAGGAjd59NFF7dQxRcHQQ8sowhHUf0e08CXiQKsP5R5Gt5H7gLmJSantcH/pJi+aeZzQZeIYqkIJqk7yMS83vp9ZQktr8A+6cK5KvTa+wDDAQ+BT4nrgJXTvsPJppzPydaI26qJNRbiebnT4DxKaaKXtMo4GrgWeL/5d/prh/Tv8v8fiUXALek9+uAqnauppWBYcQV/udEUdpZNfwcIksx90xbjERkBWNmHYB3gJXL9AuLSKIrZxGpdWa2r5k1TsOZhgOPKjGLVEzJWUTqwjFEf/5Eos/299mGI5JvatYWERHJGV05i4iI5Exmyblv375OjBnM5OfII4/0ddZZxzt37lzu/e7uxx9/vLdt29a7dOnib775Zqbx5u1n9OjR3r59e2/btq0PGzZsqfunTJnivXv39i5duvhOO+3k06ZNK73v9NNP986dO3vnzp397rvvLt3+zDPP+FZbbeVbbLGF/+IXv/AJEyY44H/729988803L90+fvz4JZ7r448/9qZNm/rll1+e+fuSlx8d33rP9ZPLn+JlNTXZ1ltv7VkaM2aM/+c///FOnTqVe//jjz/uffv29UWLFvm///1v7969ex1HmF8LFizwNm3a+MSJE/3HH3/0Ll26+LvvvrvEPvvvv7/ffPPN7u7+zDPP+CGHHOLu7o899pjvuuuuPn/+fP/uu+9866239lmzZrm7e7t27Xz8+PHu7j5ixAg/7LDD3N1L73d3f/jhh3333Xdf4rn2228/33///f2yyy6rMvZRo0b5pptu6ptssokPHTp0qfsnT57su+yyi2+++ebeq1cvnzp1aul9p512mnfq1Mk7derkI0eOLN1+2GGHeevWrb1r167etWtXf+utt9zd/b333vMePXp448aNl4pto4028s6dO3vXrl29Nj4LOr7rnt5zKYKm76xKz549WXPNNSu8/+GHH+bQQw/FzOjRowczZ87ks88+q8MI8+u1116jbdu2tGnThsaNGzNw4EAefvjhJfYZP348vXvH4j0777xz6f3jx4+nV69eNGzYkJ///Od07dqV0aNHA2BmfPvttwDMmjWL9dePWRJXW2210sedM2cOZosnoXrooYdo06YNnTpVuCBSqYULF3LssccyatQoxo8fz1133cX48eOX2OeUU07h0EMPZdy4cZx33nmceWasRPn444/z5ptvMnbsWF599VUuu+yy0lgBLrvsMsaOHcvYsWPZYotYIGvNNdfk6quv5pRTTik3nmeffZaxY8fyxhtvVBl7den4rnt5fM9Hjx5N+/btadu2LcOGDVvq/ilTptC7d2+6dOnCTjvtxLRpi5cSP/300+ncuTOdO3fm7rvvLt1+8MEH0759ezp37syRRx7J/PnzAXj//ffZbrvtWHnllbn88suXei6pnhU2OVflk08+YYMNFs8y2KpVKz75pNgZGIv3Uz48p512Gp06daJDhw4cf/zxeJnivv79+9O5c+fS2/feey+dOnVipZVWWiIh3HHHHWyxxRalPyuttBJjx46tMOZi3puuXbty//33A/Dggw8ye/ZsvvrqK7p27cqoUaOYO3cuM2bM4Nlnn2Xq1Jg6+cYbb6Rfv360atWK2267jTPOOKP08UaMGMEmm2zCaaedxtVXXw1Eoh4+fDjnn39+xW9wgdo6qajIOuuswzbbbEOjRhVNdZ2dujq+oW6P8XPPPZcuXbqwxRZb0KdPHz799FMgTp5Kju/OnTvToEEDvv7661p5vRWpy/ccau9k9OCDD+b999/nv//9L99//z033ngjUPXJqFSPknMFyn4JAEtcsdWEn/Lhefnll3nppZcYN24c77zzDq+//jpjxowp/bsHHniApk2bLvFYnTt35oEHHqBnz55LbD/44INLr/puu+02WrduXXr1V55i3pvLL7+cMWPGsOWWWzJmzBhatmxJw4YN6dOnD/369WP77bdn0KBBbLfddjRs2BCAK6+8kieeeIJp06ZxxBFHcNJJJ5U+3rHHHsvEiRMZPnw4F110EQDnn38+J5544lKvsyK1dVIBcPbZZ9OlSxdOPPFEfvzxR6piZvTp04ett96a66+/vqj4a1JdHN9Q98f4qaeeyrhx4xg7dix77rknF154Yen2kmN86NCh9OrVq9Kr3NpQV+95ido6Ge3Xrx9mhpnRvXv30pOpPJ+M1kdKzhVo1arVEl++06ZNK21mrSk/5cNjZvzwww/MmzePH3/8kfnz57PuuusC8N1333HFFVdwzjnnLPFYHTp0oH379pXGdNdddzFo0KBK9ynmvVl//fV54IEHeOutt7j44liCt1mzmC767LPPZuzYsTz11FO4O+3atWP69Om8/fbbbLttzEp54IEH8vLLLy/13AMHDuShhx4C4NVXX+W0006jdevWXHXVVVxyySVcc801FcZdWycVQ4cO5f333+f111/n66+/ZvjwqtdFeOmll3jzzTcZNWoUI0aM4Pnnn6/yb2pSXRzfUPfHeGVdICWKOcZrQ1295yVq82QUYP78+dx222307du31l7DikzJuQL9+/fn1ltvxd155ZVXaNasGS1atKjR5/gpH57tttuOnXfemRYtWtCiRQt23313OnToAETT3sknn0yTJk2qHdPdd99d5RfXNttsw//+9z8++ugj5s2bx8iRI+nfv/8S+8yYMYNFixYBkbyOPPJIIK6kvvrqKwDGjRvHuHHj6NOnD2ussQazZs3iww8/BOCpp54qfT3/+9//Sh/38ccfp127dgC88MILTJ48mcmTJ3PCCSdw1llnMXjw4Arjro2TCoAWLVpgZqy88socccQRvPbaa5W+fyXPA3G1se+++xb1NzWpLo5vyOYYP/vss9lggw244447Sq+cS8ydO5fRo0czYMCAmn6pVaqr97xEbZ2MlvjDH/5Az5492XHHHWvtNazIGla9y/Jp0KBBPPfcc8yYMYNWrVoxZMiQ0sKG3/3ud/Tr148nnniCtm3b0qRJE2666aYaj6HYD8/gwYO5+eab6dmzZ+mHZ8KECbz33nulTUq77bYbzz//PKutthoTJkzgyiuvZPLkydWK59VXX6VJkyZL9OGVp2HDhlxzzTXsvvvuLFy4kCOPPJJOnTpx3nnn0a1bN/r3789zzz3HmWeeiZnRs2dPRowYAcTZdsmHebXVVuP2228v/dDfcMMNDBgwgJVWWok11liDf/wj1j645pprePrpp2nUqBFrrLEGt9xyS7VeV4nCk4qWLVsycuRI7rzzziX2mTFjBmuuuSYrrbTSUicVM2fOZK211lripALgs88+o0WLFrg7Dz30UJXv35w5c1i0aBGrrroqc+bM4Z///CfnnXdepX9TXXk4viGbY/ziiy/m4osvZujQoVxzzTUMGTKk9L5HH32UHXbYoVaatPPynpeozskoRGvE/fffv8TJ6Nlnnw3AQQcdVHoyCjBkyBCmT5/OddddV6uvYYVWVTk3sTrMl8A7FdxvxIozE4BxwFbFlIlnPZQqD15++WXv06dP6e1LLrnEL7nkkgr3nz17trds2dLd3S+99FK/8MILS+8bMmSIDx8+3K+99lpv0aKFb7TRRt6yZUtv1KiR9+rVa4nH6dWrl7/++utLPf4JJ5zgF1988U98Vfn2+OOPe7t27bxNmzZ+0UUXubv7ueee6w8//LC7u997773etm1bb9eunR911FH+ww8/uLv7999/7x06dPAOHTr4tttuWzpcyt1955139s6dO3unTp384IMP9tmzZ7u7+2effeYtW7b0VVdd1Zs1a+YtW7b0WbNm+cSJE71Lly7epUsX79ixY2kcy6OsjnH3GBZXdljTPvvs43fcccdPfFX1w/z5833jjTf2SZMmlQ55fOedd5bYZ/r06b5w4UJ3dz/rrLP83HPPdfcYLjljxgx3d3/77be9U6dOPn/+fHd3v+GGG3y77bbzuXPnlvu8559/flHDGsv6KcMcp0yZ4rvttptvttlm3qFDB//oo4/c3X3SpEnevXt3b9u2rR9wwAH+448/VvlYp556qnfs2NE322wzP+6443zRokXVfi2VKHooVTHJuSewVSXJuR8wKiXpHsCrxTyxkvNP+/CMHDnSe/fu7fPnz/d58+b5Lrvs4o888sgSf/vRRx+VO+ayvOS8cOFCb9mypU+cOLEmX6Ks4Or6GP/www9Lf7/66qt9wIABpbdnzpzpa6yxhn/33Xc1/jrzqjZORhs0aOBt2rQpHdc/ZMgQd6/4ZLQYP2XuBPf4TvvnP//p7nGCN2fOHHd3/9WvfuV33XWXu7sfc8wxfu2111b6WC+99JJvv/32vmDBAl+wYIH36NHDn3322aJeQ5FqLjl7JODWlSTn64BBBbc/AFpU9ZhKzmFZPzwLFizwo48+uvRM8cQTT1zqsct+cT3wwAPesmVLb9y4sa+zzjpLXNE8++yzvu2229bmS5UVVF0e4/vtt5936tTJN998c99zzz192rRppffddNNNfuCBB9bmS5VlVEwLS8eOHUuvcBctWuSrrrqqu7u/++67vsMOOyz1mIsWLfK11lqr9Iq/8DkqeqyXX37Zt9pqK587d67PmTPHt95669KJkWpI0cm5qIUvzKw18Ji7L9WZZmaPAcPc/cV0+xngdHdfamYFMzsaOBpgww033HrKlClVPnexWp/xeI09Vk2ZPOyXWYdQq/L2nuv9rnvL83uu97vu3HfffYwePbp0zPRtt93Gq6++usToi4MOOohtt92WP/7xjzzwwAMMGDCAGTNm8MILL3DjjTfSuHFjPvroI3bddVeGDRvGN998Q48ePZgwYQIAU6dOZY899uCdd96p8LHWWmstTjnlFG688UbcncGDB5cWhtaQosfO1URBWHlPVm7Gd/frgesBunXrVr15RkVqiL50ZXlWH4/v8i4Siy0cXLBgAS+88AJvvfUWG264IQceeCA333zzUiNICh+zukWIZeeGqAs1MZRqGrBBwe1WwKc18LgiIrIC+CnDHFu1asWWW25JmzZtaNiwIfvssw9vvvkma6+9NjNnzmTBggVLPWZFj/Xggw/So0cPmjZtStOmTdljjz145ZVX6uItWEpNXDk/Agw2s5HAtsAsd9ckvUWqj2e5ItWhY1yq8lOGOW6zzTZ88803TJ8+nebNm/Ovf/2Lbt26YWbsvPPO3HfffQwcOJBbbrmFvffeu9LH2nDDDbnhhhs488wzcXfGjBnDCSecULdvRlLllbOZ3QX8G2hvZtPM7Cgz+52Z/S7t8gQwiRhKdQPwh1qLVkREljuFcyd06NCBAw44oHTuhEceeQSA5557jvbt27PpppvyxRdflI7BbtCgAZdffjm9e/dm8803x9357W9/C8Dw4cO54ooraNu2LV999RVHHXVUpY+1//77s8kmm7D55pvTtWtXunbtyl577ZXBO1LElbO7VzpdlEdnwbE1FpGIiKxw+vXrR79+/ZbYVjjD2/7778/+++9f7t/utttujBs3bqntbdq0KXf2vYoeq0GDBrmZWEXTd4qIiOSMkrOIiEjOrLBza4uISH6ocHBJunIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREcmZopKzmfU1sw/MbIKZnVHO/Rua2bNm9paZjTOzfjUfqoiIyIqhyuRsZg2AEcAeQEdgkJl1LLPbOcA97r4lMBC4tqYDFRERWVEUc+XcHZjg7pPcfR4wEti7zD4OrJZ+bwZ8WnMhioiIrFiKSc4tgakFt6elbYUuAA4xs2nAE8Bx5T2QmR1tZm+Y2RvTp09fhnBFRESWf8UkZytnm5e5PQi42d1bAf2A28xsqcd29+vdvZu7d2vevHn1oxUREVkBFJOcpwEbFNxuxdLN1kcB9wC4+7+BVYC1ayJAERGRFU0xyfl1oJ2ZbWxmjYmCr0fK7PMx0BvAzDoQyVnt1iIiIsugyuTs7guAwcCTwHtEVfa7ZnahmfVPu50M/NbM3gbuAg5397JN3yIiIlKEhsXs5O5PEIVehdvOK/h9PLBDzYYmIiKyYtIMYSIiIjmj5CwiIpIzSs4iIiI5o+QsIiKSM0rOIiIiOaPkLCIikjNKziIiIjmj5CwiIpIzSs4iIiI5o+QsIiKSM0rOIiIiOaPkLCIikjNKziIiIjmj5CwiIpIzSs4iIiI5o+QsIiKSM0rOIiIiOaPkLCIikjNKziIiIjmj5CwiIpIzSs4iIiI5o+QsIiKSM0rOIiIiOaPkLCIikjNKziIiIjmj5CwiIpIzSs4iIiI5o+QsIiKSM0rOIiIiOaPkLCIikjNKziIiIjmj5CwiIpIzRSVnM+trZh+Y2QQzO6OCfQ4ws/Fm9q6Z3VmzYYqIiKw4Gla1g5k1AEYAuwHTgNfN7BF3H1+wTzvgTGAHd//GzNaprYBFRESWd8VcOXcHJrj7JHefB4wE9i6zz2+BEe7+DYC7f1mzYYqIiKw4iknOLYGpBbenpW2FNgU2NbOXzOwVM+tb3gOZ2dFm9oaZvTF9+vRli1hERGQ5V0xytnK2eZnbDYF2wE7AIOBGM1t9qT9yv97du7l7t+bNm1c3VhERkRVCMcl5GrBBwe1WwKfl7POwu89394+AD4hkLSIiItVUTHJ+HWhnZhubWWNgIPBImX0eAnYGMLO1iWbuSTUZqIiIyIqiyuTs7guAwcCTwHvAPe7+rpldaGb9025PAl+Z2XjgWeBUd/+qtoIWERFZnlU5lArA3Z8Aniiz7byC3x04Kf2IiIjIT6AZwkRERHJGyVlERCRnlJxFRERyRslZREQkZ5ScRUREckbJWUREJGeUnEVERHJGyVlERCRnlJxFRERyRslZREQkZ5ScRUREckbJWUREJGeUnEVERHJGyVlERCRnlJxFRERyRslZREQkZ5ScRUREckbJWUREJGeUnEVERHJGyVlERCRnlJxFRERyRslZREQkZ5ScRUREckbJWUREJGeUnEVERHJGyVlERCRnlJxFRERyRslZREQkZ5ScRUREckbJWUREJGeUnEVERHJGyVlERCRnlJxFRERypqjkbGZ9zewDM5tgZmdUst/+ZuZm1q3mQhQREVmxVJmczawBMALYA+gIDDKzjuXstypwPPBqTQcpIiKyIinmyrk7MMHdJ7n7PGAksHc5+/0JuBT4oQbjExERWeEUk5xbAlMLbk9L20qZ2ZbABu7+WGUPZGZHm9kbZvbG9OnTqx2siIjIiqCY5GzlbPPSO81WAq4ETq7qgdz9enfv5u7dmjdvXnyUIiIiK5BikvM0YIOC262ATwturwp0Bp4zs8lAD+ARFYWJiIgsm2KS8+tAOzPb2MwaAwOBR0rudPdZ7r62u7d299bAK0B/d3+jViIWERFZzlWZnN19ATAYeBJ4D7jH3d81swvNrH9tBygiIrKiaVjMTu7+BPBEmW3nVbDvTj89LBERkRWXZggTERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnCkqOZtZXzP7wMwmmNkZ5dx/kpmNN7NxZvaMmW1U86GKiIisGKpMzmbWABgB7AF0BAaZWccyu70FdHP3LsB9wKU1HaiIiMiKopgr5+7ABHef5O7zgJHA3oU7uPuz7j433XwFaFWzYYqIiKw4iknOLYGpBbenpW0VOQoYVd4dZna0mb1hZm9Mnz69+ChFRERWIMUkZytnm5e7o9khQDfgsvLud/fr3b2bu3dr3rx58VGKiIisQBoWsc80YIOC262AT8vuZGa7AmcDvdz9x5oJT0REZMVTzJXz60A7M9vYzBoDA4FHCncwsy2B64D+7v5lzYcpIiKy4qgyObv7AmAw8CTwHnCPu79rZheaWf+022VAU+BeMxtrZo9U8HAiIiJShWKatXH3J4Anymw7r+D3XWs4LhERkRWWZggTERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyRklZxERkZxRchYREckZJWcREZGcUXIWERHJGSVnERGRnFFyFhERyZmikrOZ9TWzD8xsgpmdUc79K5vZ3en+V82sdU0HKiIisqKoMjmbWQNgBLAH0BEYZGYdy+x2FPCNu7cFrgSG13SgIiIiK4pirpy7AxPcfZK7zwNGAnuX2Wdv4Jb0+31AbzOzmgtTRERkxWHuXvkOZvsDfd39N+n2r4Ft3X1wwT7vpH2mpdsT0z4zyjzW0cDR6WZ74IOaeiE1bG1gRpV75Y/irluKu27V17ih/sauuGuwg2AxAAAgAElEQVTWDHfvW8yODYvYp7wr4LIZvZh9cPfrgeuLeM5Mmdkb7t4t6ziqS3HXLcVdt+pr3FB/Y1fc2SmmWXsasEHB7VbApxXtY2YNgWbA1zURoIiIyIqmmOT8OtDOzDY2s8bAQOCRMvs8AhyWft8f+JdX1V4uIiIi5aqyWdvdF5jZYOBJoAHwD3d/18wuBN5w90eAvwO3mdkE4op5YG0GXQdy3/ReAcVdtxR33aqvcUP9jV1xZ6TKgjARERGpW5ohTEREJGeUnEVERHJGyVlERCRnlJwl1zTTnEi2zGwzMzu44PZy+ZnM2+tScq6GvP3nLSszq0//7yXj5w83sy2yDuanqo/HUJpfv14qOdbNbFUz26Cq/fPGzFYu+H3zNI9EXTudGI1zoZk1W56GyRYe24WvKw/fkarWLpKZreTui9KHpSswE5jt7p9lHNoyM7OTgG/d/casYymPmW1MzNveAjgA2KbslLD1lZl1BuYQn8FJWcdTkYLjvhFwFjAUmF8fvqALYm8LXAU0JmYuPLy+fG7NbAdgTWBXYJG7n2hmVpfvv5ntAVwB/AyYDpzt7v9M99VpLDWpJPZ0wnMysAj40d2vLrw/q/iyOAurl9x9Ufr1JuA7YEvgfjMbXp8OTjNr4O4LzawHsahJOzPbHLisZG70vHD3j8zsI+A84HmguZktdPdvMg5tmRQki98AewGzgIVmdrK753JGvYLj/loiOcxLX2YLMgyrKAWxXwZcDawMnO/un5lZ47SQT26lq7eVgN8APYDfweIrvLpKHu4+yswcWIeY62KEmT0EnOHuC2v7+WtLwXv3Z2Bj4J/Azma2J/Bbd5+SWXCoWbta0iIgjdz9aKAp8Go682qRcWhFS4l5XeBG4E7gL8C3wLVm1ifT4AoUNP+OAw5M/54M9DOzVczseDMragL5vEiJeX3gBGA/4krkE3f/2sw2yKjJskrpRK6Dux+TTuTuMLMnzaxD1rFVxsLGxPTCHwFnA8elu89MX8K5VHIi5+4vEBNAvQhsYWZHmtkaabcdzWyVWnr+Rma2q5m1MrPVgZeA3YHPiJP6zYCxZta+Np6/tpV8v5hZb6IlqL+7X+PuA4C3ic9nppScq+dnwL/MbDgw2t2fNbNNgYvNbLWMY6uOHsCLaXa3O4Gbiea+U8zs92bWIMu+0XR172bWDzjd3Z9y9/OIM9s9iSa2M4C3sorxJ1gPeADoBazt7uek7ScCbTOLqnLfAt+Y2SXAacBo4gvsiDz3R3v4CJgN3AXc6u7/TsntV8CETAOsRMlVv5n9HngGOIpY02Br4Ggzuxc42t1/qKUQLiY+b0OB24FdgIeJk/m13X0v4j2dX0vPX+vSd9xpwElmtmvBXU8CO5lZkyy/B5Wcq+dFIjn0BoakbRcAX7j7t1kFVayCA20KsJeZ9Xf3Be4+kfjCfRPYCvh5lk316eq+GfFFcDOU9tH+h1hy9DZgV3f/IqsYf4L3gfWBu4FhAKmZext3fz/LwEqU/UJy9/HEe74IONfdb0q/z8tbs2ZhIU8qAmtGJJfGwPpmdjKx5vw/8vJ+l1VQxDYAOASY6e4z3f06YlrKeUTNy8m1GMZ9xDH6DZGs9iP6vlcGegK4+yV5rpeojC+2O1FLMcrMTkpFg72AT919bpbfgyoIq0RBH+G6wOru/oGZDQR+SSyTaUALd98l00CrUNDP3AT4nqg12JVIdNNIVxXAHsBw4Hp3H51VvADpfd7F3Y9OX6h7AtsBv3L3R7OMrToK3vu1iTVmZxJXP3sSX3RziS+DQe7+TnaRLlZw3B9DVMvvAJzl7v9O9x8PHOjuO6TbuSsKMrP/I67qDk0//yX6bL8kkt0tGYZXJYtFhsYQXTqzgV8T8Z9Um59NM1sHODMVnrUB/gAsBN4AXk6xrOzu02srhtpUcGz3JVqxfiBORLZO/zYBhrv7pWn/BlmdgCo5V6DgP3FH4HKgEdFvdSCxbGZH4urhjTxXEBd+cZrZSOI1dAL+SjSTnUIUJj0DfEAk6q3dPZOCn4IKyjZEk/t6Kdbrgd2ANu4+LIvYqqvMe/8EUem6AXE8fUkk67WB/7r725kFWg4z6wTcQhzvdwIPufvQdN8A4H/uPi7LL6+KpL7k4929j5m9BVyV92RclpmtSbz/44hipQ+IY2Yvotr8y1p63nuI4/FPBduOAnYCxgOjgHfdvd41Zxd8t3QhFmsaSXQXnO3uD6Z9riZGhvy+ZFtm3F0/ZX6AlQp+vxfon36/E/gE2D3rGJfhNf2ZqFrdFXgXaEk6OUv3r05cPe+SUXwN0r9NiaFTrYg+/s5p+3rE2fuuWb+Xy/DahgFnptc1BVgrbW+UdWyVxHwe0I8oAnombVsVOKLk81F4/OTpBxgM7EgU3t2VtrVO/w+r5zjulcrc3oTo1tkr3d6cuHptUEvPfwrwVMHti4HW6fetiIr9m4jWwszfr5/wOm8Etk3HyHNp29rAxun3gcDYsv8fdf2Ty+rQLJVcMaffDyWagCcDuPtBZnYg0T9xqrv/ObtIi5P6DxsTTfCnA9cBN7v7J2bWMzVj3U8MDxvq7u9lEaMvvvr6K/FBmUWMA74ybb+YSBJP13V8y6LgLN2IZsGXiT70P7n7V2a2LzGM7TJP3whZK9M8/TLRHNyRSNIARwLbefQ5k5e4C5lZd6Kr5lygqbtvn+46ifiynZlZcJUo6P5oTSTJecBH7n54ur8L0ar1R6+Flgoza0pcHX+WiluPBjq6+2QAd38zdS/t6PVkjHh5LEZEjCe6JvsBg9JdfyBOPk9195Fmdo8vHoqXCRWELe0YM9sw/d6Q6IMYaGl2Hne/G2hGnH3lXvoCnU8UdkwgznovS3cPA37mYUEWibkgRszsDODnwADgIqKf8NjUxHeBu5+eRXzVlU7wPBUjdQGeI5qyG/jiCV9OBT7OU4JLMW+fqrKfBVYhWjDWtRhGeDgxHCkXMyiVleoUjnD3h4hjZzMzOySNrtiaqIjPpTInp28TtQj9oXSWsHnAiNo6OXX374ji1nHAP4jjc3B6/kbpu+97T5OP1CeFx6pHd91TxInIOHf/XxpxM4B470tOlDJNzKBJSJZgMWbwf8CnZnY0cAcxvu9kom9ilJmNdfcvsiyxr4rFWNoDgQ2Bh939OTO7kujvbJqqg7sAk939tgxDLWsO8Ki7zzWz94kvqMuJprU3sw2teAUf7OFEM/bfiOPIzewmorn+Q3cfmVGIlfmYmAGvvbsfYGZnATcALwCXuPvEwtalrJnZBu4+NbUAdST6RHH3k83sNaLpchxwh+e8n9TMtgUmufsNZvYGcGy6a1/gaXe/tpae9y9EdfJwM5tO1EbMAw4xs4c9p1XtxSioHWoIHEx8p3xNjD5ob2YfEN1817j7pLR/LmooVBCWmFlLoLm7jzWzzYjy+m+IwoEPiErJPsRMWs9lFmgRzOwx4Avii7Y/UWn7pMWEDNsRTTovALe7+3dZfdmWrfJNFZTXp3hvT9ueAv7i7o/VdXzLwszapA/5dsCFRH/hDyl5/IJoOvsKGJWXLwFYqnjtBGArdz803W7i7nMzDbAcZrYTUbxzHPH5PBV4lBjuNTPtk7tK8kIFzdmrAWsBlxJDlu5z97+mZu4HieOoxmfwM7OeRAva3u4+3cxapi6v3qSx+MAL7n5XTT93XTKzu4h6oW5EN+UxRAtoE+A7z2NRb213ateXH2I84YtElR5EAclpRAHEr4k+2+5kXCRQxOs4laisLbn9a6L/qDWLizsaFNyf+esBOgCbEk2oOxKFd2OAEcCTWcdXjdcxhOi3J73nY4kmwtZZx1ZE7D2BNsQJKsA1RKvFytRSAdJPjLcR0S++c8G2X6bj5mJgozwc21W8hsLC07uJ5HwScVJ9ONCcGGN8Yi2+h9NIBa7EPPZ3Fdy/KdEVcFbW79VPfJ09Sr4TiSbtQen3LuS0ONDddeVcwmLWoG2ID8Vs4gM+DdgH6EuMDz7N3X/MKsaqpKKOR4EPiT7azyzGql5ONKtuRVQ8j3T3W7OLdIkrhr2Ik6CSprRpRBHYL4kuhjfcfXZ2kRYnXYFcCfTyaI1YnThL341oxn7Mc9hfV1C4dhlx8jmeuHJ7iSgIO8BzONGEmd1IJJP1vKAFwsxaEfUK6wJ/9npQQJiGpm3u7heY2c+I2bjOJo7/L9z9tFp63g2Jq/L3ie+Iq4lWh+cK9lmdmN5yTm3EUBvKaZHbgrj4WgtY4O6/NbO1iO/KI9z9g4xCrdQKn5zLNOcZMabwMKKA5CZ3v9/MtgTWdPdnMgy1KBbzfJ9PDNu5g6haPc1j8vp1gHbu/lKWMZawmBTlP0Rf0MfEF+pg4CVPzdr1gcWKTe8ScR9hZo089W+mL4aexNCNxzwnzYPlNfemk7tViBPSdkSyXosYSji5zoOsgMWIiROIrpneRBK+s8w+pxH9t/dlEGLRUjHS+8AT7r5nmftqfQx5qrM5k+gaGOtpQqV0TC8oe4zUB6mAbmWi2vyVVBB2ATHxz7EeU7j+jViB6o8ZhlopJefFVw7tiWkV3yJWXtkF2B+YQczv/F2GYVap7AfZYuKAI4giq75A48Kr/jz0xaV+thuBQz36ZRsTYwx7En1Ci7KOsRjpKq4ZMSnNt8R40Hc9rXpkMcNcL2JMZa1MHlFdBYUyhxOT0nwPvOxp9ilLqzaZ2TnAm+7+RIbhlkrHzEPEkJf/WAxJO50o+jrea2+u6RpTzpXdNkRXzsfAPiUtRbX1GU1X582IC47xaVtv4qR+MtGM/lVNP29dMbPriXkRNiFGqpxDDM0cRLzuFsQscful/XNT4Fgod8Mh6pItHvLSkxjsfyjwHvEf+CBRKLEw3c6t9CFemH7/q5l1d/e/E0303xIzDf3CKlhYvI5j7ZFOhPCYj3wh8LCZdUnJ7AviJMnrSWLeAtjA3X/l7gcSlaCXAgemrhLc/Qt3vydHidlSYu5CXIE+Q4z5/IOZDU0V0CXLKW5GTESSF/OIptf/pM/vg8RJdAPgGTPbLdvwqlbQUredxXrNn7p7O6L15UMzG1S4Xy0oWZHufDM7zcxWTq2CA4guvZcsxyt2VcbMriL60o9195JZ7s4mupiOY/HIm8PS/rkYNlWuuujYzvsP8ARRHLAnsdoUwBrp34ZZx1eN1zGcxTMiFRabnA/8Xw7iu5QoyPiUuMqBOEH8A9FicSsxxrPezAJGFO2sXWbbXsDT6f+jIzktTCL6GPsQYz6fJ2aPexV4jEjKRlRD5+IzUFkcRHI+krjy2zzrWCuLM/37B+Bf6fNwPXBO2v4ranF2KqLb6A5ihsABxGiUssdvybz2mb9f1XxtexAFgY3KbO9CDGk8PusYq/OzQjdrp76I1YA/EoVSFwH7uftHZnYd8Ja7/y3LGItlMVHHzUST1MR0Nvyjma3n7p+X9INmOGxqJ+BSd++erpyHA48T8wV/TqwTuw0w3jOaDKW6bMnZ5Er7mdPttYkpU5sAB3kOxtgWNGWXNFl3J1qKbgfO9xhGeD0xOcM12UZbMYtJUmYRV3mPuvvUgvvWd/dPMwuuCOmz+jwxMmEOUah5GPCsu99TW5/RVHPyOdDb3Z9N2/5OHKNfAGsQy0Q+5+6f1PTz16bUKvgS0QqxX8E2T8f8scSoiVOzjLM6VshmbVs8Y0wjj/GQKwFXAZenxNyZKL+/P6sYi2FmvzSzTQDc/WtgIrBPSWJOu12Zmrnnp/2ySMwNiHVhSyaS34gYb70Z0VpxMPCZu99f3xKzmTVLTWkjLCZ3AcDdZ7j7YUTfaOaJObnYYmGDUy1mfHrNo39zOrCvmXUgisDug9KpDnMlVTbvSvSRbwScaWalze55TcxmtomZdUw3GxIno2u7+zx3fwX4N7B3KsqrFR7dKjsBt5vZH9LmnimWScRxsFO6Xa94dOsdQEyyNCZ1ky0kWn8gZkdsnupa6oUVLjmX9DFYTK3411Rgcj1x5XxgOpP8G3Cx539ZtGbA1wX9bA8RE10MMLP9LYbHNHH31zKLMKxN6stP/WlDiRaKk4l+/U7EkKN6o+Ak52ri/+FJ4Ggzuy8dWyX7Tc4gvKWkwq8exLHdDzjAzFqY2XpEi1F/4v9lRGppWckzWpmsrJJaifTFujrROnQ1Mdf0+8CvzOz0wpqKPEmVz38nWulKkuSLwAUFJxY/Aqu4+3e1eQLt7s8Tlfj9zexb4HF3P4no9jqF6KvNZx9sJdLx+rG79yG6lB4ysxN9cZHsmUTB4zyz/M7uWGiFbdY2s4eJZusL0u0GwPZE39XcHCS0SpVUcqbKy38SzcO/AdoTV6I/IxazONfdZ2ddkZi+WI8lrnoM+G1J05mZPU0khWyXaKsmM/sFcbZ+aupCaEKs/DWAWNYv0zWxS1jMfvcyMMDd3zCzYUT1+HSigO0Gd3/JzNb2nM2UVFixbGZ/Jgp7vnH3fdK2dYgTu5nu/nh2kVbMYtjOD+5+QroybkW87/sTJ0WrpJ+j3X1cHca1DzE165Fej9ZIL2RmPyfWB5hhSw9hvJG4KJgJrOppEZF6I+tO7yx+iD6ex9Pv/YBHiKk660UhEotncWpEFG+sCVxCFFPtWbBfw/RvbgqSiGXvbgCuIL5ohwL3Zh1XNV9DyUpfdxOTdgwkFqAvuf8gYJus4yyI5xCi8KgvcXL0DjHP+qbEjHLXZB1jZe91+vf3xPKtWxNFa48Dm6T7cnN8lxP/rkR3U+t0+z5i3gGIqVybATsArTKKb21iCNeQrN+rZYz/ceJk+GclxwswsOD+a9P7v1rej5WyPyvUlXPB1WY7IiFPJiqHbwHWIfpbTvZ8zwLWiKhK3ZKouLzP00LyZrYHMWzgG6Lqc75nOH+zLZ4FrDuRlD8HXiH6C38P7EyMRezp+e9CKG98qhGtFQcRQ1Me8ByOD01XawOALYjq7NEeXQols0TdQ8wE9nF2UVYsxXgDUVD4TNp2GXFSdIHHsMFcSl1OxxDD1TYEurt773Rf5nMNlMRBjE75OutYqsPMfg/80t33LKgBWY/4LF7l7o+k/Zq5+yyrg0ldalLuCj5qQ8mHoOSD4LFM2B5E6f1d7j4zNT3NyHNiBvCouL6BqPbsRPR5ltw3ysz+S1QHZzoZQ/qwLLQY63s9ccVwLDEj2P3u/meLVYPm1YfEXMhi+cT1iP7PvxPFPJcAO5jZEHf/KMv4yvKYQOeW9H7/ADQ0s18TX2LHA6/lNTEnbYk+2XNTIduT7n6qmY0h53MQuPtTZjYV+C1RfDWypGAzXShknqDT89erxJysTkz3C7BmOgldlbgAOMnMDgauc/d/wRLLctYLK8SVc8EV8zFEhfDnRNHA2JRAzgV2c/eemQZahYIr0YbEOMn5xCxgr7r7cWmfk4j+2x/zcKZoZtcQ4zafJPrGbyMmtXiGWG1qVobhFa3gvd+OWJDjHqApMdXlWcTxdBkwzAuG9mSp4Gpia2Kd7IXEicTRxCIXmwAbuvs2hftnFnAlLGZZO4ioqXgXuDXPx07B8VLYZ74v0YLxDjAaeMdzUnRXH1ms2HUxUQT7PTEXeWPiAmAnYkrUizIK7ydb7pNzQWLegVhp5wyiH+IV4DVgJPElOz/PVw9lPuSXEtWVUy3m5j0P6EzMBva4uw/PMFTMbHtimc2ZRMHUE0Rl7eXu/i8ze5Kojj8n66uG6jKzm4lxoDen27sQyxUOyvpEqFBBYu5MXCFfC/wf0MNjdq2diWbhv3qMb878RK5EQWLrQxR77UC0TEwiTq77EcVs5+Ql5opYjMmeQ1wQ3E1MHXkcUSdymbu/nWF49VZBxfUaRItES3cfke5bCXgYOKQ+NmeXWO6HUhV8+R9N9HP+jCgQeDxtuwiYnefEnBiUftg3Som5MZEAf08scHF/SWLOarhASlZnEgvEt/BY6OE7YtKIVdJuDYE760NiNrOmFhOolBhHdCcAkJrMGhP957lRcAX8Z2IZwqnAMykxr0cUVR2fErPl5curoDtkVWLugX8Ry7b+jij0eYBYhvPWvMRcEVs8Jns2sSzqpaShYERL0n+zi65+KvleK+imnE108Z1kZsdarDb1MNFVMyvtm+vjpCLLdZ9zmf6cG1h8pXy4u39qsVzhm56TOY8rk66CNiKahPunoQKnEn2f93rBTGZZNk+mK+PGxDzl21qMG3+NqLS9zczeJ1ZvejeL+JbBrsAnZtac6M+6B3jAYjKPE4GNiUKfN7MLcWkFx8AY4qrtRqIrBKKgcK67XwXZzbNennScGzE5zXh3HwVgZo8Aj5rZRE8FkHlUcNVfOCb7JVu8OtmxxDKouZ2BrT5IxWCNiZaIq4hj+q9Ec/bn7v6ntF/mffrLarlOzqk5uy8x1/G3xPJyqwFHWIytbUtcdeZWOvue6+6j3H1KKoIZThTC3E70Iw4ys5HArHRCmUliTsU6C4i+TCOWSWxGTLF4W4p9A3cfm0V8y+gxd19gZhcR/ba3EMfR+cAoov9zmOek0rXgy6gNMSsSxBC7P7n7m2bWlThx6ptVjBVJ3TVXeqxD/iIxmU53om/2CzO7lnhduVSmBWIoaUw2cTI61sw+Bb4iWpGkmgq6avoQ9QdXEJXwL7r7Y0BHi9EsJYsA1cvm7BLLZZ+zxSxUs9z9CTN7AXiASM4bEdMTbgx8BNzo+V/vdSuiGX5vou+2OdG//B93n2RmpwKdPCcD7C3mzb7X3buk24eQ1ool9W9mGd+ySAV4HYgJI9Yl+stvI2Zfy+Ui9GZ2LzFZx9GpSPBk4mSiHbFO+c15KgCzmEziCmIc88Xu/qCZnUl0ITxDFD+eTEz48nR2kVasoL7l98SSs8OIvv4ZRBfCxDy95/VRalW5gSgE60EMAdzXzDYAWgP/Xm6K7DwHg61r+gf4NXFF8wRwUdq2CjHpwmXEAgurZB1nNV7PSsSKTW8S4/pKtu9LJIrcDLAnvkzHAFsVbBtI9NX2yDq+6rzn6d9DgPPS72sQs6+NIL54t846zkriX5MYZrddut2BOMHbLOvYyonVCn4fRpw4X0i0vuxNJO2rgd9kHWsRr2VDoj+5d8G2y4g+/6Oyjq8+/hCtnRux+GLyBKIF642S73GiafuSrGOtyZ/lsiDM3W8jmrK/Ak4wsx3d/Qd3/5D4wB/sOV+UPVUcYmY/d/dF7n4o8cU1xMyuTVdz/yGm3vvWMlqX1MzWNLPtzWwzAI++5NFAr1QhD9G0Pdpjgv96waP5rC3RfF2yAMpaRN/5FcTsbN9kFF6FSo4bj2b2ccB1ZtbJ3d9z94fd/f20X+7mF7YY0vgzouCrI/Ac8IG7n+Tux7v7jVnGV6TCMdm7A3ishPR7Yoy5VN8JxHwCO6bj+z5iAqaPgFVT1+UviO/HXB7by2K5bNYukf4jzyCWY3uYKAYbQgzpGZNlbJUpM2zqLqAr0M/dJ1vMlvQnYlWn3p7xmFozG04Upt1BXJ0NJZpONyb61rYk5isf6PWg8M7Scorp93OB79z9SjM7mbhq3oCYaWum56RJO/Uj/5foE3+ZWDr0f8SUnYOI6WrP9FiBKpdSBflTQC93/9pirvvhRKvLde7+p/pS3GP1bEx2fWBmpxF1Es8TQ2I7EiMkDiVa6u71mISpXvczF1our5xLpCvOS4j/wD2I/9g785yYk5JhUwOJop6vgbfN7ASPlVcOIwp88jDZxT3EOqp/J04aOhJ9QbsSV0FTgcH1JDE3AA6yWEwE4AXgj2b2BnGCsRMx29k2OUrMaxFTchoR42+I970nMe93T2LIYI+sYizSV8QKU92hdPjLLcSQx/FpW+4Ssy1eMauPmV1mZi8T3WYlk+50Bc6wnK6YVY+sRcwGdgxwOTDH3YcQJ/9HearqX14SM7B89jmX90NMNPLrrOOoRrxtgClEdTPE1dqnRHFb04L98tDPfCxREfxzoihjCnAUUbQxIOv4qvE6WhEnFWsB+7G4TmG3gv+T98hRvy0xnGQ1okjwfmDngvs2IZL1BVnHWeRrGUg0VZ5JnGycDwzPOq5K4i2pS1iVOIHYg5im8zHgwnTf9kCHrGOtjz8sbtk9DHgy/d6IaJ2bQgxlXD3rOGvrZ7lu1q7PzGx9og/lMC85UmNY1XlE39VB7j4xwxCXkJqdZhJFaq+6+wUFQ6tyr0xXwr7EuMlniDWypxKtTKOJYWFXZRZogTIxtydOLLYgFnS53d2nFO6Xpya/gmExOwE7EiMSxhAneNcS1dkNiBPqLzILtAqpf3MgcRK6f9q2LvAoMY1ubsdk55mZtXD3z9LvA4mREsd46poxs78QLVm7u/vnmQVai5brZu16bgZxRn53wba1gL8RfXOtsgiqEs8SM1Ft7mmNbNJ4w/qgIMn18lhX+lLiavR4YlhMyYQSuUvMyace0xeOJKrKTzGz/rD4teUwMa8K/IU4lnckhkpt4u67EnOW75nXxGxml6YE4sCLaVt3M2uSYs71mOw8M7OSpWRLPE60pPRJQ6YAFhDV2Z8vLwVgZenKOSds8cxCmxBjmZu7+6MWq2X1Jq7a+hFLL95IjOf7v+wiXlqq2D6XmMP59bwkg6oUJIt9if7ZX3ss3r4mMW1kV6IP8SbPyQem4Gp4CHHStjrwlLvfYmYtiQVRFrj7sEwDrYSZnQ984e5/M7MOxHG+KdEC89eSK6e8WR7GZOdVmkTkVGLtg5WJAt5+xInOucSxsQ6xFsKe6W/qRaFgdSk550DBF21jogjpP0TlczPgl8TBuAoxDWMX4Cx33y6reCuSzmAvJPrizs46nuowsyZEYdtvPOafbuTu89N9/YDp7v56pkEmBScTuxH9sr8lKrTnEHUJ57r7R2b2M3f/Po9fXmbWiijQnE6s5/1j+j/Yk/giHp63mGGproRhwIHEhDTnE02vvYiZF8d5/Rj6lSsWyz5eRJzonEQUOx5C1CE8TRSc/gz40N2/yVNXTU1Tcs4RMxtKFDj8Pt0+jiiG2DM13+03TwoAAAcgSURBVDQlvgxe8BiznUtmtpq7f5t1HNVhZm2IIRqHe0FlucWasI/m5fWY2Sqexuib2eXEELZORAHb8cQY0NWAvfJUkwBLX+FYLPl3JTE//Jnu/lza/nPPSTV8WQUn0ucCaxPdT12J1q5jPI0jl+orOZlMv19JDJX6A3FhchHwFnC6x/rky+0Vcwn1OefLeNJ8yGm87f8RqwdtApAOylvznJgB8pLIqsPdJwEfAr8u6ddKk0gcTqyqlTkzWx0408x2TJv+TBRS9QeGuvs0ov/zkjwnZjPrZGY7u/tkd9+XWHXqCjO7Ou0+N7NAq5AS83rEUqhDPBZYOBB4HXg6Je3lZiKMupKK6K41s4MA3P1EYiGLfYCPiZPP9YmmbdI+y21ihuV84Ys8KzgDLzz7mwpcaGaz3f36dKXcg5iVCoCSplb5aQr6+FclZnX6hBiXujdweupv3pQ4U8/LXMitiSa9krmEH3b3OWY2lVj8pB3RDHgeZLs6WUVSU3ALYsWyD4kr5ustxpIfAvXiS7dwTPbodBzdwuIhVfXhNeRND2IoWu9UEPYScayXtAJdSRz3DWH5v2oGNWtnJlV1zk2/l/abmNk2xOQLU4B5wKR0Fik1pMxV3Chi6NHuRLPwRGJ+5NWBKZ6zKUctZojrQ0y+8BnRjD2LuIr+mlhFa3ROE3MP4C/uvq2Z3U6cbGwEXObuV1f6xzmThvcMJSalGUacEDVx99MzDaweSy1CBxFrNL9PLDJzCFF/M5JYM8HzdlzXFiXnjJjZc8B/3f24dLsRceAtSLf3AV7xNIYvj1+29Z2ZHQ+0c/fjzGwKMUXqu4V9X3mQ+ma7EcWCjd19ampy70qcRDyYl2K1yqRq+LlEH+Jv3H0vM7uAKPbZ2d1fzjK+ithyMiY7j1IXwdbAc6kVaG+i//5xj6VDdyPm1n7V3S/MMta6puSckdQ88xdijuwBqbq2pKl1LWIqzDPqY/9tfZDe/5OJfuZ9iAUWLklfBvsCJ3iaYztrFlNC9iBmnppKXDU/Rgw9mk1MPHKmu9+VWZDlsJgGtRmwhru/l7YZUYXbyt1PNLNjiCLI4RmGWiFbckz2i8QQn0XA98TQtVFmtjJxYp2L46U+MbMTieS8PlGQ+RqxHsJs4Oz03he2Mq4wFylKzhkzs18Qs1D9yd3/kra9QFRkn5VpcMuhlByapLP0HYFzWHLM5GjirD03Y8gt5mUeQgyZ2pG4YluDWAhgZWKmpCPc/c2sYiyPmd1B9C9/SVTaXunu8yzWKD+DSHDbEctafpVdpFWzejomO+/S53E9YqKf84gJliYSc2i/RAwL/DS7CLOj5JwDaXzn/Szua/mFu++S7lvuCx/qQirw+i3RFNyGKLL7OzFUoxtxRdoCaODuA7KKszIWS3A+SFxR3JC2lV5J5OmqwswGE4n3NBYX+5zh7jPS/T2IYp/v8tqcXcL+v737d/E5juMA/nwOpBsc0pVBMlBWk8TlTsIgm+s6EYPBcF02y90fIJNY5McgJlJS1EknFukyGNSd2AxKplNKT8Pr/TmfvoMkd+/39/15PsZP3+E1XPd6/3i9Xu8+7cnuN2kRegbxdzGCmPFwSNKzrIFl4uRcEJLjiL7VnZIWa26wX20kHwN4B+Atojf1CuJZxQnEYIOtiPvQuaaPskSthRwBnCjx2oPkEGJgzkFJz9O3GwAGELvodYi3mmclfckV55/U0JPdr9IVggCMN4vQLnJyLkzqb/7hxPz/pF3cfkljPd9vI+5vR0s/Vu3FGI5yGcD2korXGiSHAdxDjLe8RnIB8RjEJ0SF9nrE0I4idvptvT3ZAIZai4xziJGuLyVN+mRr5XX1f6GTs1WttYsbkTSXjs7W6PeUrVkAM5Je5YzzXzBNCyvpOLst7fIfIJ5NvClpqtXfv7bUAqpWjMs92YjCwYupmn83gJOSLmQN1KrmCWFWNcUozgMA7pKckvQzJbSB9JP3KO+Fr7/SLDBKTMwAIGlJ0hEApwBMkDzW7DJLTczA8hSwPYgF3WkAbwBsAvCE5KSkeSdmW2lOzlY9SS8A7ABwmORTkhskLTHGYY4A+Jg3wrpJeogYJHGV8YpWP9gCYDr13Q5K2ocoILxEcm/e0KwLPL7TOiH1SR4lOQZgIRXfHQdwX9LrvNHVT/EE5zZEC1hxenuyFc9ANj3Zi+lnnwFMl15dbnXwnbN1DsnNiKPK75J25Y7H8qupJ9vq4ORsnZR2RYOSvuWOxfKqqSfb6uHkbGadVUNPttXJBWFm1lmtav47JM+nz8OI++UPiLe8RxHPRJqtGu+czazz+rUn2+rl5GxmlqSnWq8DOCvpUe54rLucnM3MWlI1/zyAW5Jmcsdj3eTkbGbWI1Xzb5T0NXcs1k1OzmZmZoVxtbaZmVlhnJzNzMwK4+RsZmZWGCdnMzOzwjg5m5mZFeYX974U9Cp8anAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f126ec98860>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11f93dce48>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1217f77fd0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_stats(Acc, \"Accuracy\", 8, 55)\n",
    "plot_stats(F1S, \"F1-score\", 8, 55)\n",
    "plot_stats(Prec, \"Precision\", 8, 55)\n",
    "plot_stats(Rec, \"Recall\", 8, 55)\n",
    "plot_stats(FPR, \"False Positive Rate\", 8, 55)\n",
    "plot_stats(FNR, \"False Negative Rate\", 8, 55)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Results comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred[\"RFFM\"] = class_id\n",
    "y_pred[\"Wide&Deep\"] = y_pred_NN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Metrics of baseline method\n",
    "\n",
    "name = \"baseline\"\n",
    "\n",
    "if(y_test.mean() > (1-y_test.mean())):\n",
    "    y_pred[\"baseline\"] = np.ones(len(y_test)).ravel() \n",
    "else:\n",
    "    y_pred[\"baseline\"] = np.zeros(len(y_test)).ravel()\n",
    "        \n",
    "y_pred[\"baseline\"] = y_pred[\"baseline\"].astype(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/preprocessing/data.py:625: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  return self.partial_fit(X, y)\n",
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/base.py:465: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  return self.fit(X, y, **fit_params).transform(X)\n",
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/pipeline.py:331: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  Xt = transform.transform(Xt)\n"
     ]
    }
   ],
   "source": [
    "classifiers = {\n",
    "               \"AdaBoost\":AdaBoostClassifier(DecisionTreeClassifier(max_depth = 13), n_estimators=num_trees, random_state = 0),\n",
    "               \"DecisionTree\":DecisionTreeClassifier(random_state=0, max_depth = 13),\n",
    "               \"NaiveBayes\":GaussianNB(),\n",
    "               \"RandomTree\":BaggingClassifier(DecisionTreeClassifier(random_state=0, criterion = \"entropy\", max_depth = 13)),\n",
    "               \"RandomForest\":make_pipeline(StandardScaler(),RandomForestClassifier(n_estimators=167, max_depth=38, random_state=0)),\n",
    "              }\n",
    "\n",
    "name = \"LogisticReg\"\n",
    "model = LogisticRegression(solver = 'newton-cg').fit(X_train, y_train)\n",
    "y_pred[name] = model.predict(X_test)\n",
    "\n",
    "for name, clf in classifiers.items():\n",
    "    \n",
    "    model = clf.fit(X_train, y_train)\n",
    "    \n",
    "    y_pred[name] = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RFFM62872\n",
      "Wide&Deep62872\n",
      "baseline62872\n",
      "LogisticReg62872\n",
      "AdaBoost62872\n",
      "DecisionTree62872\n",
      "NaiveBayes62872\n",
      "RandomTree62872\n",
      "RandomForest62872\n"
     ]
    }
   ],
   "source": [
    "for k in y_pred:\n",
    "    print(k + str(len(y_pred[k])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11f90f25c0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred_pand = pd.DataFrame.from_dict(y_pred)\n",
    "names = [\"LogisticReg\", \"AdaBoost\", \"DecisionTree\", \"NaiveBayes\", \"RandomTree\", \"RandomForest\", \"RFFM\", \"Wide&Deep\"]\n",
    "correlations = y_pred_pand[names].corr()\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cax = ax.matshow(correlations, vmin=0.5, vmax=1)\n",
    "fig.colorbar(cax)\n",
    "ticks = np.arange(0,8,1)\n",
    "ax.set_xticks(ticks)\n",
    "ax.set_yticks(ticks)\n",
    "ax.set_xticklabels(names, rotation='vertical')\n",
    "ax.set_yticklabels(names)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              LogisticReg  AdaBoost  DecisionTree  NaiveBayes  RandomTree  \\\n",
      "LogisticReg      1.000000  0.736223      0.681495    0.796965    0.671064   \n",
      "AdaBoost         0.736223  1.000000      0.874258    0.632475    0.881585   \n",
      "DecisionTree     0.681495  0.874258      1.000000    0.607255    0.979087   \n",
      "NaiveBayes       0.796965  0.632475      0.607255    1.000000    0.600572   \n",
      "RandomTree       0.671064  0.881585      0.979087    0.600572    1.000000   \n",
      "RandomForest     0.701221  0.911766      0.924282    0.612663    0.931205   \n",
      "RFFM             0.700537  0.826565      0.840071    0.603339    0.839829   \n",
      "Wide&Deep        0.002712 -0.002107     -0.001643    0.002297   -0.002140   \n",
      "\n",
      "              RandomForest      RFFM  Wide&Deep  \n",
      "LogisticReg       0.701221  0.700537   0.002712  \n",
      "AdaBoost          0.911766  0.826565  -0.002107  \n",
      "DecisionTree      0.924282  0.840071  -0.001643  \n",
      "NaiveBayes        0.612663  0.603339   0.002297  \n",
      "RandomTree        0.931205  0.839829  -0.002140  \n",
      "RandomForest      1.000000  0.836000  -0.002394  \n",
      "RFFM              0.836000  1.000000  -0.001524  \n",
      "Wide&Deep        -0.002394 -0.001524   1.000000  \n"
     ]
    }
   ],
   "source": [
    "print(correlations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {},
   "outputs": [],
   "source": [
    "failures = {}\n",
    "failures[\"baseline\"] = []\n",
    "failures[\"LogisticReg\"] = []\n",
    "failures[\"AdaBoost\"] = []\n",
    "failures[\"DecisionTree\"] = []\n",
    "failures[\"NaiveBayes\"] = []\n",
    "failures[\"RandomTree\"] = []\n",
    "failures[\"RandomForest\"] = []\n",
    "failures[\"RFFM\"] = []\n",
    "failures[\"Wide&Deep\"] = []\n",
    "\n",
    "for i in range(0, len(y_test)):\n",
    "    for key in y_pred.keys():\n",
    "        if y_pred[key][i] != y_test[i]:\n",
    "            failures[key].append(10)\n",
    "        else:\n",
    "            failures[key].append(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f126ee20f60>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "failures_pand = pd.DataFrame.from_dict(failures)\n",
    "names = [\"baseline\", \"LogisticReg\", \"AdaBoost\", \"DecisionTree\", \"NaiveBayes\", \"RandomTree\", \"RandomForest\", \"RFFM\", \"Wide&Deep\"]\n",
    "#names = [\"baseline\", \"LogisticReg\", \"RandomForest\"]\n",
    "correlations = failures_pand[names].corr()\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cax = ax.matshow(correlations, vmin=0, vmax=1)\n",
    "fig.colorbar(cax)\n",
    "ticks = np.arange(0,9,1)\n",
    "ax.set_xticks(ticks)\n",
    "ax.set_yticks(ticks)\n",
    "ax.set_xticklabels(names, rotation='vertical')\n",
    "ax.set_yticklabels(names)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              baseline  LogisticReg  AdaBoost  DecisionTree  NaiveBayes  \\\n",
      "baseline      1.000000     0.006450  0.195628      0.311228   -0.133475   \n",
      "LogisticReg   0.006450     1.000000  0.214494      0.083524    0.619317   \n",
      "AdaBoost      0.195628     0.214494  1.000000      0.388934    0.123356   \n",
      "DecisionTree  0.311228     0.083524  0.388934      1.000000    0.068659   \n",
      "NaiveBayes   -0.133475     0.619317  0.123356      0.068659    1.000000   \n",
      "RandomTree    0.338340     0.068396  0.451398      0.917762    0.057346   \n",
      "RandomForest  0.292199     0.080400  0.485216      0.646148    0.041950   \n",
      "RFFM          0.302155     0.238220  0.382548      0.493401    0.133097   \n",
      "Wide&Deep     0.740641     0.007839  0.142172      0.228836   -0.096606   \n",
      "\n",
      "              RandomTree  RandomForest      RFFM  Wide&Deep  \n",
      "baseline        0.338340      0.292199  0.302155   0.740641  \n",
      "LogisticReg     0.068396      0.080400  0.238220   0.007839  \n",
      "AdaBoost        0.451398      0.485216  0.382548   0.142172  \n",
      "DecisionTree    0.917762      0.646148  0.493401   0.228836  \n",
      "NaiveBayes      0.057346      0.041950  0.133097  -0.096606  \n",
      "RandomTree      1.000000      0.699414  0.506329   0.248333  \n",
      "RandomForest    0.699414      1.000000  0.436441   0.213258  \n",
      "RFFM            0.506329      0.436441  1.000000   0.222527  \n",
      "Wide&Deep       0.248333      0.213258  0.222527   1.000000  \n"
     ]
    }
   ],
   "source": [
    "print(correlations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({1: 54763, 10: 8109})\n",
      "Counter({1: 58330, 10: 4542})\n",
      "Counter({1: 61666, 10: 1206})\n",
      "Counter({1: 61277, 10: 1595})\n",
      "Counter({1: 54527, 10: 8345})\n",
      "Counter({1: 61167, 10: 1705})\n",
      "Counter({1: 61725, 10: 1147})\n",
      "Counter({1: 60341, 10: 2531})\n",
      "Counter({1: 51360, 10: 11512})\n",
      "[[54763.0, 58330.0, 61666.0, 61277.0, 54527.0, 61167.0, 61725.0, 60341.0, 51360.0], [8109.0, 4542.0, 1206.0, 1595.0, 8345.0, 1705.0, 1147.0, 2531.0, 11512.0]]\n"
     ]
    }
   ],
   "source": [
    "cont_table = [[],[]]\n",
    "for name in failures.keys():\n",
    "    print(Counter(failures[name]))\n",
    "    cont_table[0].append(float(Counter(failures[name])[1]))\n",
    "    cont_table[1].append(float(Counter(failures[name])[10]))\n",
    "print(cont_table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "probability=0.950, critical=15.507, stat=28405.719\n",
      "\n",
      "reject H0 - results dependent\n"
     ]
    }
   ],
   "source": [
    "from scipy import stats\n",
    "\n",
    "# test of independence\n",
    "# H0 : results are independent\n",
    "\n",
    "stat, p, dof, expected = stats.chi2_contingency(cont_table)\n",
    "stats.chi2_contingency(cont_table)\n",
    "prob = 0.95\n",
    "critical = stats.chi2.ppf(prob, dof)\n",
    "print('probability=%.3f, critical=%.3f, stat=%.3f' % (prob, critical, stat))\n",
    "\n",
    "if abs(stat) >= critical:\n",
    "    print('\\nreject H0 - results dependent')\n",
    "else:\n",
    "    print('\\nfailed to reject H0 - results independent')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## New data prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 411,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({-1: 210159, 1: 2830})\n"
     ]
    }
   ],
   "source": [
    "test_set = pd.read_csv(\"20151221.txt\", sep = \"\\t\", header = None)\n",
    "test_set.columns = [\"conn len\", \"service\", \"src bytes\", \"dst bytes\", \"conn numb\", \"Same srv rate\", \"Serror rate\", \"Srv serror rate\", \"Dst host count\", \"Dst host srv count\", \"Dst host same src port rate\", \"Dst host serror rate\", \"Dst host srv serror rate\", \"Conn state\", \"IDS detection\", \"Malware detection\", \"Ashula detection\", \"attack_flag\", \"src IP\", \"src port\", \"dst IP\", \"dst port\", \"start time\", \"proto\"]\n",
    "\n",
    "print(Counter(test_set[\"attack_flag\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 412,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Counter({1: 210159, 0: 2830})\n"
     ]
    }
   ],
   "source": [
    "test_set.loc[:, \"attack_flag\"] = test_set.loc[:,\"attack_flag\"].replace(-2, -1)\n",
    "test_set.loc[:, \"attack_flag\"] = test_set.loc[:, \"attack_flag\"] * -1\n",
    "test_set.loc[:, \"attack_flag\"] = test_set.loc[:,\"attack_flag\"].replace(-1, 0)\n",
    "\n",
    "print(Counter(test_set[\"attack_flag\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 413,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(212989, 24)\n",
      "(212989, 21)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1    210159\n",
       "0      2830\n",
       "Name: attack_flag, dtype: int64"
      ]
     },
     "execution_count": 413,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# slice dataset, drop additional features\n",
    "print(test_set.shape)\n",
    "test_set = test_set.drop(['start time', 'dst IP', 'src IP'], axis=1)\n",
    "print(test_set.shape)\n",
    "\n",
    "# drop software markers\n",
    "test_set = test_set.drop([\"IDS detection\", \"Malware detection\", \"Ashula detection\"], axis=1)\n",
    "test_set[\"attack_flag\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 414,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(212989, 18)"
      ]
     },
     "execution_count": 414,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# removing NaNs\n",
    "test_set = test_set.dropna()\n",
    "test_set.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 415,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_set[\"well-known src pool\"] = test_set.apply(wk, axis=1)\n",
    "test_set[\"registered src pool\"] = test_set.apply(reg, axis=1)\n",
    "test_set[\"cli src pool\"] = test_set.apply(cli, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 416,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_set[\"src port\"] = test_set[\"src port\"].apply(str)\n",
    "test_set[\"dst port\"] = test_set[\"dst port\"].apply(str)\n",
    "test_set['src port'].nunique()\n",
    "test_set['dst port'].nunique()\n",
    "\n",
    "thresh1 = 125\n",
    "thresh2 = 125\n",
    "counts1 = test_set['src port'].value_counts()\n",
    "counts2 = test_set['dst port'].value_counts()\n",
    "\n",
    "repl1 = counts1[counts1 <= thresh1].index\n",
    "repl2 = counts2[counts2 <= thresh2].index\n",
    "\n",
    "#hot one encoding for src port numbers\n",
    "test_set = pd.concat([test_set, pd.get_dummies(test_set[\"src port\"].replace(repl1, 'other_src'), prefix=\"is_src\")], axis=1)\n",
    "\n",
    "#hot one encoding for dst port numbers\n",
    "test_set = pd.concat([test_set, pd.get_dummies(test_set[\"dst port\"].replace(repl2, 'other_dst'), prefix=\"is_dst\")], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 417,
   "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>conn len</th>\n",
       "      <th>service</th>\n",
       "      <th>src bytes</th>\n",
       "      <th>dst bytes</th>\n",
       "      <th>conn numb</th>\n",
       "      <th>Same srv rate</th>\n",
       "      <th>Serror rate</th>\n",
       "      <th>Srv serror rate</th>\n",
       "      <th>Dst host count</th>\n",
       "      <th>Dst host srv count</th>\n",
       "      <th>...</th>\n",
       "      <th>state_OTH</th>\n",
       "      <th>state_REJ</th>\n",
       "      <th>state_RSTO</th>\n",
       "      <th>state_RSTOS0</th>\n",
       "      <th>state_RSTR</th>\n",
       "      <th>state_RSTRH</th>\n",
       "      <th>state_S0</th>\n",
       "      <th>state_S1</th>\n",
       "      <th>state_SF</th>\n",
       "      <th>state_SHR</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>212979</th>\n",
       "      <td>0.000061</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.05</td>\n",
       "      <td>0.15</td>\n",
       "      <td>2</td>\n",
       "      <td>85</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212980</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.22</td>\n",
       "      <td>4</td>\n",
       "      <td>6</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212981</th>\n",
       "      <td>0.000013</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>13</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.08</td>\n",
       "      <td>0.26</td>\n",
       "      <td>6</td>\n",
       "      <td>84</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212982</th>\n",
       "      <td>0.000029</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0.25</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212983</th>\n",
       "      <td>0.000039</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>14</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.07</td>\n",
       "      <td>0.24</td>\n",
       "      <td>1</td>\n",
       "      <td>84</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212984</th>\n",
       "      <td>0.000045</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>15</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.07</td>\n",
       "      <td>0.23</td>\n",
       "      <td>3</td>\n",
       "      <td>84</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212985</th>\n",
       "      <td>0.000053</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>16</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.06</td>\n",
       "      <td>0.22</td>\n",
       "      <td>19</td>\n",
       "      <td>84</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212986</th>\n",
       "      <td>0.000056</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.06</td>\n",
       "      <td>0.21</td>\n",
       "      <td>7</td>\n",
       "      <td>84</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212987</th>\n",
       "      <td>0.000058</td>\n",
       "      <td>other</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.06</td>\n",
       "      <td>0.20</td>\n",
       "      <td>20</td>\n",
       "      <td>84</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212988</th>\n",
       "      <td>0.000110</td>\n",
       "      <td>dns</td>\n",
       "      <td>36</td>\n",
       "      <td>194</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0</td>\n",
       "      <td>89</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 193 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        conn len service  src bytes  dst bytes  conn numb  Same srv rate  \\\n",
       "212979  0.000061   other          0          0         20            1.0   \n",
       "212980  0.000000   other          0          0          1            1.0   \n",
       "212981  0.000013   other          0          0         13            1.0   \n",
       "212982  0.000029   other          0          0          2            1.0   \n",
       "212983  0.000039   other          0          0         14            1.0   \n",
       "212984  0.000045   other          0          0         15            1.0   \n",
       "212985  0.000053   other          0          0         16            1.0   \n",
       "212986  0.000056   other          0          0         17            1.0   \n",
       "212987  0.000058   other          0          0         18            1.0   \n",
       "212988  0.000110     dns         36        194          1            1.0   \n",
       "\n",
       "        Serror rate  Srv serror rate  Dst host count  Dst host srv count  \\\n",
       "212979         0.05             0.15               2                  85   \n",
       "212980         0.00             0.22               4                   6   \n",
       "212981         0.08             0.26               6                  84   \n",
       "212982         0.50             0.25               1                   6   \n",
       "212983         0.07             0.24               1                  84   \n",
       "212984         0.07             0.23               3                  84   \n",
       "212985         0.06             0.22              19                  84   \n",
       "212986         0.06             0.21               7                  84   \n",
       "212987         0.06             0.20              20                  84   \n",
       "212988         0.00             0.50               0                  89   \n",
       "\n",
       "          ...      state_OTH  state_REJ  state_RSTO state_RSTOS0  state_RSTR  \\\n",
       "212979    ...              0          1           0            0           0   \n",
       "212980    ...              0          0           0            0           0   \n",
       "212981    ...              0          1           0            0           0   \n",
       "212982    ...              0          1           0            0           0   \n",
       "212983    ...              0          1           0            0           0   \n",
       "212984    ...              0          1           0            0           0   \n",
       "212985    ...              0          1           0            0           0   \n",
       "212986    ...              0          1           0            0           0   \n",
       "212987    ...              0          1           0            0           0   \n",
       "212988    ...              0          0           0            0           0   \n",
       "\n",
       "       state_RSTRH state_S0 state_S1  state_SF  state_SHR  \n",
       "212979           0        0        0         0          0  \n",
       "212980           0        1        0         0          0  \n",
       "212981           0        0        0         0          0  \n",
       "212982           0        0        0         0          0  \n",
       "212983           0        0        0         0          0  \n",
       "212984           0        0        0         0          0  \n",
       "212985           0        0        0         0          0  \n",
       "212986           0        0        0         0          0  \n",
       "212987           0        0        0         0          0  \n",
       "212988           0        0        0         1          0  \n",
       "\n",
       "[10 rows x 193 columns]"
      ]
     },
     "execution_count": 417,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### hot one encoding ###\n",
    "\n",
    "# hot one encoding for service\n",
    "test_set = pd.concat([test_set, pd.get_dummies(test_set[\"service\"], prefix=\"is\")], axis=1)\n",
    "test_set.tail(10)\n",
    "\n",
    "#hot one encoding for proto - commented due to reduced column number\n",
    "test_set = pd.concat([test_set, pd.get_dummies(test_set[\"proto\"], prefix=\"is\")], axis=1)\n",
    "test_set.tail(10)\n",
    "\n",
    "# hot one encoding for conn state\n",
    "test_set = pd.concat([test_set, pd.get_dummies(test_set[\"Conn state\"], prefix=\"state\")], axis=1)\n",
    "test_set.tail(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 418,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_set.rename(columns={'Dst host count': 'Dst_host_count', \"Dst host srv count\" : \"Dst_host_srv_count\",\n",
    "                    \"Dst host same src port rate\" : \"Dst_host_same_src_port_rate\",\n",
    "                    \"Dst host serror rate\" : \"Dst_host_serror_rate\",\n",
    "                    \"Dst host srv serror rate\" : \"Dst_host_srv_serror_rate\",\n",
    "                    \"Same srv rate\" : \"Same_srv_rate\",\n",
    "                    \"Srv serror rate\" : \"Srv_serror_rate\",\n",
    "                    \"conn len\" : \"conn_len\",\n",
    "                    \"src bytes\" : \"src_bytes\",\n",
    "                    \"dst bytes\" : \"dst_bytes\",\n",
    "                    \"conn numb\" : \"conn_numb\",\n",
    "                    \"Serror rate\" : \"Serror_rate\",\n",
    "                    \"cli src pool\" : \"cli_src_pool\",\n",
    "                    \"registered src pool\" : \"registered_src_pool\",\n",
    "                    \"well-known src pool\" : \"well_known_src_pool\",\n",
    "                    \"dst port\" : \"dst_port\",\n",
    "                    \"src port\" : \"src_port\",\n",
    "                     \"Conn state\" : \"Conn_state\" \n",
    "                    }, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 419,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['is_dst_102', 'is_dst_10200', 'is_dst_119', 'is_dst_143', 'is_dst_223',\n",
      "       'is_dst_389', 'is_dst_5432', 'is_dst_6050', 'is_dst_631', 'is_dst_82',\n",
      "       'is_dst_88', 'is_dst_9080', 'is_src_13941', 'is_src_18382',\n",
      "       'is_src_21320', 'is_src_22055', 'is_src_22207', 'is_src_2295',\n",
      "       'is_src_29174', 'is_src_300', 'is_src_37286', 'is_src_37396',\n",
      "       'is_src_37511', 'is_src_38396', 'is_src_40082', 'is_src_40718',\n",
      "       'is_src_41161', 'is_src_41287', 'is_src_41708', 'is_src_41793',\n",
      "       'is_src_42206', 'is_src_443', 'is_src_4445', 'is_src_46432',\n",
      "       'is_src_49354', 'is_src_49362', 'is_src_49376', 'is_src_49414',\n",
      "       'is_src_5061', 'is_src_5071', 'is_src_5094', 'is_src_5100',\n",
      "       'is_src_5114', 'is_src_51361', 'is_src_5208', 'is_src_5211',\n",
      "       'is_src_52131', 'is_src_5229', 'is_src_5247', 'is_src_53265',\n",
      "       'is_src_53451', 'is_src_53762', 'is_src_54110', 'is_src_54334',\n",
      "       'is_src_54499', 'is_src_54509', 'is_src_5501', 'is_src_56118',\n",
      "       'is_src_5844', 'is_src_58663', 'is_src_60000', 'is_src_6044',\n",
      "       'is_src_61605', 'is_src_6173', 'is_src_62029', 'is_src_63265',\n",
      "       'is_src_63432', 'is_src_64486'],\n",
      "      dtype='object')\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(212989, 193)"
      ]
     },
     "execution_count": 419,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check differences between datasets after hot one encoding\n",
    "\n",
    "# print(train_data.columns.difference(test_data.columns)[0])\n",
    "# columns that are not in test but are in train\n",
    "test_new_data_full = test_set\n",
    "\n",
    "# for c in data.columns.difference(test_set.columns):\n",
    "#     test_set[c] = 0\n",
    "# for c in test_set.columns.difference(data.columns):\n",
    "#     test_set = test_set.drop(c, axis=1)\n",
    "    \n",
    "print(test_set.columns.difference(data.columns))  \n",
    "test_set.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 420,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_new_data = test_set.loc[range(30000), :]\n",
    "\n",
    "for c in data.columns.difference(test_new_data.columns):\n",
    "    test_new_data[c] = 0\n",
    "for c in test_new_data.columns.difference(data.columns):\n",
    "    test_new_data = test_new_data.drop(c, axis=1)\n",
    "\n",
    "test_new_data = test_new_data.drop([\"service\", \"Conn_state\", \"proto\", \"src_port\", \"dst_port\"], axis=1)\n",
    "    \n",
    "test_new_data = test_new_data.sample(frac=1)\n",
    "y_test_new_data = test_new_data.loc[:, \"attack_flag\"].values.ravel()\n",
    "X_test_new_data = test_new_data.drop([\"attack_flag\"], axis=1).astype(np.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 421,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(30000, 286)\n",
      "(62872, 286)\n",
      "1    29498\n",
      "0      502\n",
      "Name: attack_flag, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "print(X_test_new_data.shape)\n",
    "print(X_test.shape)\n",
    "print(test_new_data[\"attack_flag\"].value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 422,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: baseline     Accuracy: 0.98327, F1-score: 0.99156, Precision: 0.98327, Recall: 1.00000, FPR: 1.00000, FNR: 0.00000\n",
      "TN:   0; FP: 502; FN:    0; TP: 29498\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Metrics of baseline method\n",
    "from sklearn import metrics\n",
    "\n",
    "name = \"baseline\"\n",
    "\n",
    "if(y_test_new_data.mean() > (1-y_test_new_data.mean())):\n",
    "    y_pred = np.ones(len(y_test_new_data)).ravel() \n",
    "else:\n",
    "    y_pred = np.zeros(len(y_test_new_data)).ravel() \n",
    "    \n",
    "Acc[name] = metrics.accuracy_score(y_test_new_data, y_pred)\n",
    "F1S[name] = metrics.f1_score(y_test_new_data, y_pred)\n",
    "Prec[name] = metrics.precision_score(y_test_new_data, y_pred)\n",
    "Rec[name] = metrics.recall_score(y_test_new_data, y_pred)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(y_test_new_data, y_pred).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 423,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: LogisticReg  Accuracy: 0.97027, F1-score: 0.98487, Precision: 0.98557, Recall: 0.98417, FPR: 0.84661, FNR: 0.01583\n",
      "TN:  77; FP: 425; FN:  467; TP: 29031\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn import metrics\n",
    "\n",
    "clf = LogisticRegression(solver = 'newton-cg')\n",
    "\n",
    "model = clf.fit(X_train, y_train)\n",
    "\n",
    "y_pred = model.predict(X_test_new_data)\n",
    "\n",
    "name = \"LogisticReg\"\n",
    "\n",
    "Acc[name] = metrics.accuracy_score(y_test_new_data, y_pred)\n",
    "F1S[name] = metrics.f1_score(y_test_new_data, y_pred)\n",
    "Prec[name] = metrics.precision_score(y_test_new_data, y_pred)\n",
    "Rec[name] = metrics.recall_score(y_test_new_data, y_pred)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(y_test_new_data, y_pred).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 424,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: AdaBoost     Accuracy: 0.98800, F1-score: 0.99392, Precision: 0.98955, Recall: 0.99834, FPR: 0.61952, FNR: 0.00166\n",
      "TN: 191; FP: 311; FN:   49; TP: 29449\n",
      "\n",
      "Algorithm: DecisionTree Accuracy: 0.97853, F1-score: 0.98915, Precision: 0.98322, Recall: 0.99515, FPR: 0.99801, FNR: 0.00485\n",
      "TN:   1; FP: 501; FN:  143; TP: 29355\n",
      "\n",
      "Algorithm: NaiveBayes   Accuracy: 0.98327, F1-score: 0.99156, Precision: 0.98327, Recall: 1.00000, FPR: 1.00000, FNR: 0.00000\n",
      "TN:   0; FP: 502; FN:    0; TP: 29498\n",
      "\n",
      "Algorithm: RandomTree   Accuracy: 0.97023, F1-score: 0.98486, Precision: 0.98534, Recall: 0.98437, FPR: 0.86056, FNR: 0.01563\n",
      "TN:  70; FP: 432; FN:  461; TP: 29037\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/preprocessing/data.py:625: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  return self.partial_fit(X, y)\n",
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/base.py:465: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  return self.fit(X, y, **fit_params).transform(X)\n",
      "/home/matz/.local/lib/python3.6/site-packages/sklearn/pipeline.py:331: DataConversionWarning: Data with input dtype float32 were all converted to float64 by StandardScaler.\n",
      "  Xt = transform.transform(Xt)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: RandomForest Accuracy: 0.98330, F1-score: 0.99158, Precision: 0.98330, Recall: 1.00000, FPR: 0.99801, FNR: 0.00000\n",
      "TN:   1; FP: 501; FN:    0; TP: 29498\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "seed = 12\n",
    "num_trees = 28\n",
    "\n",
    "classifiers = {\n",
    "               \"AdaBoost\":AdaBoostClassifier(DecisionTreeClassifier(max_depth = 13), n_estimators=num_trees, random_state = 0),\n",
    "               \"DecisionTree\":DecisionTreeClassifier(random_state=0, max_depth = 13),\n",
    "               \"NaiveBayes\":GaussianNB(),\n",
    "               \"RandomTree\":BaggingClassifier(DecisionTreeClassifier(random_state=0, criterion = \"entropy\", max_depth = 13)),\n",
    "               \"RandomForest\":make_pipeline(StandardScaler(),RandomForestClassifier(n_estimators=167, max_depth=38, random_state=0)),\n",
    "              }\n",
    "\n",
    "\n",
    "for name, clf in classifiers.items():\n",
    "    \n",
    "    model = clf.fit(X_train, y_train)\n",
    "    \n",
    "    y_pred = model.predict(X_test_new_data)\n",
    "    \n",
    "    Acc[name] = metrics.accuracy_score(y_test_new_data, y_pred)\n",
    "    F1S[name] = metrics.f1_score(y_test_new_data, y_pred)\n",
    "    Prec[name] = metrics.precision_score(y_test_new_data, y_pred)\n",
    "    Rec[name] = metrics.recall_score(y_test_new_data, y_pred)\n",
    "    tn, fp, fn, tp = metrics.confusion_matrix(y_test_new_data, y_pred).ravel()\n",
    "    FPR[name] = fp/(fp+tn)\n",
    "    FNR[name] = fn/(tp+fn)\n",
    "\n",
    "    print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "    print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 425,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_input_fn = tf.estimator.inputs.numpy_input_fn(\n",
    "    x={\"features\": X_test_new_data.values},\n",
    "    y=y_test_new_data,\n",
    "    batch_size=X_test_new_data.shape[0],\n",
    "    num_epochs=1,\n",
    "    shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 426,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from kernel_train/model.ckpt-10000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n"
     ]
    }
   ],
   "source": [
    "pred_iterator = estimator_kernel.predict(input_fn = test_input_fn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 427,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_id = []\n",
    "\n",
    "for pred_dict in pred_iterator:\n",
    "    class_id.append(pred_dict['classes'])\n",
    "    #print(pred_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 428,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: RFFM         Accuracy: 0.98457, F1-score: 0.99215, Precision: 0.99220, Recall: 0.99210, FPR: 0.45817, FNR: 0.00790\n",
      "TN: 272; FP: 230; FN:  233; TP: 29265\n",
      "\n"
     ]
    }
   ],
   "source": [
    "name = \"RFFM\"\n",
    "\n",
    "Acc[name] = metrics.accuracy_score(y_test_new_data, class_id)\n",
    "F1S[name] = metrics.f1_score(y_test_new_data, class_id)\n",
    "Prec[name] = metrics.precision_score(y_test_new_data, class_id)\n",
    "Rec[name] = metrics.recall_score(y_test_new_data, class_id)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(y_test_new_data, class_id).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 429,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n",
      "INFO:tensorflow:Done calling model_fn.\n",
      "INFO:tensorflow:Graph was finalized.\n",
      "INFO:tensorflow:Restoring parameters from /home/matz/Desktop/ml_ids/NN_model2/model.ckpt-20000\n",
      "INFO:tensorflow:Running local_init_op.\n",
      "INFO:tensorflow:Done running local_init_op.\n"
     ]
    }
   ],
   "source": [
    "results = []\n",
    "results = list(NN_model.predict(input_fn=get_input_fn(test_new_data_full.loc[range(30000), :], num_epochs=1, n_batch = X_test.shape[0], shuffle=False)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 430,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30000"
      ]
     },
     "execution_count": 430,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = []\n",
    "for el in results:\n",
    "    y_pred.append(el[\"class_ids\"])\n",
    "len(y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 431,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Algorithm: Wide&Deep    Accuracy: 0.97497, F1-score: 0.98732, Precision: 0.98332, Recall: 0.99136, FPR: 0.98805, FNR: 0.00864\n",
      "TN:   6; FP: 496; FN:  255; TP: 29243\n",
      "\n"
     ]
    }
   ],
   "source": [
    "name = \"Wide&Deep\"\n",
    "\n",
    "Acc[name] = metrics.accuracy_score(y_test_new_data, y_pred)\n",
    "F1S[name] = metrics.f1_score(y_test_new_data, y_pred)\n",
    "Prec[name] = metrics.precision_score(y_test_new_data, y_pred)\n",
    "Rec[name] = metrics.recall_score(y_test_new_data, y_pred)\n",
    "tn, fp, fn, tp = metrics.confusion_matrix(y_test_new_data, y_pred).ravel()\n",
    "FPR[name] = fp/(fp+tn)\n",
    "FNR[name] = fn/(tp+fn)\n",
    "\n",
    "print(\"Algorithm: {0:12} Accuracy: {1:.5f}, F1-score: {2:.5f}, Precision: {3:.5f}, Recall: {4:.5f}, FPR: {5:.5f}, FNR: {6:.5f}\".format(name, Acc[name], F1S[name], Prec[name], Rec[name], FPR[name], FNR[name]))\n",
    "print(\"TN: {0:3}; FP: {1:3}; FN: {2:4}; TP: {3:3}\\n\".format(tn, fp, fn, tp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 432,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11f8807898>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11f8807fd0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f11fadb9b00>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1276978ac8>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1276832a58>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f12767ec8d0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_stats(Acc, \"Accuracy\", 8, 55)\n",
    "plot_stats(F1S, \"F1-score\", 8, 55)\n",
    "plot_stats(Prec, \"Precision\", 8, 55)\n",
    "plot_stats(Rec, \"Recall\", 8, 55)\n",
    "plot_stats(FPR, \"False Positive Rate\", 8, 55)\n",
    "plot_stats(FNR, \"False Negative Rate\", 8, 55)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}