inzva_week2_anomaly_detection_plus_lightgbm.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "x275tSx2asW0"
   },
   "source": [
    "# Anomaly Detection - KDD'99 Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "k2oEv8NNPcNS"
   },
   "source": [
    "In week 2 of the inzva Applied AI program we will spend most of our time on Anomaly Detection techniques. \n",
    "\n",
    "**This notebook includes**\n",
    "\n",
    "*1- Isolation Forest*\n",
    "\n",
    "*2- Local Outlier Factor*\n",
    "\n",
    "*3- Autoencoder*\n",
    "\n",
    "*4 - XGBoost*\n",
    "\n",
    "*5 - CatBoost and Hyperparameter Optimization*\n",
    "\n",
    "*6 - TabNet*\n",
    "\n",
    "We first start with obtaining the data and creating training and test sets. In this notebook, we create our models only for Probing attack detection.\n",
    "\n",
    "In this notebook we only work with numerical features. We will drop categorical features in our dataset. You will use the full dataset in your homework in order to compare XGBoost and CatBoost algorithms.\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "sDK-7yCKbWL8"
   },
   "source": [
    "## 1. Data & Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "ZCSpFnNe1UlI"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "kdd = pd.read_csv('/Users/onurboyar/Desktop/JupyterNotebooks/Datasets/kdd.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "EI2zzt-K_2mL"
   },
   "outputs": [],
   "source": [
    "kdd = kdd.iloc[:,1:43]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "m-tUOxq4_4xU"
   },
   "outputs": [],
   "source": [
    "kdd = kdd.drop(['Protocol Type', 'Service', 'Flag'], axis = 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "uxURDlpWQcTn",
    "outputId": "60b3ced6-0062-48d4-a487-82c2f2a385e7"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(183737, 39)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Duration</th>\n",
       "      <th>src_bytes</th>\n",
       "      <th>dst_bytes</th>\n",
       "      <th>land</th>\n",
       "      <th>wrong_fragment</th>\n",
       "      <th>urgent</th>\n",
       "      <th>hot</th>\n",
       "      <th>num_failed</th>\n",
       "      <th>logged_in</th>\n",
       "      <th>num_compromised</th>\n",
       "      <th>...</th>\n",
       "      <th>dst_host_count</th>\n",
       "      <th>dst_host_srv_count</th>\n",
       "      <th>dst_host_same_srv_rate</th>\n",
       "      <th>dst_host_diff_srv_rate</th>\n",
       "      <th>dst_host_same_src_port_rate</th>\n",
       "      <th>dst_host_srv_diff_host_rate</th>\n",
       "      <th>dst_host_serror_rate</th>\n",
       "      <th>dst_host_srv_serror_rate</th>\n",
       "      <th>dst_host_rerror_rate</th>\n",
       "      <th>dst_host_srv_rerror_rate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>183737.000000</td>\n",
       "      <td>183737.000000</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>...</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "      <td>1.837370e+05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>-3.640197e-14</td>\n",
       "      <td>-1.056812e-15</td>\n",
       "      <td>9.753362e-16</td>\n",
       "      <td>0.000005</td>\n",
       "      <td>0.000697</td>\n",
       "      <td>5.397149e-16</td>\n",
       "      <td>-2.331502e-14</td>\n",
       "      <td>-9.633879e-15</td>\n",
       "      <td>1.194191e-13</td>\n",
       "      <td>-1.428721e-15</td>\n",
       "      <td>...</td>\n",
       "      <td>4.349572e-15</td>\n",
       "      <td>3.331752e-14</td>\n",
       "      <td>1.545878e-13</td>\n",
       "      <td>-7.998783e-14</td>\n",
       "      <td>1.042442e-13</td>\n",
       "      <td>-1.635035e-16</td>\n",
       "      <td>1.020946e-14</td>\n",
       "      <td>2.471166e-14</td>\n",
       "      <td>1.029228e-13</td>\n",
       "      <td>-9.611212e-16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>0.002333</td>\n",
       "      <td>0.043640</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>...</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "      <td>9.999973e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-1.501030e-01</td>\n",
       "      <td>-1.450685e-02</td>\n",
       "      <td>-8.699338e-02</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-1.025347e-02</td>\n",
       "      <td>-7.347231e-02</td>\n",
       "      <td>-9.282620e-02</td>\n",
       "      <td>-1.500984e+00</td>\n",
       "      <td>-7.873908e-03</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.803474e+00</td>\n",
       "      <td>-2.793559e+00</td>\n",
       "      <td>-3.475423e+00</td>\n",
       "      <td>-3.464397e-01</td>\n",
       "      <td>-5.074414e-01</td>\n",
       "      <td>-3.659654e-01</td>\n",
       "      <td>-1.069506e-01</td>\n",
       "      <td>-7.539644e-02</td>\n",
       "      <td>-3.017272e-01</td>\n",
       "      <td>-2.967496e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>-1.501030e-01</td>\n",
       "      <td>-1.364598e-02</td>\n",
       "      <td>-7.548494e-02</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-1.025347e-02</td>\n",
       "      <td>-7.347231e-02</td>\n",
       "      <td>-9.282620e-02</td>\n",
       "      <td>-1.173472e+00</td>\n",
       "      <td>-7.873908e-03</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.077665e+00</td>\n",
       "      <td>-7.749125e-02</td>\n",
       "      <td>3.393276e-01</td>\n",
       "      <td>-3.464397e-01</td>\n",
       "      <td>-4.753133e-01</td>\n",
       "      <td>-3.659654e-01</td>\n",
       "      <td>-1.069506e-01</td>\n",
       "      <td>-7.539644e-02</td>\n",
       "      <td>-3.017272e-01</td>\n",
       "      <td>-2.967496e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>-1.494535e-01</td>\n",
       "      <td>-4.984003e-03</td>\n",
       "      <td>-5.291073e-02</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-5.643673e-03</td>\n",
       "      <td>-7.347231e-02</td>\n",
       "      <td>-2.147289e-02</td>\n",
       "      <td>6.662233e-01</td>\n",
       "      <td>-7.208470e-03</td>\n",
       "      <td>...</td>\n",
       "      <td>7.508105e-01</td>\n",
       "      <td>4.701381e-01</td>\n",
       "      <td>3.778605e-01</td>\n",
       "      <td>-2.485595e-01</td>\n",
       "      <td>-3.463717e-01</td>\n",
       "      <td>-2.565694e-01</td>\n",
       "      <td>-1.069506e-01</td>\n",
       "      <td>-7.252428e-02</td>\n",
       "      <td>-2.235746e-01</td>\n",
       "      <td>-2.577025e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>-2.549583e-02</td>\n",
       "      <td>-4.888098e-03</td>\n",
       "      <td>-3.004341e-02</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-5.643673e-03</td>\n",
       "      <td>-5.274024e-02</td>\n",
       "      <td>-2.147289e-02</td>\n",
       "      <td>8.521615e-01</td>\n",
       "      <td>-7.208470e-03</td>\n",
       "      <td>...</td>\n",
       "      <td>7.508105e-01</td>\n",
       "      <td>6.532273e-01</td>\n",
       "      <td>5.365422e-01</td>\n",
       "      <td>-1.918766e-01</td>\n",
       "      <td>-2.504167e-01</td>\n",
       "      <td>4.609461e-02</td>\n",
       "      <td>-9.664964e-02</td>\n",
       "      <td>-7.252428e-02</td>\n",
       "      <td>-2.146693e-01</td>\n",
       "      <td>-2.047013e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.134389e+02</td>\n",
       "      <td>3.196954e+02</td>\n",
       "      <td>1.653286e+02</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>2.480652e+02</td>\n",
       "      <td>1.782544e+02</td>\n",
       "      <td>1.468001e+02</td>\n",
       "      <td>8.521615e-01</td>\n",
       "      <td>2.241221e+02</td>\n",
       "      <td>...</td>\n",
       "      <td>1.019572e+00</td>\n",
       "      <td>6.532273e-01</td>\n",
       "      <td>5.365422e-01</td>\n",
       "      <td>5.419733e+00</td>\n",
       "      <td>4.120075e+00</td>\n",
       "      <td>1.487663e+01</td>\n",
       "      <td>2.105394e+01</td>\n",
       "      <td>1.764532e+01</td>\n",
       "      <td>5.592885e+00</td>\n",
       "      <td>5.161322e+00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 38 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Duration     src_bytes     dst_bytes           land  \\\n",
       "count  1.837370e+05  1.837370e+05  1.837370e+05  183737.000000   \n",
       "mean  -3.640197e-14 -1.056812e-15  9.753362e-16       0.000005   \n",
       "std    9.999973e-01  9.999973e-01  9.999973e-01       0.002333   \n",
       "min   -1.501030e-01 -1.450685e-02 -8.699338e-02       0.000000   \n",
       "25%   -1.501030e-01 -1.364598e-02 -7.548494e-02       0.000000   \n",
       "50%   -1.494535e-01 -4.984003e-03 -5.291073e-02       0.000000   \n",
       "75%   -2.549583e-02 -4.888098e-03 -3.004341e-02       0.000000   \n",
       "max    1.134389e+02  3.196954e+02  1.653286e+02       1.000000   \n",
       "\n",
       "       wrong_fragment        urgent           hot    num_failed     logged_in  \\\n",
       "count   183737.000000  1.837370e+05  1.837370e+05  1.837370e+05  1.837370e+05   \n",
       "mean         0.000697  5.397149e-16 -2.331502e-14 -9.633879e-15  1.194191e-13   \n",
       "std          0.043640  9.999973e-01  9.999973e-01  9.999973e-01  9.999973e-01   \n",
       "min          0.000000 -1.025347e-02 -7.347231e-02 -9.282620e-02 -1.500984e+00   \n",
       "25%          0.000000 -1.025347e-02 -7.347231e-02 -9.282620e-02 -1.173472e+00   \n",
       "50%          0.000000 -5.643673e-03 -7.347231e-02 -2.147289e-02  6.662233e-01   \n",
       "75%          0.000000 -5.643673e-03 -5.274024e-02 -2.147289e-02  8.521615e-01   \n",
       "max          3.000000  2.480652e+02  1.782544e+02  1.468001e+02  8.521615e-01   \n",
       "\n",
       "       num_compromised  ...  dst_host_count  dst_host_srv_count  \\\n",
       "count     1.837370e+05  ...    1.837370e+05        1.837370e+05   \n",
       "mean     -1.428721e-15  ...    4.349572e-15        3.331752e-14   \n",
       "std       9.999973e-01  ...    9.999973e-01        9.999973e-01   \n",
       "min      -7.873908e-03  ...   -1.803474e+00       -2.793559e+00   \n",
       "25%      -7.873908e-03  ...   -1.077665e+00       -7.749125e-02   \n",
       "50%      -7.208470e-03  ...    7.508105e-01        4.701381e-01   \n",
       "75%      -7.208470e-03  ...    7.508105e-01        6.532273e-01   \n",
       "max       2.241221e+02  ...    1.019572e+00        6.532273e-01   \n",
       "\n",
       "       dst_host_same_srv_rate  dst_host_diff_srv_rate  \\\n",
       "count            1.837370e+05            1.837370e+05   \n",
       "mean             1.545878e-13           -7.998783e-14   \n",
       "std              9.999973e-01            9.999973e-01   \n",
       "min             -3.475423e+00           -3.464397e-01   \n",
       "25%              3.393276e-01           -3.464397e-01   \n",
       "50%              3.778605e-01           -2.485595e-01   \n",
       "75%              5.365422e-01           -1.918766e-01   \n",
       "max              5.365422e-01            5.419733e+00   \n",
       "\n",
       "       dst_host_same_src_port_rate  dst_host_srv_diff_host_rate  \\\n",
       "count                 1.837370e+05                 1.837370e+05   \n",
       "mean                  1.042442e-13                -1.635035e-16   \n",
       "std                   9.999973e-01                 9.999973e-01   \n",
       "min                  -5.074414e-01                -3.659654e-01   \n",
       "25%                  -4.753133e-01                -3.659654e-01   \n",
       "50%                  -3.463717e-01                -2.565694e-01   \n",
       "75%                  -2.504167e-01                 4.609461e-02   \n",
       "max                   4.120075e+00                 1.487663e+01   \n",
       "\n",
       "       dst_host_serror_rate  dst_host_srv_serror_rate  dst_host_rerror_rate  \\\n",
       "count          1.837370e+05              1.837370e+05          1.837370e+05   \n",
       "mean           1.020946e-14              2.471166e-14          1.029228e-13   \n",
       "std            9.999973e-01              9.999973e-01          9.999973e-01   \n",
       "min           -1.069506e-01             -7.539644e-02         -3.017272e-01   \n",
       "25%           -1.069506e-01             -7.539644e-02         -3.017272e-01   \n",
       "50%           -1.069506e-01             -7.252428e-02         -2.235746e-01   \n",
       "75%           -9.664964e-02             -7.252428e-02         -2.146693e-01   \n",
       "max            2.105394e+01              1.764532e+01          5.592885e+00   \n",
       "\n",
       "       dst_host_srv_rerror_rate  \n",
       "count              1.837370e+05  \n",
       "mean              -9.611212e-16  \n",
       "std                9.999973e-01  \n",
       "min               -2.967496e-01  \n",
       "25%               -2.967496e-01  \n",
       "50%               -2.577025e-01  \n",
       "75%               -2.047013e-01  \n",
       "max                5.161322e+00  \n",
       "\n",
       "[8 rows x 38 columns]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "kdd['src_bytes'].hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "kdd.boxplot(column = ['src_bytes'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Duration                       0\n",
       "src_bytes                      0\n",
       "dst_bytes                      0\n",
       "land                           0\n",
       "wrong_fragment                 0\n",
       "urgent                         0\n",
       "hot                            0\n",
       "num_failed                     0\n",
       "logged_in                      0\n",
       "num_compromised                0\n",
       "root_shell                     0\n",
       "su_attempted                   0\n",
       "num_root                       0\n",
       "num_file_creations             0\n",
       "num_shells                     0\n",
       "num_access_files               0\n",
       "num_outbound_cmds              0\n",
       "is_host_login                  0\n",
       "is_guest_login                 0\n",
       "count                          0\n",
       "srv_count                      0\n",
       "serror_rate                    0\n",
       "srv_error_rate                 0\n",
       "rerror_rate                    0\n",
       "srv_rerror_rate                0\n",
       "same_srv_rate                  0\n",
       "diff_srv_rate                  0\n",
       "srv_diff_host_rate             0\n",
       "dst_host_count                 0\n",
       "dst_host_srv_count             0\n",
       "dst_host_same_srv_rate         0\n",
       "dst_host_diff_srv_rate         0\n",
       "dst_host_same_src_port_rate    0\n",
       "dst_host_srv_diff_host_rate    0\n",
       "dst_host_serror_rate           0\n",
       "dst_host_srv_serror_rate       0\n",
       "dst_host_rerror_rate           0\n",
       "dst_host_srv_rerror_rate       0\n",
       "Type_Groups                    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 183737 entries, 0 to 183736\n",
      "Data columns (total 39 columns):\n",
      " #   Column                       Non-Null Count   Dtype  \n",
      "---  ------                       --------------   -----  \n",
      " 0   Duration                     183737 non-null  float64\n",
      " 1   src_bytes                    183737 non-null  float64\n",
      " 2   dst_bytes                    183737 non-null  float64\n",
      " 3   land                         183737 non-null  int64  \n",
      " 4   wrong_fragment               183737 non-null  int64  \n",
      " 5   urgent                       183737 non-null  float64\n",
      " 6   hot                          183737 non-null  float64\n",
      " 7   num_failed                   183737 non-null  float64\n",
      " 8   logged_in                    183737 non-null  float64\n",
      " 9   num_compromised              183737 non-null  float64\n",
      " 10  root_shell                   183737 non-null  float64\n",
      " 11  su_attempted                 183737 non-null  float64\n",
      " 12  num_root                     183737 non-null  float64\n",
      " 13  num_file_creations           183737 non-null  float64\n",
      " 14  num_shells                   183737 non-null  float64\n",
      " 15  num_access_files             183737 non-null  float64\n",
      " 16  num_outbound_cmds            183737 non-null  int64  \n",
      " 17  is_host_login                183737 non-null  int64  \n",
      " 18  is_guest_login               183737 non-null  float64\n",
      " 19  count                        183737 non-null  float64\n",
      " 20  srv_count                    183737 non-null  float64\n",
      " 21  serror_rate                  183737 non-null  float64\n",
      " 22  srv_error_rate               183737 non-null  float64\n",
      " 23  rerror_rate                  183737 non-null  float64\n",
      " 24  srv_rerror_rate              183737 non-null  float64\n",
      " 25  same_srv_rate                183737 non-null  float64\n",
      " 26  diff_srv_rate                183737 non-null  float64\n",
      " 27  srv_diff_host_rate           183737 non-null  float64\n",
      " 28  dst_host_count               183737 non-null  float64\n",
      " 29  dst_host_srv_count           183737 non-null  float64\n",
      " 30  dst_host_same_srv_rate       183737 non-null  float64\n",
      " 31  dst_host_diff_srv_rate       183737 non-null  float64\n",
      " 32  dst_host_same_src_port_rate  183737 non-null  float64\n",
      " 33  dst_host_srv_diff_host_rate  183737 non-null  float64\n",
      " 34  dst_host_serror_rate         183737 non-null  float64\n",
      " 35  dst_host_srv_serror_rate     183737 non-null  float64\n",
      " 36  dst_host_rerror_rate         183737 non-null  float64\n",
      " 37  dst_host_srv_rerror_rate     183737 non-null  float64\n",
      " 38  Type_Groups                  183737 non-null  object \n",
      "dtypes: float64(34), int64(4), object(1)\n",
      "memory usage: 54.7+ MB\n"
     ]
    }
   ],
   "source": [
    "kdd.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd.duplicated().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "kdd.drop_duplicates(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Duration</th>\n",
       "      <th>src_bytes</th>\n",
       "      <th>dst_bytes</th>\n",
       "      <th>land</th>\n",
       "      <th>wrong_fragment</th>\n",
       "      <th>urgent</th>\n",
       "      <th>hot</th>\n",
       "      <th>num_failed</th>\n",
       "      <th>logged_in</th>\n",
       "      <th>num_compromised</th>\n",
       "      <th>...</th>\n",
       "      <th>dst_host_count</th>\n",
       "      <th>dst_host_srv_count</th>\n",
       "      <th>dst_host_same_srv_rate</th>\n",
       "      <th>dst_host_diff_srv_rate</th>\n",
       "      <th>dst_host_same_src_port_rate</th>\n",
       "      <th>dst_host_srv_diff_host_rate</th>\n",
       "      <th>dst_host_serror_rate</th>\n",
       "      <th>dst_host_srv_serror_rate</th>\n",
       "      <th>dst_host_rerror_rate</th>\n",
       "      <th>dst_host_srv_rerror_rate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Duration</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.004578</td>\n",
       "      <td>0.021672</td>\n",
       "      <td>-0.000351</td>\n",
       "      <td>-0.000347</td>\n",
       "      <td>0.003690</td>\n",
       "      <td>0.012719</td>\n",
       "      <td>0.001486</td>\n",
       "      <td>-0.146037</td>\n",
       "      <td>0.098938</td>\n",
       "      <td>...</td>\n",
       "      <td>0.068190</td>\n",
       "      <td>-0.189572</td>\n",
       "      <td>-0.214620</td>\n",
       "      <td>0.255215</td>\n",
       "      <td>0.204651</td>\n",
       "      <td>-0.024366</td>\n",
       "      <td>0.012989</td>\n",
       "      <td>0.006247</td>\n",
       "      <td>0.013741</td>\n",
       "      <td>0.020110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>src_bytes</th>\n",
       "      <td>0.004578</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.007928</td>\n",
       "      <td>-0.000014</td>\n",
       "      <td>-0.000233</td>\n",
       "      <td>-0.000045</td>\n",
       "      <td>0.002330</td>\n",
       "      <td>-0.000535</td>\n",
       "      <td>-0.012170</td>\n",
       "      <td>-0.000056</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000027</td>\n",
       "      <td>-0.017917</td>\n",
       "      <td>-0.015160</td>\n",
       "      <td>0.001398</td>\n",
       "      <td>0.008754</td>\n",
       "      <td>0.002703</td>\n",
       "      <td>0.007055</td>\n",
       "      <td>0.022344</td>\n",
       "      <td>-0.001297</td>\n",
       "      <td>0.000915</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_bytes</th>\n",
       "      <td>0.021672</td>\n",
       "      <td>0.007928</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.000176</td>\n",
       "      <td>-0.001709</td>\n",
       "      <td>0.011452</td>\n",
       "      <td>0.000287</td>\n",
       "      <td>0.025406</td>\n",
       "      <td>0.019351</td>\n",
       "      <td>0.026384</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.019988</td>\n",
       "      <td>-0.015013</td>\n",
       "      <td>-0.002238</td>\n",
       "      <td>-0.016368</td>\n",
       "      <td>0.007375</td>\n",
       "      <td>0.002882</td>\n",
       "      <td>0.003843</td>\n",
       "      <td>0.007759</td>\n",
       "      <td>-0.015288</td>\n",
       "      <td>-0.015691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>land</th>\n",
       "      <td>-0.000351</td>\n",
       "      <td>-0.000014</td>\n",
       "      <td>-0.000176</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.000047</td>\n",
       "      <td>-0.000019</td>\n",
       "      <td>-0.000208</td>\n",
       "      <td>-0.000062</td>\n",
       "      <td>-0.005424</td>\n",
       "      <td>-0.000022</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.003197</td>\n",
       "      <td>-0.005172</td>\n",
       "      <td>0.001413</td>\n",
       "      <td>-0.000864</td>\n",
       "      <td>0.007509</td>\n",
       "      <td>0.008351</td>\n",
       "      <td>0.055836</td>\n",
       "      <td>0.003844</td>\n",
       "      <td>-0.000759</td>\n",
       "      <td>-0.000738</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>wrong_fragment</th>\n",
       "      <td>-0.000347</td>\n",
       "      <td>-0.000233</td>\n",
       "      <td>-0.001709</td>\n",
       "      <td>-0.000047</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.000202</td>\n",
       "      <td>-0.001089</td>\n",
       "      <td>-0.001643</td>\n",
       "      <td>-0.030348</td>\n",
       "      <td>-0.000159</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.001357</td>\n",
       "      <td>-0.039489</td>\n",
       "      <td>-0.032805</td>\n",
       "      <td>0.002834</td>\n",
       "      <td>0.030775</td>\n",
       "      <td>0.007797</td>\n",
       "      <td>0.011099</td>\n",
       "      <td>-0.001459</td>\n",
       "      <td>0.006002</td>\n",
       "      <td>-0.003239</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>urgent</th>\n",
       "      <td>0.003690</td>\n",
       "      <td>-0.000045</td>\n",
       "      <td>0.011452</td>\n",
       "      <td>-0.000019</td>\n",
       "      <td>-0.000202</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.024185</td>\n",
       "      <td>0.084431</td>\n",
       "      <td>0.004443</td>\n",
       "      <td>0.039020</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.007518</td>\n",
       "      <td>-0.019645</td>\n",
       "      <td>-0.015221</td>\n",
       "      <td>0.013500</td>\n",
       "      <td>0.003135</td>\n",
       "      <td>0.002830</td>\n",
       "      <td>0.001669</td>\n",
       "      <td>-0.000639</td>\n",
       "      <td>-0.000892</td>\n",
       "      <td>0.003740</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hot</th>\n",
       "      <td>0.012719</td>\n",
       "      <td>0.002330</td>\n",
       "      <td>0.000287</td>\n",
       "      <td>-0.000208</td>\n",
       "      <td>-0.001089</td>\n",
       "      <td>0.024185</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.007537</td>\n",
       "      <td>0.035556</td>\n",
       "      <td>0.009142</td>\n",
       "      <td>...</td>\n",
       "      <td>0.023482</td>\n",
       "      <td>-0.126645</td>\n",
       "      <td>-0.134481</td>\n",
       "      <td>0.016140</td>\n",
       "      <td>-0.021403</td>\n",
       "      <td>-0.019025</td>\n",
       "      <td>0.018972</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>0.002000</td>\n",
       "      <td>-0.014772</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_failed</th>\n",
       "      <td>0.001486</td>\n",
       "      <td>-0.000535</td>\n",
       "      <td>0.025406</td>\n",
       "      <td>-0.000062</td>\n",
       "      <td>-0.001643</td>\n",
       "      <td>0.084431</td>\n",
       "      <td>0.007537</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.087611</td>\n",
       "      <td>0.004282</td>\n",
       "      <td>...</td>\n",
       "      <td>0.017104</td>\n",
       "      <td>-0.060760</td>\n",
       "      <td>-0.043562</td>\n",
       "      <td>-0.002525</td>\n",
       "      <td>-0.014387</td>\n",
       "      <td>-0.003188</td>\n",
       "      <td>0.027317</td>\n",
       "      <td>0.017950</td>\n",
       "      <td>0.032950</td>\n",
       "      <td>0.036186</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>logged_in</th>\n",
       "      <td>-0.146037</td>\n",
       "      <td>-0.012170</td>\n",
       "      <td>0.019351</td>\n",
       "      <td>-0.005424</td>\n",
       "      <td>-0.030348</td>\n",
       "      <td>0.004443</td>\n",
       "      <td>0.035556</td>\n",
       "      <td>-0.087611</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.004389</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.148581</td>\n",
       "      <td>0.462860</td>\n",
       "      <td>0.435803</td>\n",
       "      <td>-0.398185</td>\n",
       "      <td>-0.397248</td>\n",
       "      <td>-0.165710</td>\n",
       "      <td>-0.102782</td>\n",
       "      <td>-0.112004</td>\n",
       "      <td>-0.414882</td>\n",
       "      <td>-0.409466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_compromised</th>\n",
       "      <td>0.098938</td>\n",
       "      <td>-0.000056</td>\n",
       "      <td>0.026384</td>\n",
       "      <td>-0.000022</td>\n",
       "      <td>-0.000159</td>\n",
       "      <td>0.039020</td>\n",
       "      <td>0.009142</td>\n",
       "      <td>0.004282</td>\n",
       "      <td>0.004389</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.006233</td>\n",
       "      <td>-0.018696</td>\n",
       "      <td>-0.017727</td>\n",
       "      <td>0.001994</td>\n",
       "      <td>-0.001598</td>\n",
       "      <td>-0.001139</td>\n",
       "      <td>0.007625</td>\n",
       "      <td>0.010023</td>\n",
       "      <td>-0.001596</td>\n",
       "      <td>-0.001197</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>root_shell</th>\n",
       "      <td>0.027823</td>\n",
       "      <td>-0.000163</td>\n",
       "      <td>0.025422</td>\n",
       "      <td>-0.000071</td>\n",
       "      <td>-0.000555</td>\n",
       "      <td>0.172920</td>\n",
       "      <td>0.067311</td>\n",
       "      <td>0.022762</td>\n",
       "      <td>0.014683</td>\n",
       "      <td>0.261008</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.007163</td>\n",
       "      <td>-0.038762</td>\n",
       "      <td>-0.024500</td>\n",
       "      <td>0.003803</td>\n",
       "      <td>0.008618</td>\n",
       "      <td>0.012709</td>\n",
       "      <td>0.010551</td>\n",
       "      <td>0.012970</td>\n",
       "      <td>0.002892</td>\n",
       "      <td>0.001308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>su_attempted</th>\n",
       "      <td>0.088947</td>\n",
       "      <td>0.000016</td>\n",
       "      <td>0.052575</td>\n",
       "      <td>-0.000029</td>\n",
       "      <td>-0.000154</td>\n",
       "      <td>-0.000085</td>\n",
       "      <td>0.001754</td>\n",
       "      <td>0.069621</td>\n",
       "      <td>0.005313</td>\n",
       "      <td>0.736226</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.001939</td>\n",
       "      <td>-0.019681</td>\n",
       "      <td>-0.021385</td>\n",
       "      <td>0.000534</td>\n",
       "      <td>-0.003931</td>\n",
       "      <td>-0.001855</td>\n",
       "      <td>0.018427</td>\n",
       "      <td>0.018738</td>\n",
       "      <td>-0.000072</td>\n",
       "      <td>-0.002064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_root</th>\n",
       "      <td>0.097136</td>\n",
       "      <td>-0.000070</td>\n",
       "      <td>0.024942</td>\n",
       "      <td>-0.000035</td>\n",
       "      <td>-0.000160</td>\n",
       "      <td>0.047759</td>\n",
       "      <td>0.009173</td>\n",
       "      <td>0.001904</td>\n",
       "      <td>0.006070</td>\n",
       "      <td>0.995128</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.007071</td>\n",
       "      <td>-0.023233</td>\n",
       "      <td>-0.023059</td>\n",
       "      <td>0.002970</td>\n",
       "      <td>0.000331</td>\n",
       "      <td>-0.000678</td>\n",
       "      <td>0.005444</td>\n",
       "      <td>0.008525</td>\n",
       "      <td>-0.002151</td>\n",
       "      <td>-0.002218</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_file_creations</th>\n",
       "      <td>0.063465</td>\n",
       "      <td>-0.000063</td>\n",
       "      <td>0.002815</td>\n",
       "      <td>-0.000066</td>\n",
       "      <td>-0.000216</td>\n",
       "      <td>0.018094</td>\n",
       "      <td>0.277273</td>\n",
       "      <td>0.003842</td>\n",
       "      <td>0.010776</td>\n",
       "      <td>0.011638</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.004307</td>\n",
       "      <td>-0.034042</td>\n",
       "      <td>-0.031373</td>\n",
       "      <td>0.007161</td>\n",
       "      <td>0.001465</td>\n",
       "      <td>0.005422</td>\n",
       "      <td>0.028576</td>\n",
       "      <td>0.016124</td>\n",
       "      <td>-0.001866</td>\n",
       "      <td>-0.001514</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_shells</th>\n",
       "      <td>0.001948</td>\n",
       "      <td>-0.000075</td>\n",
       "      <td>0.000923</td>\n",
       "      <td>-0.000061</td>\n",
       "      <td>-0.000269</td>\n",
       "      <td>0.033402</td>\n",
       "      <td>0.021819</td>\n",
       "      <td>-0.000895</td>\n",
       "      <td>0.010520</td>\n",
       "      <td>0.026640</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.000468</td>\n",
       "      <td>-0.034438</td>\n",
       "      <td>-0.031464</td>\n",
       "      <td>-0.000599</td>\n",
       "      <td>0.013197</td>\n",
       "      <td>0.009246</td>\n",
       "      <td>0.015544</td>\n",
       "      <td>0.023794</td>\n",
       "      <td>-0.001310</td>\n",
       "      <td>-0.003008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_access_files</th>\n",
       "      <td>0.023541</td>\n",
       "      <td>-0.000462</td>\n",
       "      <td>0.006244</td>\n",
       "      <td>-0.000176</td>\n",
       "      <td>-0.001052</td>\n",
       "      <td>0.040267</td>\n",
       "      <td>0.011132</td>\n",
       "      <td>0.004123</td>\n",
       "      <td>0.033241</td>\n",
       "      <td>0.309364</td>\n",
       "      <td>...</td>\n",
       "      <td>0.025922</td>\n",
       "      <td>-0.017639</td>\n",
       "      <td>-0.015384</td>\n",
       "      <td>-0.001492</td>\n",
       "      <td>-0.020680</td>\n",
       "      <td>-0.010681</td>\n",
       "      <td>0.002807</td>\n",
       "      <td>0.005124</td>\n",
       "      <td>-0.009146</td>\n",
       "      <td>-0.012994</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>num_outbound_cmds</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is_host_login</th>\n",
       "      <td>0.000628</td>\n",
       "      <td>-0.000120</td>\n",
       "      <td>-0.000663</td>\n",
       "      <td>-0.000024</td>\n",
       "      <td>-0.000164</td>\n",
       "      <td>-0.000102</td>\n",
       "      <td>0.022067</td>\n",
       "      <td>0.006568</td>\n",
       "      <td>0.005823</td>\n",
       "      <td>-0.000080</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.007949</td>\n",
       "      <td>-0.024007</td>\n",
       "      <td>-0.017637</td>\n",
       "      <td>0.013872</td>\n",
       "      <td>0.012326</td>\n",
       "      <td>-0.002889</td>\n",
       "      <td>-0.000882</td>\n",
       "      <td>-0.000739</td>\n",
       "      <td>-0.001759</td>\n",
       "      <td>-0.001640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is_guest_login</th>\n",
       "      <td>0.024576</td>\n",
       "      <td>-0.000872</td>\n",
       "      <td>-0.005921</td>\n",
       "      <td>-0.000248</td>\n",
       "      <td>-0.001937</td>\n",
       "      <td>-0.000915</td>\n",
       "      <td>0.674614</td>\n",
       "      <td>0.054679</td>\n",
       "      <td>0.036790</td>\n",
       "      <td>-0.000810</td>\n",
       "      <td>...</td>\n",
       "      <td>0.034730</td>\n",
       "      <td>-0.186555</td>\n",
       "      <td>-0.196786</td>\n",
       "      <td>0.040652</td>\n",
       "      <td>-0.036207</td>\n",
       "      <td>-0.036194</td>\n",
       "      <td>0.040115</td>\n",
       "      <td>-0.006210</td>\n",
       "      <td>0.025195</td>\n",
       "      <td>-0.021636</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>-0.021264</td>\n",
       "      <td>0.000158</td>\n",
       "      <td>-0.014496</td>\n",
       "      <td>-0.000659</td>\n",
       "      <td>0.006650</td>\n",
       "      <td>-0.002331</td>\n",
       "      <td>-0.018483</td>\n",
       "      <td>-0.013245</td>\n",
       "      <td>-0.234428</td>\n",
       "      <td>-0.002185</td>\n",
       "      <td>...</td>\n",
       "      <td>0.180280</td>\n",
       "      <td>-0.138073</td>\n",
       "      <td>-0.202353</td>\n",
       "      <td>0.441878</td>\n",
       "      <td>0.041525</td>\n",
       "      <td>-0.075755</td>\n",
       "      <td>0.173170</td>\n",
       "      <td>0.259073</td>\n",
       "      <td>0.310146</td>\n",
       "      <td>0.268546</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>srv_count</th>\n",
       "      <td>-0.040256</td>\n",
       "      <td>-0.002799</td>\n",
       "      <td>-0.014572</td>\n",
       "      <td>-0.001177</td>\n",
       "      <td>0.010474</td>\n",
       "      <td>-0.003312</td>\n",
       "      <td>-0.029951</td>\n",
       "      <td>-0.017527</td>\n",
       "      <td>-0.122827</td>\n",
       "      <td>-0.003409</td>\n",
       "      <td>...</td>\n",
       "      <td>0.154687</td>\n",
       "      <td>0.176394</td>\n",
       "      <td>0.158120</td>\n",
       "      <td>-0.109335</td>\n",
       "      <td>0.033510</td>\n",
       "      <td>-0.062443</td>\n",
       "      <td>-0.028618</td>\n",
       "      <td>-0.023342</td>\n",
       "      <td>-0.096552</td>\n",
       "      <td>-0.091356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>serror_rate</th>\n",
       "      <td>0.012695</td>\n",
       "      <td>0.032286</td>\n",
       "      <td>0.002691</td>\n",
       "      <td>0.055010</td>\n",
       "      <td>-0.001529</td>\n",
       "      <td>-0.000804</td>\n",
       "      <td>-0.002418</td>\n",
       "      <td>0.015621</td>\n",
       "      <td>-0.111272</td>\n",
       "      <td>-0.000461</td>\n",
       "      <td>...</td>\n",
       "      <td>0.001016</td>\n",
       "      <td>-0.131988</td>\n",
       "      <td>-0.155080</td>\n",
       "      <td>0.198254</td>\n",
       "      <td>0.042990</td>\n",
       "      <td>-0.009009</td>\n",
       "      <td>0.727948</td>\n",
       "      <td>0.624134</td>\n",
       "      <td>0.073745</td>\n",
       "      <td>0.076238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>srv_error_rate</th>\n",
       "      <td>0.000556</td>\n",
       "      <td>0.020482</td>\n",
       "      <td>0.000897</td>\n",
       "      <td>0.041417</td>\n",
       "      <td>-0.001558</td>\n",
       "      <td>-0.000784</td>\n",
       "      <td>-0.002973</td>\n",
       "      <td>0.005104</td>\n",
       "      <td>-0.111494</td>\n",
       "      <td>-0.000733</td>\n",
       "      <td>...</td>\n",
       "      <td>0.015448</td>\n",
       "      <td>-0.132549</td>\n",
       "      <td>-0.159816</td>\n",
       "      <td>0.226609</td>\n",
       "      <td>0.029486</td>\n",
       "      <td>-0.011452</td>\n",
       "      <td>0.556522</td>\n",
       "      <td>0.876172</td>\n",
       "      <td>0.120649</td>\n",
       "      <td>0.003944</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rerror_rate</th>\n",
       "      <td>0.024791</td>\n",
       "      <td>-0.000136</td>\n",
       "      <td>-0.016999</td>\n",
       "      <td>-0.000719</td>\n",
       "      <td>-0.003256</td>\n",
       "      <td>-0.001920</td>\n",
       "      <td>-0.015689</td>\n",
       "      <td>0.039348</td>\n",
       "      <td>-0.438489</td>\n",
       "      <td>-0.001872</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.072895</td>\n",
       "      <td>-0.231681</td>\n",
       "      <td>-0.236533</td>\n",
       "      <td>0.290000</td>\n",
       "      <td>0.060870</td>\n",
       "      <td>0.198658</td>\n",
       "      <td>0.072024</td>\n",
       "      <td>0.115323</td>\n",
       "      <td>0.904562</td>\n",
       "      <td>0.926933</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>srv_rerror_rate</th>\n",
       "      <td>0.025714</td>\n",
       "      <td>0.000916</td>\n",
       "      <td>-0.016476</td>\n",
       "      <td>-0.000716</td>\n",
       "      <td>-0.003178</td>\n",
       "      <td>-0.001892</td>\n",
       "      <td>-0.016011</td>\n",
       "      <td>0.039899</td>\n",
       "      <td>-0.427260</td>\n",
       "      <td>-0.001867</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.075301</td>\n",
       "      <td>-0.220182</td>\n",
       "      <td>-0.222191</td>\n",
       "      <td>0.266824</td>\n",
       "      <td>0.061757</td>\n",
       "      <td>0.199597</td>\n",
       "      <td>0.078206</td>\n",
       "      <td>0.002159</td>\n",
       "      <td>0.873826</td>\n",
       "      <td>0.945762</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>same_srv_rate</th>\n",
       "      <td>-0.041337</td>\n",
       "      <td>-0.012252</td>\n",
       "      <td>0.014384</td>\n",
       "      <td>0.000488</td>\n",
       "      <td>0.003602</td>\n",
       "      <td>0.001704</td>\n",
       "      <td>0.004633</td>\n",
       "      <td>0.006845</td>\n",
       "      <td>0.309230</td>\n",
       "      <td>0.001489</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.159736</td>\n",
       "      <td>0.395865</td>\n",
       "      <td>0.487583</td>\n",
       "      <td>-0.602360</td>\n",
       "      <td>-0.144716</td>\n",
       "      <td>0.057155</td>\n",
       "      <td>-0.150115</td>\n",
       "      <td>-0.217234</td>\n",
       "      <td>-0.365158</td>\n",
       "      <td>-0.348882</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>diff_srv_rate</th>\n",
       "      <td>0.036048</td>\n",
       "      <td>0.003624</td>\n",
       "      <td>-0.014178</td>\n",
       "      <td>-0.000492</td>\n",
       "      <td>-0.003687</td>\n",
       "      <td>-0.001741</td>\n",
       "      <td>0.000676</td>\n",
       "      <td>-0.004745</td>\n",
       "      <td>-0.286986</td>\n",
       "      <td>-0.001585</td>\n",
       "      <td>...</td>\n",
       "      <td>0.143637</td>\n",
       "      <td>-0.381193</td>\n",
       "      <td>-0.466588</td>\n",
       "      <td>0.531331</td>\n",
       "      <td>0.105483</td>\n",
       "      <td>-0.050784</td>\n",
       "      <td>0.127415</td>\n",
       "      <td>0.185534</td>\n",
       "      <td>0.345461</td>\n",
       "      <td>0.349024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>srv_diff_host_rate</th>\n",
       "      <td>-0.037919</td>\n",
       "      <td>-0.003569</td>\n",
       "      <td>-0.013117</td>\n",
       "      <td>0.007200</td>\n",
       "      <td>-0.005214</td>\n",
       "      <td>-0.004446</td>\n",
       "      <td>-0.032686</td>\n",
       "      <td>-0.025858</td>\n",
       "      <td>-0.019771</td>\n",
       "      <td>-0.004230</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.119273</td>\n",
       "      <td>-0.085060</td>\n",
       "      <td>-0.051782</td>\n",
       "      <td>-0.058600</td>\n",
       "      <td>-0.010034</td>\n",
       "      <td>0.081464</td>\n",
       "      <td>0.030203</td>\n",
       "      <td>0.007719</td>\n",
       "      <td>0.075779</td>\n",
       "      <td>0.086385</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_count</th>\n",
       "      <td>0.068190</td>\n",
       "      <td>0.000027</td>\n",
       "      <td>-0.019988</td>\n",
       "      <td>-0.003197</td>\n",
       "      <td>-0.001357</td>\n",
       "      <td>-0.007518</td>\n",
       "      <td>0.023482</td>\n",
       "      <td>0.017104</td>\n",
       "      <td>-0.148581</td>\n",
       "      <td>-0.006233</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.006110</td>\n",
       "      <td>-0.184441</td>\n",
       "      <td>0.168316</td>\n",
       "      <td>-0.205418</td>\n",
       "      <td>-0.389799</td>\n",
       "      <td>0.016860</td>\n",
       "      <td>0.019382</td>\n",
       "      <td>-0.079978</td>\n",
       "      <td>-0.079868</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_srv_count</th>\n",
       "      <td>-0.189572</td>\n",
       "      <td>-0.017917</td>\n",
       "      <td>-0.015013</td>\n",
       "      <td>-0.005172</td>\n",
       "      <td>-0.039489</td>\n",
       "      <td>-0.019645</td>\n",
       "      <td>-0.126645</td>\n",
       "      <td>-0.060760</td>\n",
       "      <td>0.462860</td>\n",
       "      <td>-0.018696</td>\n",
       "      <td>...</td>\n",
       "      <td>0.006110</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.838576</td>\n",
       "      <td>-0.581844</td>\n",
       "      <td>-0.410819</td>\n",
       "      <td>-0.126110</td>\n",
       "      <td>-0.173095</td>\n",
       "      <td>-0.150779</td>\n",
       "      <td>-0.233564</td>\n",
       "      <td>-0.228777</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_same_srv_rate</th>\n",
       "      <td>-0.214620</td>\n",
       "      <td>-0.015160</td>\n",
       "      <td>-0.002238</td>\n",
       "      <td>0.001413</td>\n",
       "      <td>-0.032805</td>\n",
       "      <td>-0.015221</td>\n",
       "      <td>-0.134481</td>\n",
       "      <td>-0.043562</td>\n",
       "      <td>0.435803</td>\n",
       "      <td>-0.017727</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.184441</td>\n",
       "      <td>0.838576</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.686633</td>\n",
       "      <td>-0.280306</td>\n",
       "      <td>0.043227</td>\n",
       "      <td>-0.205075</td>\n",
       "      <td>-0.177016</td>\n",
       "      <td>-0.238685</td>\n",
       "      <td>-0.229339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_diff_srv_rate</th>\n",
       "      <td>0.255215</td>\n",
       "      <td>0.001398</td>\n",
       "      <td>-0.016368</td>\n",
       "      <td>-0.000864</td>\n",
       "      <td>0.002834</td>\n",
       "      <td>0.013500</td>\n",
       "      <td>0.016140</td>\n",
       "      <td>-0.002525</td>\n",
       "      <td>-0.398185</td>\n",
       "      <td>0.001994</td>\n",
       "      <td>...</td>\n",
       "      <td>0.168316</td>\n",
       "      <td>-0.581844</td>\n",
       "      <td>-0.686633</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.379532</td>\n",
       "      <td>-0.038841</td>\n",
       "      <td>0.205924</td>\n",
       "      <td>0.238647</td>\n",
       "      <td>0.342405</td>\n",
       "      <td>0.274681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_same_src_port_rate</th>\n",
       "      <td>0.204651</td>\n",
       "      <td>0.008754</td>\n",
       "      <td>0.007375</td>\n",
       "      <td>0.007509</td>\n",
       "      <td>0.030775</td>\n",
       "      <td>0.003135</td>\n",
       "      <td>-0.021403</td>\n",
       "      <td>-0.014387</td>\n",
       "      <td>-0.397248</td>\n",
       "      <td>-0.001598</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.205418</td>\n",
       "      <td>-0.410819</td>\n",
       "      <td>-0.280306</td>\n",
       "      <td>0.379532</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.273264</td>\n",
       "      <td>0.036013</td>\n",
       "      <td>0.030742</td>\n",
       "      <td>0.067507</td>\n",
       "      <td>0.056124</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_srv_diff_host_rate</th>\n",
       "      <td>-0.024366</td>\n",
       "      <td>0.002703</td>\n",
       "      <td>0.002882</td>\n",
       "      <td>0.008351</td>\n",
       "      <td>0.007797</td>\n",
       "      <td>0.002830</td>\n",
       "      <td>-0.019025</td>\n",
       "      <td>-0.003188</td>\n",
       "      <td>-0.165710</td>\n",
       "      <td>-0.001139</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.389799</td>\n",
       "      <td>-0.126110</td>\n",
       "      <td>0.043227</td>\n",
       "      <td>-0.038841</td>\n",
       "      <td>0.273264</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.012589</td>\n",
       "      <td>-0.011162</td>\n",
       "      <td>0.209053</td>\n",
       "      <td>0.214990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_serror_rate</th>\n",
       "      <td>0.012989</td>\n",
       "      <td>0.007055</td>\n",
       "      <td>0.003843</td>\n",
       "      <td>0.055836</td>\n",
       "      <td>0.011099</td>\n",
       "      <td>0.001669</td>\n",
       "      <td>0.018972</td>\n",
       "      <td>0.027317</td>\n",
       "      <td>-0.102782</td>\n",
       "      <td>0.007625</td>\n",
       "      <td>...</td>\n",
       "      <td>0.016860</td>\n",
       "      <td>-0.173095</td>\n",
       "      <td>-0.205075</td>\n",
       "      <td>0.205924</td>\n",
       "      <td>0.036013</td>\n",
       "      <td>-0.012589</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.612745</td>\n",
       "      <td>0.073076</td>\n",
       "      <td>0.077448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_srv_serror_rate</th>\n",
       "      <td>0.006247</td>\n",
       "      <td>0.022344</td>\n",
       "      <td>0.007759</td>\n",
       "      <td>0.003844</td>\n",
       "      <td>-0.001459</td>\n",
       "      <td>-0.000639</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>0.017950</td>\n",
       "      <td>-0.112004</td>\n",
       "      <td>0.010023</td>\n",
       "      <td>...</td>\n",
       "      <td>0.019382</td>\n",
       "      <td>-0.150779</td>\n",
       "      <td>-0.177016</td>\n",
       "      <td>0.238647</td>\n",
       "      <td>0.030742</td>\n",
       "      <td>-0.011162</td>\n",
       "      <td>0.612745</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.118429</td>\n",
       "      <td>-0.006152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_rerror_rate</th>\n",
       "      <td>0.013741</td>\n",
       "      <td>-0.001297</td>\n",
       "      <td>-0.015288</td>\n",
       "      <td>-0.000759</td>\n",
       "      <td>0.006002</td>\n",
       "      <td>-0.000892</td>\n",
       "      <td>0.002000</td>\n",
       "      <td>0.032950</td>\n",
       "      <td>-0.414882</td>\n",
       "      <td>-0.001596</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.079978</td>\n",
       "      <td>-0.233564</td>\n",
       "      <td>-0.238685</td>\n",
       "      <td>0.342405</td>\n",
       "      <td>0.067507</td>\n",
       "      <td>0.209053</td>\n",
       "      <td>0.073076</td>\n",
       "      <td>0.118429</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.886568</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dst_host_srv_rerror_rate</th>\n",
       "      <td>0.020110</td>\n",
       "      <td>0.000915</td>\n",
       "      <td>-0.015691</td>\n",
       "      <td>-0.000738</td>\n",
       "      <td>-0.003239</td>\n",
       "      <td>0.003740</td>\n",
       "      <td>-0.014772</td>\n",
       "      <td>0.036186</td>\n",
       "      <td>-0.409466</td>\n",
       "      <td>-0.001197</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.079868</td>\n",
       "      <td>-0.228777</td>\n",
       "      <td>-0.229339</td>\n",
       "      <td>0.274681</td>\n",
       "      <td>0.056124</td>\n",
       "      <td>0.214990</td>\n",
       "      <td>0.077448</td>\n",
       "      <td>-0.006152</td>\n",
       "      <td>0.886568</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>38 rows × 38 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                             Duration  src_bytes  dst_bytes      land  \\\n",
       "Duration                     1.000000   0.004578   0.021672 -0.000351   \n",
       "src_bytes                    0.004578   1.000000   0.007928 -0.000014   \n",
       "dst_bytes                    0.021672   0.007928   1.000000 -0.000176   \n",
       "land                        -0.000351  -0.000014  -0.000176  1.000000   \n",
       "wrong_fragment              -0.000347  -0.000233  -0.001709 -0.000047   \n",
       "urgent                       0.003690  -0.000045   0.011452 -0.000019   \n",
       "hot                          0.012719   0.002330   0.000287 -0.000208   \n",
       "num_failed                   0.001486  -0.000535   0.025406 -0.000062   \n",
       "logged_in                   -0.146037  -0.012170   0.019351 -0.005424   \n",
       "num_compromised              0.098938  -0.000056   0.026384 -0.000022   \n",
       "root_shell                   0.027823  -0.000163   0.025422 -0.000071   \n",
       "su_attempted                 0.088947   0.000016   0.052575 -0.000029   \n",
       "num_root                     0.097136  -0.000070   0.024942 -0.000035   \n",
       "num_file_creations           0.063465  -0.000063   0.002815 -0.000066   \n",
       "num_shells                   0.001948  -0.000075   0.000923 -0.000061   \n",
       "num_access_files             0.023541  -0.000462   0.006244 -0.000176   \n",
       "num_outbound_cmds                 NaN        NaN        NaN       NaN   \n",
       "is_host_login                0.000628  -0.000120  -0.000663 -0.000024   \n",
       "is_guest_login               0.024576  -0.000872  -0.005921 -0.000248   \n",
       "count                       -0.021264   0.000158  -0.014496 -0.000659   \n",
       "srv_count                   -0.040256  -0.002799  -0.014572 -0.001177   \n",
       "serror_rate                  0.012695   0.032286   0.002691  0.055010   \n",
       "srv_error_rate               0.000556   0.020482   0.000897  0.041417   \n",
       "rerror_rate                  0.024791  -0.000136  -0.016999 -0.000719   \n",
       "srv_rerror_rate              0.025714   0.000916  -0.016476 -0.000716   \n",
       "same_srv_rate               -0.041337  -0.012252   0.014384  0.000488   \n",
       "diff_srv_rate                0.036048   0.003624  -0.014178 -0.000492   \n",
       "srv_diff_host_rate          -0.037919  -0.003569  -0.013117  0.007200   \n",
       "dst_host_count               0.068190   0.000027  -0.019988 -0.003197   \n",
       "dst_host_srv_count          -0.189572  -0.017917  -0.015013 -0.005172   \n",
       "dst_host_same_srv_rate      -0.214620  -0.015160  -0.002238  0.001413   \n",
       "dst_host_diff_srv_rate       0.255215   0.001398  -0.016368 -0.000864   \n",
       "dst_host_same_src_port_rate  0.204651   0.008754   0.007375  0.007509   \n",
       "dst_host_srv_diff_host_rate -0.024366   0.002703   0.002882  0.008351   \n",
       "dst_host_serror_rate         0.012989   0.007055   0.003843  0.055836   \n",
       "dst_host_srv_serror_rate     0.006247   0.022344   0.007759  0.003844   \n",
       "dst_host_rerror_rate         0.013741  -0.001297  -0.015288 -0.000759   \n",
       "dst_host_srv_rerror_rate     0.020110   0.000915  -0.015691 -0.000738   \n",
       "\n",
       "                             wrong_fragment    urgent       hot  num_failed  \\\n",
       "Duration                          -0.000347  0.003690  0.012719    0.001486   \n",
       "src_bytes                         -0.000233 -0.000045  0.002330   -0.000535   \n",
       "dst_bytes                         -0.001709  0.011452  0.000287    0.025406   \n",
       "land                              -0.000047 -0.000019 -0.000208   -0.000062   \n",
       "wrong_fragment                     1.000000 -0.000202 -0.001089   -0.001643   \n",
       "urgent                            -0.000202  1.000000  0.024185    0.084431   \n",
       "hot                               -0.001089  0.024185  1.000000    0.007537   \n",
       "num_failed                        -0.001643  0.084431  0.007537    1.000000   \n",
       "logged_in                         -0.030348  0.004443  0.035556   -0.087611   \n",
       "num_compromised                   -0.000159  0.039020  0.009142    0.004282   \n",
       "root_shell                        -0.000555  0.172920  0.067311    0.022762   \n",
       "su_attempted                      -0.000154 -0.000085  0.001754    0.069621   \n",
       "num_root                          -0.000160  0.047759  0.009173    0.001904   \n",
       "num_file_creations                -0.000216  0.018094  0.277273    0.003842   \n",
       "num_shells                        -0.000269  0.033402  0.021819   -0.000895   \n",
       "num_access_files                  -0.001052  0.040267  0.011132    0.004123   \n",
       "num_outbound_cmds                       NaN       NaN       NaN         NaN   \n",
       "is_host_login                     -0.000164 -0.000102  0.022067    0.006568   \n",
       "is_guest_login                    -0.001937 -0.000915  0.674614    0.054679   \n",
       "count                              0.006650 -0.002331 -0.018483   -0.013245   \n",
       "srv_count                          0.010474 -0.003312 -0.029951   -0.017527   \n",
       "serror_rate                       -0.001529 -0.000804 -0.002418    0.015621   \n",
       "srv_error_rate                    -0.001558 -0.000784 -0.002973    0.005104   \n",
       "rerror_rate                       -0.003256 -0.001920 -0.015689    0.039348   \n",
       "srv_rerror_rate                   -0.003178 -0.001892 -0.016011    0.039899   \n",
       "same_srv_rate                      0.003602  0.001704  0.004633    0.006845   \n",
       "diff_srv_rate                     -0.003687 -0.001741  0.000676   -0.004745   \n",
       "srv_diff_host_rate                -0.005214 -0.004446 -0.032686   -0.025858   \n",
       "dst_host_count                    -0.001357 -0.007518  0.023482    0.017104   \n",
       "dst_host_srv_count                -0.039489 -0.019645 -0.126645   -0.060760   \n",
       "dst_host_same_srv_rate            -0.032805 -0.015221 -0.134481   -0.043562   \n",
       "dst_host_diff_srv_rate             0.002834  0.013500  0.016140   -0.002525   \n",
       "dst_host_same_src_port_rate        0.030775  0.003135 -0.021403   -0.014387   \n",
       "dst_host_srv_diff_host_rate        0.007797  0.002830 -0.019025   -0.003188   \n",
       "dst_host_serror_rate               0.011099  0.001669  0.018972    0.027317   \n",
       "dst_host_srv_serror_rate          -0.001459 -0.000639 -0.000583    0.017950   \n",
       "dst_host_rerror_rate               0.006002 -0.000892  0.002000    0.032950   \n",
       "dst_host_srv_rerror_rate          -0.003239  0.003740 -0.014772    0.036186   \n",
       "\n",
       "                             logged_in  num_compromised  ...  dst_host_count  \\\n",
       "Duration                     -0.146037         0.098938  ...        0.068190   \n",
       "src_bytes                    -0.012170        -0.000056  ...        0.000027   \n",
       "dst_bytes                     0.019351         0.026384  ...       -0.019988   \n",
       "land                         -0.005424        -0.000022  ...       -0.003197   \n",
       "wrong_fragment               -0.030348        -0.000159  ...       -0.001357   \n",
       "urgent                        0.004443         0.039020  ...       -0.007518   \n",
       "hot                           0.035556         0.009142  ...        0.023482   \n",
       "num_failed                   -0.087611         0.004282  ...        0.017104   \n",
       "logged_in                     1.000000         0.004389  ...       -0.148581   \n",
       "num_compromised               0.004389         1.000000  ...       -0.006233   \n",
       "root_shell                    0.014683         0.261008  ...       -0.007163   \n",
       "su_attempted                  0.005313         0.736226  ...       -0.001939   \n",
       "num_root                      0.006070         0.995128  ...       -0.007071   \n",
       "num_file_creations            0.010776         0.011638  ...       -0.004307   \n",
       "num_shells                    0.010520         0.026640  ...       -0.000468   \n",
       "num_access_files              0.033241         0.309364  ...        0.025922   \n",
       "num_outbound_cmds                  NaN              NaN  ...             NaN   \n",
       "is_host_login                 0.005823        -0.000080  ...       -0.007949   \n",
       "is_guest_login                0.036790        -0.000810  ...        0.034730   \n",
       "count                        -0.234428        -0.002185  ...        0.180280   \n",
       "srv_count                    -0.122827        -0.003409  ...        0.154687   \n",
       "serror_rate                  -0.111272        -0.000461  ...        0.001016   \n",
       "srv_error_rate               -0.111494        -0.000733  ...        0.015448   \n",
       "rerror_rate                  -0.438489        -0.001872  ...       -0.072895   \n",
       "srv_rerror_rate              -0.427260        -0.001867  ...       -0.075301   \n",
       "same_srv_rate                 0.309230         0.001489  ...       -0.159736   \n",
       "diff_srv_rate                -0.286986        -0.001585  ...        0.143637   \n",
       "srv_diff_host_rate           -0.019771        -0.004230  ...       -0.119273   \n",
       "dst_host_count               -0.148581        -0.006233  ...        1.000000   \n",
       "dst_host_srv_count            0.462860        -0.018696  ...        0.006110   \n",
       "dst_host_same_srv_rate        0.435803        -0.017727  ...       -0.184441   \n",
       "dst_host_diff_srv_rate       -0.398185         0.001994  ...        0.168316   \n",
       "dst_host_same_src_port_rate  -0.397248        -0.001598  ...       -0.205418   \n",
       "dst_host_srv_diff_host_rate  -0.165710        -0.001139  ...       -0.389799   \n",
       "dst_host_serror_rate         -0.102782         0.007625  ...        0.016860   \n",
       "dst_host_srv_serror_rate     -0.112004         0.010023  ...        0.019382   \n",
       "dst_host_rerror_rate         -0.414882        -0.001596  ...       -0.079978   \n",
       "dst_host_srv_rerror_rate     -0.409466        -0.001197  ...       -0.079868   \n",
       "\n",
       "                             dst_host_srv_count  dst_host_same_srv_rate  \\\n",
       "Duration                              -0.189572               -0.214620   \n",
       "src_bytes                             -0.017917               -0.015160   \n",
       "dst_bytes                             -0.015013               -0.002238   \n",
       "land                                  -0.005172                0.001413   \n",
       "wrong_fragment                        -0.039489               -0.032805   \n",
       "urgent                                -0.019645               -0.015221   \n",
       "hot                                   -0.126645               -0.134481   \n",
       "num_failed                            -0.060760               -0.043562   \n",
       "logged_in                              0.462860                0.435803   \n",
       "num_compromised                       -0.018696               -0.017727   \n",
       "root_shell                            -0.038762               -0.024500   \n",
       "su_attempted                          -0.019681               -0.021385   \n",
       "num_root                              -0.023233               -0.023059   \n",
       "num_file_creations                    -0.034042               -0.031373   \n",
       "num_shells                            -0.034438               -0.031464   \n",
       "num_access_files                      -0.017639               -0.015384   \n",
       "num_outbound_cmds                           NaN                     NaN   \n",
       "is_host_login                         -0.024007               -0.017637   \n",
       "is_guest_login                        -0.186555               -0.196786   \n",
       "count                                 -0.138073               -0.202353   \n",
       "srv_count                              0.176394                0.158120   \n",
       "serror_rate                           -0.131988               -0.155080   \n",
       "srv_error_rate                        -0.132549               -0.159816   \n",
       "rerror_rate                           -0.231681               -0.236533   \n",
       "srv_rerror_rate                       -0.220182               -0.222191   \n",
       "same_srv_rate                          0.395865                0.487583   \n",
       "diff_srv_rate                         -0.381193               -0.466588   \n",
       "srv_diff_host_rate                    -0.085060               -0.051782   \n",
       "dst_host_count                         0.006110               -0.184441   \n",
       "dst_host_srv_count                     1.000000                0.838576   \n",
       "dst_host_same_srv_rate                 0.838576                1.000000   \n",
       "dst_host_diff_srv_rate                -0.581844               -0.686633   \n",
       "dst_host_same_src_port_rate           -0.410819               -0.280306   \n",
       "dst_host_srv_diff_host_rate           -0.126110                0.043227   \n",
       "dst_host_serror_rate                  -0.173095               -0.205075   \n",
       "dst_host_srv_serror_rate              -0.150779               -0.177016   \n",
       "dst_host_rerror_rate                  -0.233564               -0.238685   \n",
       "dst_host_srv_rerror_rate              -0.228777               -0.229339   \n",
       "\n",
       "                             dst_host_diff_srv_rate  \\\n",
       "Duration                                   0.255215   \n",
       "src_bytes                                  0.001398   \n",
       "dst_bytes                                 -0.016368   \n",
       "land                                      -0.000864   \n",
       "wrong_fragment                             0.002834   \n",
       "urgent                                     0.013500   \n",
       "hot                                        0.016140   \n",
       "num_failed                                -0.002525   \n",
       "logged_in                                 -0.398185   \n",
       "num_compromised                            0.001994   \n",
       "root_shell                                 0.003803   \n",
       "su_attempted                               0.000534   \n",
       "num_root                                   0.002970   \n",
       "num_file_creations                         0.007161   \n",
       "num_shells                                -0.000599   \n",
       "num_access_files                          -0.001492   \n",
       "num_outbound_cmds                               NaN   \n",
       "is_host_login                              0.013872   \n",
       "is_guest_login                             0.040652   \n",
       "count                                      0.441878   \n",
       "srv_count                                 -0.109335   \n",
       "serror_rate                                0.198254   \n",
       "srv_error_rate                             0.226609   \n",
       "rerror_rate                                0.290000   \n",
       "srv_rerror_rate                            0.266824   \n",
       "same_srv_rate                             -0.602360   \n",
       "diff_srv_rate                              0.531331   \n",
       "srv_diff_host_rate                        -0.058600   \n",
       "dst_host_count                             0.168316   \n",
       "dst_host_srv_count                        -0.581844   \n",
       "dst_host_same_srv_rate                    -0.686633   \n",
       "dst_host_diff_srv_rate                     1.000000   \n",
       "dst_host_same_src_port_rate                0.379532   \n",
       "dst_host_srv_diff_host_rate               -0.038841   \n",
       "dst_host_serror_rate                       0.205924   \n",
       "dst_host_srv_serror_rate                   0.238647   \n",
       "dst_host_rerror_rate                       0.342405   \n",
       "dst_host_srv_rerror_rate                   0.274681   \n",
       "\n",
       "                             dst_host_same_src_port_rate  \\\n",
       "Duration                                        0.204651   \n",
       "src_bytes                                       0.008754   \n",
       "dst_bytes                                       0.007375   \n",
       "land                                            0.007509   \n",
       "wrong_fragment                                  0.030775   \n",
       "urgent                                          0.003135   \n",
       "hot                                            -0.021403   \n",
       "num_failed                                     -0.014387   \n",
       "logged_in                                      -0.397248   \n",
       "num_compromised                                -0.001598   \n",
       "root_shell                                      0.008618   \n",
       "su_attempted                                   -0.003931   \n",
       "num_root                                        0.000331   \n",
       "num_file_creations                              0.001465   \n",
       "num_shells                                      0.013197   \n",
       "num_access_files                               -0.020680   \n",
       "num_outbound_cmds                                    NaN   \n",
       "is_host_login                                   0.012326   \n",
       "is_guest_login                                 -0.036207   \n",
       "count                                           0.041525   \n",
       "srv_count                                       0.033510   \n",
       "serror_rate                                     0.042990   \n",
       "srv_error_rate                                  0.029486   \n",
       "rerror_rate                                     0.060870   \n",
       "srv_rerror_rate                                 0.061757   \n",
       "same_srv_rate                                  -0.144716   \n",
       "diff_srv_rate                                   0.105483   \n",
       "srv_diff_host_rate                             -0.010034   \n",
       "dst_host_count                                 -0.205418   \n",
       "dst_host_srv_count                             -0.410819   \n",
       "dst_host_same_srv_rate                         -0.280306   \n",
       "dst_host_diff_srv_rate                          0.379532   \n",
       "dst_host_same_src_port_rate                     1.000000   \n",
       "dst_host_srv_diff_host_rate                     0.273264   \n",
       "dst_host_serror_rate                            0.036013   \n",
       "dst_host_srv_serror_rate                        0.030742   \n",
       "dst_host_rerror_rate                            0.067507   \n",
       "dst_host_srv_rerror_rate                        0.056124   \n",
       "\n",
       "                             dst_host_srv_diff_host_rate  \\\n",
       "Duration                                       -0.024366   \n",
       "src_bytes                                       0.002703   \n",
       "dst_bytes                                       0.002882   \n",
       "land                                            0.008351   \n",
       "wrong_fragment                                  0.007797   \n",
       "urgent                                          0.002830   \n",
       "hot                                            -0.019025   \n",
       "num_failed                                     -0.003188   \n",
       "logged_in                                      -0.165710   \n",
       "num_compromised                                -0.001139   \n",
       "root_shell                                      0.012709   \n",
       "su_attempted                                   -0.001855   \n",
       "num_root                                       -0.000678   \n",
       "num_file_creations                              0.005422   \n",
       "num_shells                                      0.009246   \n",
       "num_access_files                               -0.010681   \n",
       "num_outbound_cmds                                    NaN   \n",
       "is_host_login                                  -0.002889   \n",
       "is_guest_login                                 -0.036194   \n",
       "count                                          -0.075755   \n",
       "srv_count                                      -0.062443   \n",
       "serror_rate                                    -0.009009   \n",
       "srv_error_rate                                 -0.011452   \n",
       "rerror_rate                                     0.198658   \n",
       "srv_rerror_rate                                 0.199597   \n",
       "same_srv_rate                                   0.057155   \n",
       "diff_srv_rate                                  -0.050784   \n",
       "srv_diff_host_rate                              0.081464   \n",
       "dst_host_count                                 -0.389799   \n",
       "dst_host_srv_count                             -0.126110   \n",
       "dst_host_same_srv_rate                          0.043227   \n",
       "dst_host_diff_srv_rate                         -0.038841   \n",
       "dst_host_same_src_port_rate                     0.273264   \n",
       "dst_host_srv_diff_host_rate                     1.000000   \n",
       "dst_host_serror_rate                           -0.012589   \n",
       "dst_host_srv_serror_rate                       -0.011162   \n",
       "dst_host_rerror_rate                            0.209053   \n",
       "dst_host_srv_rerror_rate                        0.214990   \n",
       "\n",
       "                             dst_host_serror_rate  dst_host_srv_serror_rate  \\\n",
       "Duration                                 0.012989                  0.006247   \n",
       "src_bytes                                0.007055                  0.022344   \n",
       "dst_bytes                                0.003843                  0.007759   \n",
       "land                                     0.055836                  0.003844   \n",
       "wrong_fragment                           0.011099                 -0.001459   \n",
       "urgent                                   0.001669                 -0.000639   \n",
       "hot                                      0.018972                 -0.000583   \n",
       "num_failed                               0.027317                  0.017950   \n",
       "logged_in                               -0.102782                 -0.112004   \n",
       "num_compromised                          0.007625                  0.010023   \n",
       "root_shell                               0.010551                  0.012970   \n",
       "su_attempted                             0.018427                  0.018738   \n",
       "num_root                                 0.005444                  0.008525   \n",
       "num_file_creations                       0.028576                  0.016124   \n",
       "num_shells                               0.015544                  0.023794   \n",
       "num_access_files                         0.002807                  0.005124   \n",
       "num_outbound_cmds                             NaN                       NaN   \n",
       "is_host_login                           -0.000882                 -0.000739   \n",
       "is_guest_login                           0.040115                 -0.006210   \n",
       "count                                    0.173170                  0.259073   \n",
       "srv_count                               -0.028618                 -0.023342   \n",
       "serror_rate                              0.727948                  0.624134   \n",
       "srv_error_rate                           0.556522                  0.876172   \n",
       "rerror_rate                              0.072024                  0.115323   \n",
       "srv_rerror_rate                          0.078206                  0.002159   \n",
       "same_srv_rate                           -0.150115                 -0.217234   \n",
       "diff_srv_rate                            0.127415                  0.185534   \n",
       "srv_diff_host_rate                       0.030203                  0.007719   \n",
       "dst_host_count                           0.016860                  0.019382   \n",
       "dst_host_srv_count                      -0.173095                 -0.150779   \n",
       "dst_host_same_srv_rate                  -0.205075                 -0.177016   \n",
       "dst_host_diff_srv_rate                   0.205924                  0.238647   \n",
       "dst_host_same_src_port_rate              0.036013                  0.030742   \n",
       "dst_host_srv_diff_host_rate             -0.012589                 -0.011162   \n",
       "dst_host_serror_rate                     1.000000                  0.612745   \n",
       "dst_host_srv_serror_rate                 0.612745                  1.000000   \n",
       "dst_host_rerror_rate                     0.073076                  0.118429   \n",
       "dst_host_srv_rerror_rate                 0.077448                 -0.006152   \n",
       "\n",
       "                             dst_host_rerror_rate  dst_host_srv_rerror_rate  \n",
       "Duration                                 0.013741                  0.020110  \n",
       "src_bytes                               -0.001297                  0.000915  \n",
       "dst_bytes                               -0.015288                 -0.015691  \n",
       "land                                    -0.000759                 -0.000738  \n",
       "wrong_fragment                           0.006002                 -0.003239  \n",
       "urgent                                  -0.000892                  0.003740  \n",
       "hot                                      0.002000                 -0.014772  \n",
       "num_failed                               0.032950                  0.036186  \n",
       "logged_in                               -0.414882                 -0.409466  \n",
       "num_compromised                         -0.001596                 -0.001197  \n",
       "root_shell                               0.002892                  0.001308  \n",
       "su_attempted                            -0.000072                 -0.002064  \n",
       "num_root                                -0.002151                 -0.002218  \n",
       "num_file_creations                      -0.001866                 -0.001514  \n",
       "num_shells                              -0.001310                 -0.003008  \n",
       "num_access_files                        -0.009146                 -0.012994  \n",
       "num_outbound_cmds                             NaN                       NaN  \n",
       "is_host_login                           -0.001759                 -0.001640  \n",
       "is_guest_login                           0.025195                 -0.021636  \n",
       "count                                    0.310146                  0.268546  \n",
       "srv_count                               -0.096552                 -0.091356  \n",
       "serror_rate                              0.073745                  0.076238  \n",
       "srv_error_rate                           0.120649                  0.003944  \n",
       "rerror_rate                              0.904562                  0.926933  \n",
       "srv_rerror_rate                          0.873826                  0.945762  \n",
       "same_srv_rate                           -0.365158                 -0.348882  \n",
       "diff_srv_rate                            0.345461                  0.349024  \n",
       "srv_diff_host_rate                       0.075779                  0.086385  \n",
       "dst_host_count                          -0.079978                 -0.079868  \n",
       "dst_host_srv_count                      -0.233564                 -0.228777  \n",
       "dst_host_same_srv_rate                  -0.238685                 -0.229339  \n",
       "dst_host_diff_srv_rate                   0.342405                  0.274681  \n",
       "dst_host_same_src_port_rate              0.067507                  0.056124  \n",
       "dst_host_srv_diff_host_rate              0.209053                  0.214990  \n",
       "dst_host_serror_rate                     0.073076                  0.077448  \n",
       "dst_host_srv_serror_rate                 0.118429                 -0.006152  \n",
       "dst_host_rerror_rate                     1.000000                  0.886568  \n",
       "dst_host_srv_rerror_rate                 0.886568                  1.000000  \n",
       "\n",
       "[38 rows x 38 columns]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#kdd.corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "qZTBqq6xxzdS"
   },
   "outputs": [],
   "source": [
    "kdd_train = kdd.iloc[0:102563, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "JQ2KBBexxzdU"
   },
   "outputs": [],
   "source": [
    "kdd_test = kdd.iloc[102563:183737, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "gemL8-7axzdW"
   },
   "outputs": [],
   "source": [
    "kdd_train_probe = kdd_train[(kdd_train.Type_Groups == 'Normal') | (kdd_train.Type_Groups == 'Probe')]\n",
    "kdd_test_probe = kdd_test[(kdd_test.Type_Groups == 'Normal') | (kdd_test.Type_Groups == 'Probe')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "id": "phc3gRD7X59o"
   },
   "outputs": [],
   "source": [
    "kdd = pd.concat([kdd_train_probe, kdd_test_probe])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ILHqBF0SX9-4",
    "outputId": "aae01604-3e98-4d25-fb5e-0a386ba90b07"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(166144, 39)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "rV7NgFHAYAYS",
    "outputId": "917689a6-3432-4739-b106-98fe26ae6b9b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Type_Groups\n",
       "Normal    157871\n",
       "Probe       8273\n",
       "Name: Type_Groups, dtype: int64"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd.groupby(['Type_Groups'])['Type_Groups'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4E3ST5hCYMiC",
    "outputId": "7e66298d-55a4-44e0-fa3a-ecc10e504979"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "19.082678593013416"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "157871 / 8273"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "id": "9sRyGK7gV5Oy"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import itertools\n",
    "from sklearn.ensemble import IsolationForest\n",
    "from sklearn.neighbors import LocalOutlierFactor\n",
    "from sklearn import svm\n",
    "from sklearn.neighbors import NearestNeighbors\n",
    "import seaborn as sns\n",
    "\n",
    "\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "from sklearn.metrics import recall_score\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import make_scorer\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "h0VhI_Yohn5v"
   },
   "source": [
    "## 1.1 Isolation Forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "nWsHXjLbPcNa"
   },
   "source": [
    "How many lines should I draw to separate a point from the others? If the point is not in a dense region, it won't take that much lines even if we draw these lines randomly. This is the logic behind the Isolation Forest algorithm. It uses number of lines necessary to separate a point from others to calculate its anomaly score. The most basic (and actually incorrect) way to think of this anomaly score is\n",
    "\n",
    "$AnomalyScore = e^{\\#Lines} $"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "VGnCZb2kghyf"
   },
   "source": [
    "Normals are encoded as 1, anomalies are encoded as -1. Use this version if you want to apply Isolation Forest or Local Outlier Factor."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-diWExxfcH8j"
   },
   "source": [
    "- n_estimators, default=100: The number of base estimators in the ensemble.\n",
    "- max_samples “auto”, int or float, default=”auto”: The number of samples to draw from X to train each base estimator\n",
    "- contamination ‘auto’ or float, default=’auto’: The amount of contamination of the data set, i.e. the proportion of outliers in the data set. Used when fitting to define the threshold on the scores of the samples.\n",
    "- max_features int or float, default=1.0: The number of features to draw from X to train each base estimator.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "QcZqbLtJ3RgW",
    "outputId": "036cdb1e-d308-4fa3-9e69-8f2a227aa417"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n",
      "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \n"
     ]
    }
   ],
   "source": [
    "kdd_train_probe['Type_Groups'] = np.where(kdd_train_probe['Type_Groups'] == 'Normal', 1, -1)\n",
    "kdd_test_probe['Type_Groups'] = np.where(kdd_test_probe['Type_Groups'] == 'Normal', 1, -1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "hPRl1mULdBK8"
   },
   "outputs": [],
   "source": [
    "x_train = kdd_train_probe.drop(['Type_Groups'], axis = 1)\n",
    "y_train = kdd_train_probe['Type_Groups']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "xTk1whWyiSr9"
   },
   "outputs": [],
   "source": [
    "x_test = kdd_test_probe.drop(['Type_Groups'], axis = 1)\n",
    "y_test = kdd_test_probe['Type_Groups']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "tkh6vw8tdh8A"
   },
   "outputs": [],
   "source": [
    "clfIF = IsolationForest(max_samples=0.25, random_state=11, contamination=0.15, n_estimators=100, n_jobs=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "kHrohm2_dmap",
    "outputId": "acb3c5b2-46c1-4305-fc10-788a887149b1"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "IsolationForest(behaviour='deprecated', bootstrap=False, contamination=0.15,\n",
       "                max_features=1.0, max_samples=0.25, n_estimators=100, n_jobs=-1,\n",
       "                random_state=11, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 17,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clfIF.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "euBiTJ5LeGl8"
   },
   "outputs": [],
   "source": [
    "y_pred_train = clfIF.predict(x_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "goT-4ZQBfL1H",
    "outputId": "33afdd7c-0633-456e-8771-35f83c6c6e90"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-1,  1])"
      ]
     },
     "execution_count": 19,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.unique(y_pred_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "TxmfACrjeNl8",
    "outputId": "f1fd6e32-59c1-4552-e7c5-952194c96ca5"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 3996,   111],\n",
       "       [11212, 86066]])"
      ]
     },
     "execution_count": 20,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conf_matrix = confusion_matrix(y_train, y_pred_train)\n",
    "conf_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "TJJDsmz7iNAR"
   },
   "outputs": [],
   "source": [
    "y_pred_test_if = clfIF.predict(x_test)\n",
    "conf_matrix_if_test = confusion_matrix(y_test, y_pred_test_if)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "8BM4PcCElPf6",
    "outputId": "e7cb4cab-b637-45e5-bd69-b5499cfc0f10"
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'conf_matrix_if_test' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m/var/folders/6j/xfjz279j5w73v97s3944dbj80000gn/T/ipykernel_5422/3273571125.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mconf_matrix_if_test\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'conf_matrix_if_test' is not defined"
     ]
    }
   ],
   "source": [
    "conf_matrix_if_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "5gNlUF0DmtoO",
    "outputId": "e7bd78a8-c236-46dc-8a22-c21fd01269b9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "          -1       0.34      1.00      0.50      4166\n",
      "           1       1.00      0.86      0.93     60593\n",
      "\n",
      "    accuracy                           0.87     64759\n",
      "   macro avg       0.67      0.93      0.72     64759\n",
      "weighted avg       0.96      0.87      0.90     64759\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred_test_if))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "oDNzbAGJhd4I"
   },
   "source": [
    "## 1.3 Local Outlier Factor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "## density based methods works better with scaling\n",
    "std_scale = StandardScaler()\n",
    "kdd[['src_bytes']]=std_scale.fit_transform(kdd[['src_bytes']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "KxpmhajEhGyL"
   },
   "outputs": [],
   "source": [
    "clfLOF = LocalOutlierFactor(n_neighbors=2, metric='euclidean', algorithm='auto', contamination=0.15, n_jobs=-1)\n",
    "clfLOF.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "2fO9j-HPh3Fi"
   },
   "outputs": [],
   "source": [
    "y_pred_train_lof = clfLOF.fit_predict(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "id": "brYaF_LgjX72",
    "outputId": "1fb9f3e8-06d7-4b78-9523-f34b8b5dbb7e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  546,  3561],\n",
       "       [14662, 82616]])"
      ]
     },
     "execution_count": 70,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conf_matrix_lof = confusion_matrix(y_train, y_pred_train_lof)\n",
    "conf_matrix_lof"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "id": "fMymZ_MNlG1O",
    "outputId": "c91baf3f-41ce-43db-94e7-7aa137b4b2eb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 1082,  3084],\n",
       "       [ 8632, 51961]])"
      ]
     },
     "execution_count": 74,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred_test_lof = clfLOF.fit_predict(x_test, y_test)\n",
    "conf_matrix_lof_test = confusion_matrix(y_test, y_pred_test_lof)\n",
    "conf_matrix_lof_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "id": "oQHzcfiRlmP0",
    "outputId": "d1722fd8-3782-4aaf-d336-a23bb4296acf"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "          -1       0.11      0.26      0.16      4166\n",
      "           1       0.94      0.86      0.90     60593\n",
      "\n",
      "    accuracy                           0.82     64759\n",
      "   macro avg       0.53      0.56      0.53     64759\n",
      "weighted avg       0.89      0.82      0.85     64759\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred_test_lof))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FrY00ZRxn9lS"
   },
   "source": [
    "## 1.4 Autoencoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "QzQ5iH-OpgQP"
   },
   "outputs": [],
   "source": [
    "import pickle\n",
    "from scipy import stats\n",
    "import tensorflow as tf\n",
    "from pylab import rcParams\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.models import Model, load_model\n",
    "from keras.layers import Input, Dense\n",
    "from keras.callbacks import ModelCheckpoint, TensorBoard\n",
    "from keras import regularizers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6qZu26TppGkF"
   },
   "source": [
    "## 1.4.1 Probing Attack Detection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "8YwMEh2B3cRg",
    "outputId": "5ae444d3-304c-4642-bf88-29094f7b12e4"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:4: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  after removing the cwd from sys.path.\n",
      "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:5: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \"\"\"\n"
     ]
    }
   ],
   "source": [
    "kdd_train_probe = kdd_train[(kdd_train.Type_Groups == 'Normal') | (kdd_train.Type_Groups == 'Probe')]\n",
    "kdd_test_probe = kdd_test[(kdd_test.Type_Groups == 'Normal') | (kdd_test.Type_Groups == 'Probe')]\n",
    "\n",
    "kdd_train_probe['Type_Groups'] = np.where(kdd_train_probe['Type_Groups'] == 'Normal', 0, 1)\n",
    "kdd_test_probe['Type_Groups'] = np.where(kdd_test_probe['Type_Groups'] == 'Normal', 0, 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "PEijPpzhXh5t"
   },
   "source": [
    "## We need to create Validation Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 162
    },
    "id": "RXYZxz5DXmat",
    "outputId": "ccdb4279-f983-455a-8171-254c4d25e4e0"
   },
   "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>Duration</th>\n",
       "      <th>src_bytes</th>\n",
       "      <th>dst_bytes</th>\n",
       "      <th>land</th>\n",
       "      <th>wrong_fragment</th>\n",
       "      <th>urgent</th>\n",
       "      <th>hot</th>\n",
       "      <th>num_failed</th>\n",
       "      <th>logged_in</th>\n",
       "      <th>num_compromised</th>\n",
       "      <th>root_shell</th>\n",
       "      <th>su_attempted</th>\n",
       "      <th>num_root</th>\n",
       "      <th>num_file_creations</th>\n",
       "      <th>num_shells</th>\n",
       "      <th>num_access_files</th>\n",
       "      <th>num_outbound_cmds</th>\n",
       "      <th>is_host_login</th>\n",
       "      <th>is_guest_login</th>\n",
       "      <th>count</th>\n",
       "      <th>srv_count</th>\n",
       "      <th>serror_rate</th>\n",
       "      <th>srv_error_rate</th>\n",
       "      <th>rerror_rate</th>\n",
       "      <th>srv_rerror_rate</th>\n",
       "      <th>same_srv_rate</th>\n",
       "      <th>diff_srv_rate</th>\n",
       "      <th>srv_diff_host_rate</th>\n",
       "      <th>dst_host_count</th>\n",
       "      <th>dst_host_srv_count</th>\n",
       "      <th>dst_host_same_srv_rate</th>\n",
       "      <th>dst_host_diff_srv_rate</th>\n",
       "      <th>dst_host_same_src_port_rate</th>\n",
       "      <th>dst_host_srv_diff_host_rate</th>\n",
       "      <th>dst_host_serror_rate</th>\n",
       "      <th>dst_host_srv_serror_rate</th>\n",
       "      <th>dst_host_rerror_rate</th>\n",
       "      <th>dst_host_srv_rerror_rate</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Type_Groups</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "      <td>60593</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "      <td>4166</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Duration  ...  dst_host_srv_rerror_rate\n",
       "Type_Groups            ...                          \n",
       "0               60593  ...                     60593\n",
       "1                4166  ...                      4166\n",
       "\n",
       "[2 rows x 38 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd_test_probe.groupby('Type_Groups').count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 162
    },
    "id": "K_uA_W-XX67U",
    "outputId": "7dadd279-381f-411a-ad5f-2ca666d28612"
   },
   "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>Duration</th>\n",
       "      <th>src_bytes</th>\n",
       "      <th>dst_bytes</th>\n",
       "      <th>land</th>\n",
       "      <th>wrong_fragment</th>\n",
       "      <th>urgent</th>\n",
       "      <th>hot</th>\n",
       "      <th>num_failed</th>\n",
       "      <th>logged_in</th>\n",
       "      <th>num_compromised</th>\n",
       "      <th>root_shell</th>\n",
       "      <th>su_attempted</th>\n",
       "      <th>num_root</th>\n",
       "      <th>num_file_creations</th>\n",
       "      <th>num_shells</th>\n",
       "      <th>num_access_files</th>\n",
       "      <th>num_outbound_cmds</th>\n",
       "      <th>is_host_login</th>\n",
       "      <th>is_guest_login</th>\n",
       "      <th>count</th>\n",
       "      <th>srv_count</th>\n",
       "      <th>serror_rate</th>\n",
       "      <th>srv_error_rate</th>\n",
       "      <th>rerror_rate</th>\n",
       "      <th>srv_rerror_rate</th>\n",
       "      <th>same_srv_rate</th>\n",
       "      <th>diff_srv_rate</th>\n",
       "      <th>srv_diff_host_rate</th>\n",
       "      <th>dst_host_count</th>\n",
       "      <th>dst_host_srv_count</th>\n",
       "      <th>dst_host_same_srv_rate</th>\n",
       "      <th>dst_host_diff_srv_rate</th>\n",
       "      <th>dst_host_same_src_port_rate</th>\n",
       "      <th>dst_host_srv_diff_host_rate</th>\n",
       "      <th>dst_host_serror_rate</th>\n",
       "      <th>dst_host_srv_serror_rate</th>\n",
       "      <th>dst_host_rerror_rate</th>\n",
       "      <th>dst_host_srv_rerror_rate</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Type_Groups</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "      <td>18756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "      <td>1244</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Duration  ...  dst_host_srv_rerror_rate\n",
       "Type_Groups            ...                          \n",
       "0               18756  ...                     18756\n",
       "1                1244  ...                      1244\n",
       "\n",
       "[2 rows x 38 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd_test_probe.iloc[14000:34000,:].groupby('Type_Groups').count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "0NnV4TglYmJD"
   },
   "outputs": [],
   "source": [
    "kdd_valid_probe = kdd_test_probe.iloc[14000:34000,:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "E_AkIPVsZeBL"
   },
   "outputs": [],
   "source": [
    "kdd_test_v2_probe = pd.concat([kdd_test_probe.iloc[0:14000,:], kdd_test_probe.iloc[34001:64759,:]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 162
    },
    "id": "firflwnBY1yR",
    "outputId": "4e4ce367-7a7d-4b8f-b75c-37192b183fa2"
   },
   "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>Duration</th>\n",
       "      <th>src_bytes</th>\n",
       "      <th>dst_bytes</th>\n",
       "      <th>land</th>\n",
       "      <th>wrong_fragment</th>\n",
       "      <th>urgent</th>\n",
       "      <th>hot</th>\n",
       "      <th>num_failed</th>\n",
       "      <th>logged_in</th>\n",
       "      <th>num_compromised</th>\n",
       "      <th>root_shell</th>\n",
       "      <th>su_attempted</th>\n",
       "      <th>num_root</th>\n",
       "      <th>num_file_creations</th>\n",
       "      <th>num_shells</th>\n",
       "      <th>num_access_files</th>\n",
       "      <th>num_outbound_cmds</th>\n",
       "      <th>is_host_login</th>\n",
       "      <th>is_guest_login</th>\n",
       "      <th>count</th>\n",
       "      <th>srv_count</th>\n",
       "      <th>serror_rate</th>\n",
       "      <th>srv_error_rate</th>\n",
       "      <th>rerror_rate</th>\n",
       "      <th>srv_rerror_rate</th>\n",
       "      <th>same_srv_rate</th>\n",
       "      <th>diff_srv_rate</th>\n",
       "      <th>srv_diff_host_rate</th>\n",
       "      <th>dst_host_count</th>\n",
       "      <th>dst_host_srv_count</th>\n",
       "      <th>dst_host_same_srv_rate</th>\n",
       "      <th>dst_host_diff_srv_rate</th>\n",
       "      <th>dst_host_same_src_port_rate</th>\n",
       "      <th>dst_host_srv_diff_host_rate</th>\n",
       "      <th>dst_host_serror_rate</th>\n",
       "      <th>dst_host_srv_serror_rate</th>\n",
       "      <th>dst_host_rerror_rate</th>\n",
       "      <th>dst_host_srv_rerror_rate</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Type_Groups</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "      <td>41836</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "      <td>2922</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Duration  ...  dst_host_srv_rerror_rate\n",
       "Type_Groups            ...                          \n",
       "0               41836  ...                     41836\n",
       "1                2922  ...                      2922\n",
       "\n",
       "[2 rows x 38 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd_test_v2_probe.groupby('Type_Groups').count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "iKgW7zMsecSp"
   },
   "source": [
    "Now that we obtained Train, Test and Validation sets, we can train our data and optimize the reconstruction error threshold using test set. Then, we will perform our actual prediction task by using Validation set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "zZF3-B2GzG4H",
    "outputId": "bff13cfb-d57d-40fa-ef95-29ae77a536e4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(97278, 38)"
      ]
     },
     "execution_count": 17,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train, X_test = kdd_train_probe, kdd_test_v2_probe\n",
    "X_train = X_train[X_train.Type_Groups == 0]\n",
    "X_train = X_train.drop(['Type_Groups'], axis=1)\n",
    "y_test = X_test['Type_Groups']\n",
    "X_test = X_test.drop(['Type_Groups'], axis=1)\n",
    "X_train = X_train.values\n",
    "X_test = X_test.values\n",
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ddmJ-5CSpPrw"
   },
   "outputs": [],
   "source": [
    "input_dim = X_train.shape[1]\n",
    "encoding_dim = 14\n",
    "input_layer = Input(shape=(input_dim, ))\n",
    "encoder = Dense(encoding_dim, activation=\"tanh\", \n",
    "                activity_regularizer=regularizers.l1(10e-5))(input_layer)\n",
    "encoder = Dense(int(encoding_dim / 2), activation=\"relu\")(encoder)\n",
    "decoder = Dense(int(encoding_dim), activation='tanh')(encoder)\n",
    "decoder = Dense(input_dim, activation='relu')(decoder)\n",
    "\n",
    "autoencoder_corr = Model(inputs=input_layer, outputs=decoder)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "5CWGL4-b2umu",
    "outputId": "91d6bc5b-7fa4-4eaf-bb7a-c235171a66a4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "973/973 [==============================] - 4s 3ms/step - loss: 0.4853 - accuracy: 0.4874 - val_loss: 0.6084 - val_accuracy: 0.6571\n",
      "Epoch 2/5\n",
      "973/973 [==============================] - 2s 2ms/step - loss: 0.2893 - accuracy: 0.8466 - val_loss: 0.5704 - val_accuracy: 0.6501\n",
      "Epoch 3/5\n",
      "973/973 [==============================] - 2s 2ms/step - loss: 0.2898 - accuracy: 0.8700 - val_loss: 0.5663 - val_accuracy: 0.6442\n",
      "Epoch 4/5\n",
      "973/973 [==============================] - 2s 2ms/step - loss: 0.2627 - accuracy: 0.8628 - val_loss: 0.5470 - val_accuracy: 0.5853\n",
      "Epoch 5/5\n",
      "973/973 [==============================] - 2s 2ms/step - loss: 0.2615 - accuracy: 0.8545 - val_loss: 0.5453 - val_accuracy: 0.6316\n"
     ]
    }
   ],
   "source": [
    "nb_epoch = 5\n",
    "batch_size = 100\n",
    "batch_index=0\n",
    "autoencoder_corr.compile(optimizer='adam', \n",
    "                    loss='mean_squared_error', \n",
    "                    metrics=['accuracy'])\n",
    "checkpointer = ModelCheckpoint(filepath=\"model.h5\",  #TO SAVE THE MODEL\n",
    "                               verbose=0,\n",
    "                               save_best_only=True)\n",
    "tensorboard = TensorBoard(log_dir='./logs',\n",
    "                          histogram_freq=0,\n",
    "                          write_graph=True,\n",
    "                          write_images=True)\n",
    "history = autoencoder_corr.fit(X_train, X_train, #INPUT AND EXPECTED OUTPUT ARE THE SAME.\n",
    "                    epochs=nb_epoch,\n",
    "                    batch_size=batch_size,\n",
    "                    shuffle=True,\n",
    "                    validation_data=(X_test, X_test),\n",
    "                    verbose=1,\n",
    "                    callbacks=[checkpointer, tensorboard]).history"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "p1wb7_MsBO05"
   },
   "source": [
    "# Tune the threshold value using test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "id": "v_WijHwGBOjs",
    "outputId": "78964106-971b-4815-fa96-859d7bcac498"
   },
   "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>reconstruction_error</th>\n",
       "      <th>true_class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>102563</th>\n",
       "      <td>0.065080</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102564</th>\n",
       "      <td>0.065080</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102565</th>\n",
       "      <td>0.065080</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102569</th>\n",
       "      <td>0.653062</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102570</th>\n",
       "      <td>0.064486</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        reconstruction_error  true_class\n",
       "102563              0.065080           0\n",
       "102564              0.065080           0\n",
       "102565              0.065080           0\n",
       "102569              0.653062           0\n",
       "102570              0.064486           0"
      ]
     },
     "execution_count": 21,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions = autoencoder_corr.predict(X_test)\n",
    "mse = np.mean(np.power(X_test - predictions, 2), axis=1)\n",
    "error_df = pd.DataFrame({'reconstruction_error': mse,\n",
    "                        'true_class': y_test})\n",
    "error_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "NWhNe_P7BOVz"
   },
   "outputs": [],
   "source": [
    "threshold = np.arange(0, 1, 0.01)\n",
    "results = np.zeros(100)\n",
    "probe = np.zeros(100)\n",
    "i = 0\n",
    "for t in threshold:\n",
    "  y_pred = [1 if e > t else 0 for e in error_df.reconstruction_error.values]\n",
    "  conf_matrix = confusion_matrix(error_df.true_class, y_pred)\n",
    "  results[i] = (conf_matrix[0,0] + conf_matrix[1,1]) / 44758 #64759\n",
    "  probe[i] = conf_matrix[1,1] / 2922 #4166\n",
    "  i = i + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "id": "6ZxbaJfHG24M",
    "outputId": "ab061fdf-1369-4f48-d1ca-9e6c83168e5e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7fbd147906a0>]"
      ]
     },
     "execution_count": 44,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(threshold, probe)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "gU-cutBesUna",
    "outputId": "ff7a7478-7cf8-4031-beb4-bb1e9cc6f952"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 1.        ,\n",
       "       1.        , 1.        , 1.        , 1.        , 0.99931554,\n",
       "       0.99931554, 0.99931554, 0.99931554, 0.99931554, 0.99897331,\n",
       "       0.99897331, 0.99828884, 0.99828884, 0.99828884, 0.99794661,\n",
       "       0.99760438, 0.99726215, 0.99691992, 0.99691992, 0.99657769,\n",
       "       0.99486653, 0.9945243 , 0.99315537, 0.99212868, 0.99075975,\n",
       "       0.99075975, 0.99007529, 0.98973306, 0.9890486 , 0.9890486 ,\n",
       "       0.98870637, 0.98836413, 0.98733744, 0.98631075, 0.98631075,\n",
       "       0.98562628, 0.98494182, 0.98391513, 0.98391513, 0.98323066,\n",
       "       0.9825462 , 0.98220397, 0.98186174, 0.98117728, 0.98117728,\n",
       "       0.98083504, 0.98015058, 0.98015058, 0.97912389, 0.97843943,\n",
       "       0.97843943, 0.97741273, 0.97672827, 0.97604381, 0.97604381,\n",
       "       0.97570157, 0.97535934, 0.97501711, 0.97433265, 0.97433265])"
      ]
     },
     "execution_count": 23,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "probe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "zm9H6m_-sWD8"
   },
   "outputs": [],
   "source": [
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "UlWYF4Cf3XLL",
    "outputId": "ddd7c227-9449-4a15-a6de-8d1270409b9d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[36569,  5267],\n",
       "       [    0,  2922]])"
      ]
     },
     "execution_count": 39,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "threshold = 0.13 # 0.20 de iyi\n",
    "LABELS = [\"Normal\", \"Probing\"]\n",
    "y_pred = [1 if e > threshold else 0 for e in error_df.reconstruction_error.values]\n",
    "conf_matrix = confusion_matrix(error_df.true_class, y_pred)\n",
    "#plt.figure(figsize=(12, 12))\n",
    "#sns.heatmap(conf_matrix, xticklabels=LABELS, yticklabels=LABELS, annot=True, fmt=\"d\");\n",
    "#plt.title(\"Confusion matrix\")\n",
    "#plt.ylabel('True class')\n",
    "#plt.xlabel('Predicted class')\n",
    "#plt.show()\n",
    "conf_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "id": "wJEkAmeC3v1v"
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import (confusion_matrix, precision_recall_curve, auc,\n",
    "                             roc_curve, recall_score, classification_report, f1_score,\n",
    "                             precision_recall_fscore_support)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "oyHacenx3VJu",
    "outputId": "9fcfdfc2-f907-40d3-bb4f-917f4eca04ad"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fpr, tpr, thresholds = roc_curve(error_df.true_class, error_df.reconstruction_error)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr, tpr, label='AUC = %0.4f'% roc_auc)\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.001, 1])\n",
    "plt.ylim([0, 1.001])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "id": "pS90TZJcHGGD",
    "outputId": "ac5ddb99-b422-4fc7-d15b-397b55499eff"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7fbd146fb160>]"
      ]
     },
     "execution_count": 45,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(threshold, results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rXN_FzmBjUfP"
   },
   "source": [
    "Let's try if we can find a better threshold value between 0.14 and 0.18."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "8kLPqf61BAf6"
   },
   "outputs": [],
   "source": [
    "threshold = np.arange(0.3, 0.5, 0.001)\n",
    "results = np.zeros(200)\n",
    "probe = np.zeros(200)\n",
    "i = 0\n",
    "for t in threshold:\n",
    "  y_pred = [1 if e > t else 0 for e in error_df.reconstruction_error.values]\n",
    "  conf_matrix = confusion_matrix(error_df.true_class, y_pred)\n",
    "  results[i] = (conf_matrix[0,0] + conf_matrix[1,1]) / 44758\n",
    "  probe[i] = conf_matrix[1,1] / 2922\n",
    "  i = i + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "zI558plHkG08"
   },
   "outputs": [],
   "source": [
    "threshold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Cj9GcA5gkUrH"
   },
   "outputs": [],
   "source": [
    "probe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "cmJ2X6z7klyj"
   },
   "outputs": [],
   "source": [
    "results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vydMN6zBmjDq"
   },
   "source": [
    "Let's fix our threshold value to 0.499 and apply it to Validation Set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "JvofCF2WmnyC"
   },
   "outputs": [],
   "source": [
    "X_valid = kdd_valid_probe\n",
    "y_valid = X_valid['Type_Groups']\n",
    "X_valid = X_valid.drop(['Type_Groups'], axis=1)\n",
    "X_valid = X_valid.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 297
    },
    "id": "zc_9M1VNm38D",
    "outputId": "eca32605-5914-4e72-d1e7-7bcad0cdf0f9"
   },
   "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>reconstruction_error</th>\n",
       "      <th>true_class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>20000.000000</td>\n",
       "      <td>20000.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.553160</td>\n",
       "      <td>0.062200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>19.441038</td>\n",
       "      <td>0.241524</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.019957</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.041338</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.086328</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.131369</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1940.447577</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       reconstruction_error    true_class\n",
       "count          20000.000000  20000.000000\n",
       "mean               0.553160      0.062200\n",
       "std               19.441038      0.241524\n",
       "min                0.019957      0.000000\n",
       "25%                0.041338      0.000000\n",
       "50%                0.086328      0.000000\n",
       "75%                0.131369      0.000000\n",
       "max             1940.447577      1.000000"
      ]
     },
     "execution_count": 112,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions = autoencoder_corr.predict(X_valid)\n",
    "mse = np.mean(np.power(X_valid - predictions, 2), axis=1)\n",
    "error_df = pd.DataFrame({'reconstruction_error': mse,\n",
    "                        'true_class': y_valid})\n",
    "error_df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "id": "Tb9UyZXmnA03",
    "outputId": "e0e6f281-8b93-43f1-e1f7-3e4c57f4d3aa"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[18290,   466],\n",
       "       [   94,  1150]])"
      ]
     },
     "execution_count": 117,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "threshold = 0.499 # 0.20 de iyi\n",
    "LABELS = [\"Normal\", \"Probing\"]\n",
    "y_pred = [1 if e > threshold else 0 for e in error_df.reconstruction_error.values]\n",
    "conf_matrix = confusion_matrix(error_df.true_class, y_pred)\n",
    "conf_matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "2a7efGoZLk1O"
   },
   "source": [
    "## XGBoosting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "tzb8DuA_7GDw"
   },
   "outputs": [
    {
     "ename": "ModuleNotFoundError",
     "evalue": "No module named 'imblearn'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)",
      "\u001b[0;32m/var/folders/6j/xfjz279j5w73v97s3944dbj80000gn/T/ipykernel_5422/3686472688.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0msklearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel_selection\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mGridSearchCV\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0msklearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mneural_network\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mMLPClassifier\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 12\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0mimblearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpipeline\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mmake_pipeline\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mmake_pipeline_imb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     13\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mimblearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mover_sampling\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mSMOTE\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     14\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mimblearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmetrics\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mclassification_report_imbalanced\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mModuleNotFoundError\u001b[0m: No module named 'imblearn'"
     ]
    }
   ],
   "source": [
    "from numpy import loadtxt\n",
    "from sklearn.metrics import accuracy_score\n",
    "import os\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import f1_score, roc_auc_score\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from imblearn.pipeline import make_pipeline as make_pipeline_imb\n",
    "from imblearn.over_sampling import SMOTE\n",
    "from imblearn.metrics import classification_report_imbalanced\n",
    "import xgboost as xgb\n",
    "from xgboost import XGBClassifier\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from collections import Counter\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 226
    },
    "id": "7wFt0NEX6gWA",
    "outputId": "09032ed4-32a6-49d4-e699-a76a0feabbe7"
   },
   "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>Duration</th>\n",
       "      <th>src_bytes</th>\n",
       "      <th>dst_bytes</th>\n",
       "      <th>land</th>\n",
       "      <th>wrong_fragment</th>\n",
       "      <th>urgent</th>\n",
       "      <th>hot</th>\n",
       "      <th>num_failed</th>\n",
       "      <th>logged_in</th>\n",
       "      <th>num_compromised</th>\n",
       "      <th>root_shell</th>\n",
       "      <th>su_attempted</th>\n",
       "      <th>num_root</th>\n",
       "      <th>num_file_creations</th>\n",
       "      <th>num_shells</th>\n",
       "      <th>num_access_files</th>\n",
       "      <th>num_outbound_cmds</th>\n",
       "      <th>is_host_login</th>\n",
       "      <th>is_guest_login</th>\n",
       "      <th>count</th>\n",
       "      <th>srv_count</th>\n",
       "      <th>serror_rate</th>\n",
       "      <th>srv_error_rate</th>\n",
       "      <th>rerror_rate</th>\n",
       "      <th>srv_rerror_rate</th>\n",
       "      <th>same_srv_rate</th>\n",
       "      <th>diff_srv_rate</th>\n",
       "      <th>srv_diff_host_rate</th>\n",
       "      <th>dst_host_count</th>\n",
       "      <th>dst_host_srv_count</th>\n",
       "      <th>dst_host_same_srv_rate</th>\n",
       "      <th>dst_host_diff_srv_rate</th>\n",
       "      <th>dst_host_same_src_port_rate</th>\n",
       "      <th>dst_host_srv_diff_host_rate</th>\n",
       "      <th>dst_host_serror_rate</th>\n",
       "      <th>dst_host_srv_serror_rate</th>\n",
       "      <th>dst_host_rerror_rate</th>\n",
       "      <th>dst_host_srv_rerror_rate</th>\n",
       "      <th>Type_Groups</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.150103</td>\n",
       "      <td>-0.004917</td>\n",
       "      <td>0.019931</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.005644</td>\n",
       "      <td>-0.073472</td>\n",
       "      <td>-0.021473</td>\n",
       "      <td>0.666223</td>\n",
       "      <td>-0.007208</td>\n",
       "      <td>-0.023163</td>\n",
       "      <td>-0.010262</td>\n",
       "      <td>-0.012379</td>\n",
       "      <td>-0.024657</td>\n",
       "      <td>-0.021773</td>\n",
       "      <td>-0.060732</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.081998</td>\n",
       "      <td>-0.112107</td>\n",
       "      <td>-0.125503</td>\n",
       "      <td>-0.093846</td>\n",
       "      <td>-0.078612</td>\n",
       "      <td>-0.291137</td>\n",
       "      <td>-0.290134</td>\n",
       "      <td>0.198961</td>\n",
       "      <td>-0.201086</td>\n",
       "      <td>-0.478157</td>\n",
       "      <td>-1.336229</td>\n",
       "      <td>-1.990255</td>\n",
       "      <td>0.536542</td>\n",
       "      <td>-0.34644</td>\n",
       "      <td>-0.154032</td>\n",
       "      <td>-0.365965</td>\n",
       "      <td>-0.106951</td>\n",
       "      <td>-0.072524</td>\n",
       "      <td>-0.301727</td>\n",
       "      <td>-0.29675</td>\n",
       "      <td>Normal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.150103</td>\n",
       "      <td>-0.004890</td>\n",
       "      <td>-0.048616</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.005644</td>\n",
       "      <td>-0.073472</td>\n",
       "      <td>-0.021473</td>\n",
       "      <td>0.666223</td>\n",
       "      <td>-0.007208</td>\n",
       "      <td>-0.023163</td>\n",
       "      <td>-0.010262</td>\n",
       "      <td>-0.012379</td>\n",
       "      <td>-0.024657</td>\n",
       "      <td>-0.021773</td>\n",
       "      <td>-0.060732</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.081998</td>\n",
       "      <td>-0.112107</td>\n",
       "      <td>-0.125503</td>\n",
       "      <td>-0.093846</td>\n",
       "      <td>-0.078612</td>\n",
       "      <td>-0.291137</td>\n",
       "      <td>-0.290134</td>\n",
       "      <td>0.198961</td>\n",
       "      <td>-0.201086</td>\n",
       "      <td>-0.478157</td>\n",
       "      <td>-1.240464</td>\n",
       "      <td>-1.882796</td>\n",
       "      <td>0.536542</td>\n",
       "      <td>-0.34644</td>\n",
       "      <td>-0.346801</td>\n",
       "      <td>-0.365965</td>\n",
       "      <td>-0.106951</td>\n",
       "      <td>-0.072524</td>\n",
       "      <td>-0.301727</td>\n",
       "      <td>-0.29675</td>\n",
       "      <td>Normal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.150103</td>\n",
       "      <td>-0.004892</td>\n",
       "      <td>-0.036865</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.005644</td>\n",
       "      <td>-0.073472</td>\n",
       "      <td>-0.021473</td>\n",
       "      <td>0.666223</td>\n",
       "      <td>-0.007208</td>\n",
       "      <td>-0.023163</td>\n",
       "      <td>-0.010262</td>\n",
       "      <td>-0.012379</td>\n",
       "      <td>-0.024657</td>\n",
       "      <td>-0.021773</td>\n",
       "      <td>-0.060732</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.081998</td>\n",
       "      <td>-0.112107</td>\n",
       "      <td>-0.125503</td>\n",
       "      <td>-0.093846</td>\n",
       "      <td>-0.078612</td>\n",
       "      <td>-0.291137</td>\n",
       "      <td>-0.290134</td>\n",
       "      <td>0.198961</td>\n",
       "      <td>-0.201086</td>\n",
       "      <td>-0.478157</td>\n",
       "      <td>-1.144700</td>\n",
       "      <td>-1.775337</td>\n",
       "      <td>0.536542</td>\n",
       "      <td>-0.34644</td>\n",
       "      <td>-0.411057</td>\n",
       "      <td>-0.365965</td>\n",
       "      <td>-0.106951</td>\n",
       "      <td>-0.072524</td>\n",
       "      <td>-0.301727</td>\n",
       "      <td>-0.29675</td>\n",
       "      <td>Normal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.150103</td>\n",
       "      <td>-0.004899</td>\n",
       "      <td>-0.036865</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.005644</td>\n",
       "      <td>-0.073472</td>\n",
       "      <td>-0.021473</td>\n",
       "      <td>0.666223</td>\n",
       "      <td>-0.007208</td>\n",
       "      <td>-0.023163</td>\n",
       "      <td>-0.010262</td>\n",
       "      <td>-0.012379</td>\n",
       "      <td>-0.024657</td>\n",
       "      <td>-0.021773</td>\n",
       "      <td>-0.060732</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.081998</td>\n",
       "      <td>-0.145850</td>\n",
       "      <td>-0.218540</td>\n",
       "      <td>-0.093846</td>\n",
       "      <td>-0.078612</td>\n",
       "      <td>-0.291137</td>\n",
       "      <td>-0.290134</td>\n",
       "      <td>0.198961</td>\n",
       "      <td>-0.201086</td>\n",
       "      <td>-0.478157</td>\n",
       "      <td>-1.048936</td>\n",
       "      <td>-1.667879</td>\n",
       "      <td>0.536542</td>\n",
       "      <td>-0.34644</td>\n",
       "      <td>-0.411057</td>\n",
       "      <td>-0.365965</td>\n",
       "      <td>-0.106951</td>\n",
       "      <td>-0.072524</td>\n",
       "      <td>-0.301727</td>\n",
       "      <td>-0.29675</td>\n",
       "      <td>Normal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.150103</td>\n",
       "      <td>-0.004900</td>\n",
       "      <td>-0.027268</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.005644</td>\n",
       "      <td>-0.073472</td>\n",
       "      <td>-0.021473</td>\n",
       "      <td>0.666223</td>\n",
       "      <td>-0.007208</td>\n",
       "      <td>-0.023163</td>\n",
       "      <td>-0.010262</td>\n",
       "      <td>-0.012379</td>\n",
       "      <td>-0.024657</td>\n",
       "      <td>-0.021773</td>\n",
       "      <td>-0.060732</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.081998</td>\n",
       "      <td>-0.145850</td>\n",
       "      <td>-0.218540</td>\n",
       "      <td>-0.093846</td>\n",
       "      <td>-0.078612</td>\n",
       "      <td>-0.291137</td>\n",
       "      <td>-0.290134</td>\n",
       "      <td>0.198961</td>\n",
       "      <td>-0.201086</td>\n",
       "      <td>-0.478157</td>\n",
       "      <td>-0.953172</td>\n",
       "      <td>-1.560420</td>\n",
       "      <td>0.536542</td>\n",
       "      <td>-0.34644</td>\n",
       "      <td>-0.443185</td>\n",
       "      <td>-0.365965</td>\n",
       "      <td>-0.106951</td>\n",
       "      <td>-0.072524</td>\n",
       "      <td>-0.301727</td>\n",
       "      <td>-0.29675</td>\n",
       "      <td>Normal</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Duration  src_bytes  ...  dst_host_srv_rerror_rate  Type_Groups\n",
       "0 -0.150103  -0.004917  ...                  -0.29675       Normal\n",
       "1 -0.150103  -0.004890  ...                  -0.29675       Normal\n",
       "2 -0.150103  -0.004892  ...                  -0.29675       Normal\n",
       "3 -0.150103  -0.004899  ...                  -0.29675       Normal\n",
       "4 -0.150103  -0.004900  ...                  -0.29675       Normal\n",
       "\n",
       "[5 rows x 39 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd_train_probe.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "id": "Dj2ak4hnLmbq"
   },
   "outputs": [],
   "source": [
    "def xgb_classifier(X_train, X_test, y_train, y_test, useTrainCV=True, cv_folds=5, \n",
    "                   early_stopping_rounds=50):\n",
    "    \"\"\"\n",
    "    # {'learning_rate': 0.1, 'max_depth': 5, 'min_child_weight': 3} 0.862920874517388\n",
    "    # {'colsample_bytree': 1.0, 'gamma': 0.2} 0.871\n",
    "    # {'gamma': 0.2, 'scale_pos_weight': 1} 0.8702009952422571\n",
    "    # {'subsample': 0.6} 0.864310306628855\n",
    "    \"\"\"\n",
    "    alg = XGBClassifier(learning_rate=0.1, \n",
    "                        n_estimators=140, \n",
    "                        max_depth=5, # maximum number of features used in each tree\n",
    "                        min_child_weight=3, \n",
    "                        gamma=0.2, \n",
    "                        subsample=0.6, #hich is for each tree the % of rows taken to build the tree. \n",
    "                                        #I recommend not taking out too many rows, as performance will drop a lot. Take values from 0.8 to 1.\n",
    "                        colsample_bytree=0.6, # number of columns used by each tree. \n",
    "                        #In order to avoid some columns to take too much credit for the prediction \n",
    "                        #(think of it like in recommender systems when you recommend the most purchased products\n",
    "                        #and forget about the long tail), take out a good proportion of columns. \n",
    "                        #Values from 0.3 to 0.8 if you have many columns (especially if you did one-hot encoding), \n",
    "                        #or 0.8 to 1 if you only have a few columns.\n",
    "                        objective='binary:logistic', \n",
    "                        nthread=4, \n",
    "                        scale_pos_weight=20, \n",
    "                        seed=27)\n",
    "\n",
    "    if useTrainCV:\n",
    "        print(\"Start Feeding Data\")\n",
    "        xgb_param = alg.get_xgb_params()\n",
    "        xgtrain = xgb.DMatrix(X_train.values, label=y_train.values)\n",
    "        # xgtest = xgb.DMatrix(X_test.values, label=y_test.values)\n",
    "        cvresult = xgb.cv(xgb_param, xgtrain, num_boost_round=alg.get_params()['n_estimators'], nfold=cv_folds,\n",
    "                          early_stopping_rounds=early_stopping_rounds)\n",
    "        alg.set_params(n_estimators=cvresult.shape[0])\n",
    "\n",
    "    print('Start Training')\n",
    "    alg.fit(X_train, y_train, eval_metric='auc')\n",
    "\n",
    "    # param_test1 = {}\n",
    "    # gsearch1 = GridSearchCV(estimator=XGBClassifier(learning_rate=0.1, n_estimators=140, max_depth=5,\n",
    "    #                                                 min_child_weight=3, gamma=0.2, subsample=0.8,\n",
    "    #                                                 colsample_bytree=1.0,\n",
    "    #                                                 objective='binary:logistic', nthread=4, scale_pos_weight=1,\n",
    "    #                                                 seed=27),\n",
    "    #                         param_grid=param_test1,\n",
    "    #                         scoring='f1',\n",
    "    #                         n_jobs=4, iid=False, cv=5)\n",
    "    # gsearch1.fit(X_train, y_train)\n",
    "    # print(gsearch1.cv_results_, gsearch1.best_params_, gsearch1.best_score_)\n",
    "\n",
    "    print(\"Start Predicting\")\n",
    "    predictions = alg.predict(X_test)\n",
    "    pred_proba = alg.predict_proba(X_test)[:, 1]\n",
    "\n",
    "    print(\"\\nResulting Metrics\")\n",
    "    print(\"Accuracy : %.4g\" % metrics.accuracy_score(y_test, predictions))\n",
    "    print(\"AUC : %f\" % metrics.roc_auc_score(y_test, pred_proba))\n",
    "    print(\"F1 Score : %f\" % metrics.f1_score(y_test, predictions))\n",
    "\n",
    "    feat_imp = alg.feature_importances_\n",
    "    feat = X_train.columns.tolist()\n",
    "    # clf.best_estimator_.booster().get_fscore()\n",
    "    res_df = pd.DataFrame({'Features': feat, 'Importance': feat_imp}).sort_values(by='Importance', ascending=False)\n",
    "    res_df.plot('Features', 'Importance', kind='bar', title='Feature Importances')\n",
    "    plt.ylabel('Feature Importance Score')\n",
    "    plt.show()\n",
    "    print(res_df)\n",
    "    print(res_df[\"Features\"].tolist())\n",
    "    return alg\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "MzD0O6KV2KHP"
   },
   "outputs": [],
   "source": [
    "kdd_train_probe = kdd_train[(kdd_train.Type_Groups == 'Normal') | (kdd_train.Type_Groups == 'Probe')]\n",
    "kdd_test_probe = kdd_test[(kdd_test.Type_Groups == 'Normal') | (kdd_test.Type_Groups == 'Probe')]\n",
    "\n",
    "kdd_train_probe['Type_Groups'] = np.where(kdd_train_probe['Type_Groups'] == 'Normal', 0, 1)\n",
    "kdd_test_probe['Type_Groups'] = np.where(kdd_test_probe['Type_Groups'] == 'Normal', 0, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "id": "B4gndvi-6FsL"
   },
   "outputs": [],
   "source": [
    "y_train = kdd_train_probe['Type_Groups']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "F_gWM69b6vdw"
   },
   "outputs": [],
   "source": [
    "X_train = kdd_train_probe.drop(['Type_Groups'], axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "id": "VX90qotY6z-L"
   },
   "outputs": [],
   "source": [
    "y_test = kdd_test_probe['Type_Groups']\n",
    "X_test = kdd_test_probe.drop(['Type_Groups'], axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "1sikZut16_gj",
    "outputId": "78f75f65-8081-4188-d2f5-de2b63101521"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start Feeding Data\n",
      "Start Training\n",
      "Start Predicting\n",
      "\n",
      "Resulting Metrics\n",
      "Accuracy : 0.9614\n",
      "AUC : 0.988588\n",
      "F1 Score : 0.754472\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAGdCAYAAADnmo8wAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOydd7gdVdX/P9+EQKghgYhISEIwgCAtJlFBioWiIvi+IkVUQJAXkeKLDRtN359gARERRWkiSFMxIgpIU4oSQgomgCQhQLBTQugJWb8/1j73zp3MOWefm3uSw836PM88Z2bXtWfmzG5rry0zIwiCIAiqGLCiBQiCIAg6l6gkgiAIgrpEJREEQRDUJSqJIAiCoC5RSQRBEAR1iUoiCIIgqEtUEkEQBEFdopIIljuS5kl6QdKzheN1fZDmu/pKxoz8Tpb00+WVXyMkHSLp9hUtR9A/iUoiWFG8z8zWKhx/W5HCSFplRebfW16tcgevHqKSCDoGSUMknS/p75Iel/Q1SQOT36aSbpb0hKT/SLpU0rrJ7xJgJPDr1Cv5nKRdJc0vpd/V20g9gasl/VTSM8AhjfLPkN0kHSXpIUkLJX01yXynpGckXSlp1RR2V0nzJX0xlWWepINK9+Enkv4t6RFJX5Y0IPkdIukOSWdKegK4AvgB8NZU9qdTuPdKmpryfkzSyYX0Ryd5D5b0aJLhSwX/gUm2OaksUyRtnPy2kHSjpCclPShpv0K890ialeI8Lukz2Q8/6Fiikgg6iYuAxcDrge2B3YHDk5+ArwOvA94AbAycDGBmHwEepbt38o3M/PYBrgbWBS5tkn8OewBvAt4CfA44D/hwkvWNwIGFsK8F1gc2Ag4GzpO0efI7GxgCjAF2AT4KHFqI+2ZgLrBBSv9I4K5U9nVTmOdSvHWB9wKfkPT+krxvAzYH3gmcKOkNyf34JOt7gHWAjwHPS1oTuBG4DHgNcADwfUlbpnjnA/9jZmun8t6cddeCjiYqiWBFcY2kp9NxjaQN8I/Sp8zsOTP7F3Am/iHCzGab2Y1m9pKZ/Rs4A/+ALgt3mdk1ZrYE/xjWzT+Tb5jZM2Y2E/gLcIOZzTWzBcBv8YqnyFdSeW4DfgPsl3ouBwBfMLOFZjYP+DbwkUK8v5nZ2Wa22MxeqBLEzG41s/vMbImZzQB+xtL36xQze8HMpgPTgW2T++HAl83sQXOmm9kTwF7APDO7MOU9Ffg58MEUbxGwpaR1zOwpM7u3hXsXdCgxnhmsKN5vZr+vXUiaCAwC/i6p5jwAeCz5bwCcBewErJ38nlpGGR4rnI9qlH8m/yycv1Bx/drC9VNm9lzh+hG8l7R+kuORkt9GdeSuRNKbgdPwFv2qwGrAVaVg/yicPw+slc43BuZUJDsKeHNtSCuxCnBJOv8A8GXgNEkzgBPM7K5msgadTfQkgk7hMeAlYH0zWzcd65jZVsn//wEGbG1m6+DDLCrEL5szfg5Yo3aRWujDS2GKcZrl39cMTcM3NUYCfwP+g7fIR5X8Hq8jd9U1+JDQJGBjMxuCz1uoIlwVjwGb1nG/rXB/1k1DXJ8AMLPJZrYPPhR1DXBlZn5BBxOVRNARmNnfgRuAb0taR9KANPFbGyJZG3gWWCBpI+CzpST+iY/h1/grMDhN4A7CW7irLUP+7eAUSatK2gkfyrnKzF7BP67/J2ltSaPwOYJG6rb/BEbUJsYTawNPmtmLqZf2oRbk+jHwVUlj5WwjaT3gWmAzSR+RNCgdEyS9IZXjIElDzGwR8AywpIU8gw4lKomgk/goPjQyCx9KuhrYMPmdAowDFuDj978oxf068OU0x/GZNA9wFP7BexzvWcynMY3y72v+kfL4Gz5pfqSZPZD8jsHlnQvcjvcKLmiQ1s3ATOAfkv6T3I4CTpW0EDiR1lr1Z6TwN+Af+/OB1c1sIT6Zf0CS+x/A6XRXvh8B5iVtsSOBgwhe9Sg2HQqC5YukXYGfmtmIFS1LEDQjehJBEARBXaKSCIIgCOoSw01BEARBXaInEQRBENQlKokgCIKgLv1mxfX6669vo0ePXtFiBEEQvKqYMmXKf8ysvNC0i7ZWEpL2xE0pDAR+bGanlfyPBD4JvIIvlDrCzGYlvy8AhyW/Y83s+kZ5jR49mnvuuafvCxEEQdCPkfRII/+2DTclMwjnAO8GtgQOLFiLrHGZmW1tZtsB38AX8ZDCHQBsBeyJW5rMMtkcBEEQ9B3tnJOYCMxOVjBfBi7HTTN3YWbPFC7XpNsGzT7A5clC5sPA7JReEARBsBxp53DTRvS0Vjkft4PfA0mfxG3TrAq8oxD3T6W4GxEEQRAsV1b4xLWZnQOcI+lDuBG2g3PjSjoCOAJg5MiR7REwCIK2sGjRIubPn8+LL764okVZKRg8eDAjRoxg0KBBLcVrZyXxOG6XvsYIepo7LnM5cG4rcc3sPHz3L8aPHx+rAoPgVcT8+fNZe+21GT16NIU9PII2YGY88cQTzJ8/n0022aSluO2ck5gMjJW0STJhfABu374LSWMLl+8FHkrnk4ADJK0maRNgLHB3G2UNgmA58+KLL7LeeutFBbEckMR6663Xq15b23oSZrZY0tHA9bgK7AVmNlPSqcA9ZjYJOFq+Mf0i3GzywSnuTElX4iabFwOfTHb2gyDoR0QFsfzo7b1u65yEmV0HXFdyO7FwflyDuP8H/F/7pAuCYGVnrbXW4tlnn11u+c2bN48777yTD32olT2gViwrfOK6HYw+4Tc9rued9t4VJEkQBLmU/7fLSqf97xcvXsy8efO47LLLXlWVRNhuCoJgpefWW29ll112YZ999mHMmDGccMIJXHrppUycOJGtt96aOXPmAHDIIYdw5JFHMn78eDbbbDOuvfZawOdXDj30ULbeemu23357brnlFgAuuugi9t57b97xjnfwzne+kxNOOIE//vGPbLfddpx55pnMmzePnXbaiXHjxjFu3DjuvPPOLnl23XVX9t13X7bYYgsOOuggaha7J0+ezA477MC2227LxIkTWbhwIa+88gqf/exnmTBhAttssw0//OEP++ze9MueRBAEQatMnz6d+++/n2HDhjFmzBgOP/xw7r77bs466yzOPvtsvvOd7wA+ZHT33XczZ84c3v72tzN79mzOOeccJHHffffxwAMPsPvuu/PXv/4VgHvvvZcZM2YwbNgwbr31Vr71rW91VS7PP/88N954I4MHD+ahhx7iwAMP7DIvNHXqVGbOnMnrXvc6dtxxR+644w4mTpzI/vvvzxVXXMGECRN45plnWH311Tn//PMZMmQIkydP5qWXXmLHHXdk9913b1mTqYqoJIIgCIAJEyaw4Ya+pfmmm27K7rvvDsDWW2/d1TMA2G+//RgwYABjx45lzJgxPPDAA9x+++0cc8wxAGyxxRaMGjWqq5LYbbfdGDZsWGWeixYt4uijj2batGkMHDiwKw7AxIkTGTHCd7jdbrvtmDdvHkOGDGHDDTdkwoQJAKyzzjoA3HDDDcyYMYOrr74agAULFvDQQw9FJREEQdBXrLbaal3nAwYM6LoeMGAAixcv7vIrawk10xpac8016/qdeeaZbLDBBkyfPp0lS5YwePDgSnkGDhzYQ4YyZsbZZ5/NHnvs0VCW3hBzEkEQBC1w1VVXsWTJEubMmcPcuXPZfPPN2Wmnnbj00ksB+Otf/8qjjz7K5ptvvlTctddem4ULF3ZdL1iwgA033JABAwZwySWX8MorjTX9N998c/7+978zefJkABYuXMjixYvZY489OPfcc1m0aFGXDM8991yflDd6EkEQBC0wcuRIJk6cyDPPPMMPfvADBg8ezFFHHcUnPvEJtt56a1ZZZRUuuuiiHj2BGttssw0DBw5k22235ZBDDuGoo47iAx/4AD/5yU/Yc889G/Y6AFZddVWuuOIKjjnmGF544QVWX311fv/733P44Yczb948xo0bh5kxfPhwrrnmmj4pb7/Z43r8+PFWm/AJFdgg6Hzuv/9+3vCGN6xoMVrikEMOYa+99mLfffdd0aL0iqp7LmmKmY2vFyeGm4IgCIK6xHBTEARBJhdddNGKFmG5Ez2JIAiCoC5RSQRBsMLoL3OirwZ6e6+jkgiCYIUwePBgnnjiiagolgO1/SSK6zByiTmJIAhWCCNGjGD+/Pn8+9//XtGirBTUdqZrlagkgiBYIQwaNKhPzEYE7SWGm4IgCIK6RCURBEEQ1CUqiSAIgqAuUUkEQRAEdYlKIgiCIKhLVBJBEARBXaKSCIIgCOoSlUQQBEFQl6gkgiAIgrpEJREEQRDUJSqJIAiCoC5trSQk7SnpQUmzJZ1Q4X+8pFmSZki6SdKogt8rkqalY1I75QyCIAiqaZuBP0kDgXOA3YD5wGRJk8xsViHYVGC8mT0v6RPAN4D9k98LZrZdu+QLgiAImtPOnsREYLaZzTWzl4HLgX2KAczsFjN7Pl3+CWjdjm0QBEHQNtpZSWwEPFa4np/c6nEY8NvC9WBJ90j6k6T3t0PAIAiCoDEdsZ+EpA8D44FdCs6jzOxxSWOAmyXdZ2ZzSvGOAI4AGDly5HKTNwiCYGWhnT2Jx4GNC9cjklsPJL0L+BKwt5m9VHM3s8fT71zgVmD7clwzO8/MxpvZ+OHDh/et9EEQBEFbK4nJwFhJm0haFTgA6KGlJGl74Id4BfGvgvtQSaul8/WBHYHihHcQBEGwHGjbcJOZLZZ0NHA9MBC4wMxmSjoVuMfMJgHfBNYCrpIE8KiZ7Q28AfihpCV4RXZaSSsqCIIgWA60dU7CzK4Driu5nVg4f1edeHcCW7dTtiAIgqA52cNNktZopyBBEARB59G0kpC0g6RZwAPpeltJ32+7ZEEQBMEKJ6cncSawB/AEgJlNB3Zup1BBEARBZ5A13GRmj5WcXmmDLEEQBEGHkTNx/ZikHQCTNAg4Dri/vWIFQRAEnUBOT+JI4JO4SY3Hge3SdRAEQdDPadiTSJZczzKzg5aTPEEQBEEH0bAnYWavAKPSiukgCIJgJSNnTmIucEfa+Oe5mqOZndE2qYIgCIKOIKeSmJOOAcDa7RUnCIIg6CSaVhJmdgqApLXS9bPtFioIgiDoDHJWXL9R0lRgJjBT0hRJW7VftCAIgmBFk6MCex5wvJmNMrNRwKeBH7VXrCAIgqATyKkk1jSzW2oXZnYrsGbbJAqCIAg6hiztJklfAS5J1x/GNZ6CIAiCfk5OT+JjwHDgF8DPgfWTWxAEQdDPydFuego4djnIEgRBEHQYOdpNN0pat3A9VNL17RUrCIIg6ARyhpvWN7OnaxepZ/Ga9okUBEEQdAo5lcQSSSNrF5JGAdY+kYIgCIJOIUe76UvA7ZJuAwTsBBzRVqmCIAiCjiBn4vp3ksYBb8F7EJ8ys/+0XbIgCIJghVN3uEnSKElDAFKl8BywO/DRMB0eBEGwctBoTuJK0spqSdsBVwGPAtsC32+/aEEQBMGKptFw0+pm9rd0/mHgAjP7tqQBwLT2ixYEQRCsaBr1JFQ4fwdwE4CZLWmrREEQBEHH0KgncbOkK4G/A0OBmwEkbQi8vBxkC4IgCFYwjXoSn8LtNc0D3mZmi5L7a3G12KZI2lPSg5JmSzqhwv94SbMkzZB0U1qDUfM7WNJD6Tg4u0RBEARBn1G3J2FmBlxe4T41J2FJA4FzgN2A+cBkSZPMbFYh2FRgvJk9L+kTwDeA/SUNA04CxuNqt1NS3KcyyxUEQRD0ATkrrnvLRGC2mc01s5fxCmefYgAzu8XMnk+XfwJGpPM9gBvN7MlUMdwI7NlGWYMgCIIK2llJbAQ8Vrien9zqcRjw217GDYIgCNpAjlkOJK0OjDSzB9shhKQP40NLu7QY7wiSiZCRI0c2CR0EQRC0So6p8Pfh6yJ+l663kzQpI+3HgY0L1yOSWzn9d+ET4Xub2UutxDWz88xsvJmNHz58eIZIQRAEQSvkDDedjM8vPA1gZtOATTLiTQbGStokmfE4AOhRuUjaHvghXkH8q+B1PbB72rtiKG4OJPawCIIgWM7kDDctMrMFUnFtXXNT4Wa2WNLR+Md9IL5ie6akU4F7zGwS8E1gLeCqlP6jZra3mT0p6at4RQNwqpk9mV+sIAiCoC/IqSRmSvoQMFDSWHwr0ztzEjez64DrSm4nFs7f1SDuBcAFOfkEQRAE7SFnuOkYYCvgJeAyYAG+0C4IgiDo5+TsJ/E8PrGctco6CIIg6D/kaDfdKGndwvVQSTGJHARBsBKQM9y0vpk9XbtIK6Bf0z6RgiAIgk4hp5JYIqlrpVoywtdUuykIgiB49ZOj3fQl4HZJt+F7TOxEWuUcBEEQ9G9yJq5/J2kc8Jbk9Km053UQBEHQz8my3QSsBjyZwm8pCTP7Q/vECoIgCDqBppWEpNOB/YGZQG3rUgOikgiCIOjn5PQk3g9sXjC+FwRBEKwk5Gg3zQUGtVuQIAiCoPPI6Uk8D0yTdBNumgMAMzu2bVIFQRAEHUFOJTGJkonvIAiCYOUgRwX24uUhSBAEQdB55Gg3jQW+DmwJDK65m9mYNsoVBEEQdAA5E9cXAucCi4G3Az8BftpOoYIgCILOIKeSWN3MbgJkZo+Y2cnAe9srVhAEQdAJ5ExcvyRpAPBQ2o70cXzL0SAIgqCfk9OTOA5YA9+29E3Ah4GPtlOoIAiCoDPIqSRGm9mzZjbfzA41sw8AI5vGCoIgCF715FQSX8h0C4IgCPoZdeckJL0beA+wkaTvFrzWwTWdgiAIgn5Oo4nrvwH3AHsDUwruC4H/badQQRAEQWdQt5Iws+mS/gLsEauugyAIVk4azkmY2SvAxpJWXU7yBEEQBB1EzjqJh4E7JE0Cnqs5mtkZbZMqCIIg6AhytJvmANemsGsXjqZI2lPSg5JmSzqhwn9nSfdKWixp35LfK5KmpSOs0AZBEKwAcqzAngIgaa10/WxOwpIGAucAuwHzgcmSJpnZrEKwR4FDgM9UJPGCmW2Xk1cQBEHQHpr2JCS9UdJUfI/rmZKmSNoqI+2JwGwzm2tmLwOXA/sUA5jZPDObQffe2UEQBEEHkTPcdB5wvJmNMrNRwKeBH2XE2wh4rHA9P7nlMljSPZL+JOn9LcQLgiAI+oicies1zeyW2oWZ3SppzTbKVGOUmT0uaQxws6T7zGxOMYCkI4AjAEaODEshQRAEfU1OT2KupK9IGp2OLwNzM+I9DmxcuB6R3LIws8fT71zgVmD7ijDnmdl4Mxs/fPjw3KSDIAiCTHIqiY8Bw4FfpGN4cmvGZGCspE3SOosDyNwrW9JQSaul8/WBHYFZjWMFQRAEfU2OdtNTwLGShgBLzGxhTsJmtjjtP3E9MBC4wMxmSjoVuMfMJkmaAPwSGAq8T9IpZrYV8Abgh5KW4BXZaSWtqCAIgmA5kLPH9QTgAtLaCEkLgI+Z2ZSGEQEzuw64ruR2YuF8Mj4MVY53J7B1s/R7y+gTfrOU27zTYrO9IAiCMjkT1+cDR5nZHwEkvQ3f93qbdgoWBEEQrHhy5iReqVUQAGZ2O2EqPAiCYKUgpydxm6QfAj8DDNgfuFXSOAAzu7eN8gVBEAQrkJxKYtv0e1LJfXu80nhHn0oUBEEQdAw52k1vXx6CdBoxuR0EQZCn3bQu8FFgdDG8mR3bPrGCIAiCTiBnuOk64E/AfYQhviAIgpWKnEpisJkd33ZJgiAIgo4jRwX2Ekkfl7ShpGG1o+2SBUEQBCucnJ7Ey8A3gS/h2kyk3zHtEioIgiDoDHIqiU8Drzez/7RbmCAIgqCzyBlumg08325BgiAIgs4jpyfxHDBN0i3ASzXHUIENgiDo/+RUEtekIwiCIFjJyFlxffHyECQIgiDoPOpWEpLuo1ubaSnMLEyFB0EQ9HMa9ST2Wm5SBEEQBB1J3UrCzB5ZnoIEQRAEnUeOCmwQBEGwkhKVRBAEQVCXrEpC0uqSNm+3MEEQBEFn0bSSkPQ+YBrwu3S9naRJ7RYsCIIgWPHk9CROBiYCTwOY2TRgkzbKFARBEHQIOZXEIjNbUHKru34iCIIg6D/kmOWYKelDwEBJY4FjgTvbK1YQBEHQCeT0JI4BtsKN+10GLAA+1U6hgiAIgs6gYU9C0kDgN2b2dnzToSAIgmAlomFPwsxeAZZIGtKbxCXtKelBSbMlnVDhv7OkeyUtlrRvye9gSQ+l4+De5B8EQRAsGzlzEs8C90m6Ed9bAmi+n0TqhZwD7AbMByZLmmRmswrBHgUOAT5TijsMOAkYj0+ST0lxn8qQNwiCIOgjciqJX6SjVSYCs81sLoCky4F9gK5KwszmJb8lpbh7ADea2ZPJ/0ZgT+BnvZAjCIIg6CXt3E9iI+CxwvV84M3LEHejXsoRBEEQ9JKmlYSkh6lYF2FmY9oiUQtIOgI4AmDkyJErWJogCIL+R85w0/jC+WDgg8CwjHiPAxsXrkcktxweB3Ytxb21HMjMzgPOAxg/fnws8AuCIOhjmq6TMLMnCsfjZvYd4L0ZaU8GxkraRNKqwAFArs2n64HdJQ2VNBTYPbkFQRAEy5Gc4aZxhcsBeM8iZy5jsaSj8Y/7QOACM5sp6VTgHjObJGkC8EtgKPA+SaeY2VZm9qSkr+IVDcCptUnsIAiCYPmRM9z07cL5YuBhYL+cxM3sOuC6ktuJhfPJ+FBSVdwLgAty8gmCIAjaQ04lcVhNjbWGpLACGwRBsBKQY7vp6ky3IAiCoJ9RtychaQvcsN8QSf9d8FoH13IKgiAI+jmNhps2B/YC1gXeV3BfCHy8nUIFQRAEnUHdSsLMfgX8StJbzeyu5ShTEARB0CHkTFxPlfRJfOipa5jJzD7WNqmCIAiCjiBn4voS4LW40b3bcJXVhe0UKgiCIOgMciqJ15vZV4DnkrG/95JvqC8IgiB4FZNTSSxKv09LeiMwBHhN+0QKgiAIOoWcOYnzkv2kr+C2l9YCTmwcJQiCIOgP5Nhg+nE6vQ1Y4ebBgyAIguVH0+EmSRtIOl/Sb9P1lpIOa79oQRAEwYomZ07iItyS6+vS9V+BT7VLoCAIgqBzyKkk1jezK4El4CbAgVfaKlUQBEHQEeRUEs9JWo+0hamktwAL2ipVEARB0BHkaDcdj2s1bSrpDmA4sG9bpQqCIAg6gkZWYEea2aNmdq+kXXCDfwIeNLNF9eKtTIw+4Tc9ruedlrOraxAEwauHRsNN1xTOrzCzmWb2l6gggiAIVh4aVRIqnMf6iCAIgpWQRpWE1TkPgiAIVhIaTVxvK+kZvEexejonXZuZrdN26YIgCIIVSqNNhwYuT0GCIAiCziNHBTZYBkIDKgiCVzM5i+mCIAiClZSoJIIgCIK6RCURBEEQ1KWtlYSkPSU9KGm2pBMq/FeTdEXy/7Ok0cl9tKQXJE1Lxw/aKWcQBEFQTdsmriUNBM4BdgPmA5MlTTKzWYVghwFPmdnrJR0AnA7sn/zmmNl27ZIvCIIgaE47exITgdlmNtfMXgYuB/YphdkHuDidXw28U5IIgiAIOoJ2VhIbAY8Vrucnt8owaZ+KBcB6yW8TSVMl3SZppzbKGQRBENShU9dJ/B0YaWZPSHoTcI2krczsmWIgSUcARwCMHDlyBYgZBEHQv2lnT+JxYOPC9YjkVhlG0irAEOAJM3vJzJ4AMLMpwBxgs3IGZnaemY03s/HDhw9vQxGCIAhWbtrZk5gMjJW0CV4ZHAB8qBRmEnAwcBe+kdHNZmaShgNPmtkrksYAY4G5bZR1hRKrsoMg6FTaVkmY2WJJRwPXAwOBC8xspqRTgXvMbBJwPnCJpNnAk3hFArAzcKqkRfje2kea2ZPtkjUIgiCopq1zEmZ2HXBdye3EwvmLwAcr4v0c+Hk7ZQuCIAiaEyuugyAIgrpEJREEQRDUpVNVYIMSMbkdBMGKIHoSQRAEQV2iJ9FPKPc0IHobQRAsO9GTCIIgCOoSlUQQBEFQl6gkgiAIgrpEJREEQRDUJSqJIAiCoC5RSQRBEAR1iUoiCIIgqEtUEkEQBEFdopIIgiAI6hIrrlciYlV2EAStEj2JIAiCoC5RSQRBEAR1iUoiCIIgqEtUEkEQBEFdYuI66EFsbhQEQZGoJIKWiYokCFYeopII2kKziiTUcYPg1UFUEkHHEhVJEKx4opIIXtXkDH31RZiosIKVlagkgqCPiIok6I9EJREEy5GY9A9ebUQlEQQdRlQkQSfR1sV0kvaU9KCk2ZJOqPBfTdIVyf/PkkYX/L6Q3B+UtEc75QyCIAiqaVslIWkgcA7wbmBL4EBJW5aCHQY8ZWavB84ETk9xtwQOALYC9gS+n9ILgiAIliPtHG6aCMw2s7kAki4H9gFmFcLsA5yczq8GvidJyf1yM3sJeFjS7JTeXW2UNwheNfSFNlZMtAc5tHO4aSPgscL1/ORWGcbMFgMLgPUy4wZBEARtRmbWnoSlfYE9zezwdP0R4M1mdnQhzF9SmPnpeg7wZrx38Scz+2lyPx/4rZldXcrjCOCIdLk58GBJjPWB/zQRtVmYvkij08J0kiw5YTpJlpwwnSRLTphOkqWvwnSSLDlhVqQso8xseN3QZtaWA3grcH3h+gvAF0phrgfems5XSYKrHLYYrkUZ7lnWMH2RRqeF6SRZQt4VH6aTZFlZy9RJspSPdg43TQbGStpE0qr4RPSkUphJwMHpfF/gZvNSTAIOSNpPmwBjgbvbKGsQBEFQQdsmrs1ssaSj8V7AQOACM5sp6VS8JpsEnA9ckiamn8QrElK4K/FJ7sXAJ83slXbJGgRBEFTT1sV0ZnYdcF3J7cTC+YvAB+vE/T/g/5ZRhPP6IExfpNFpYTpJlpwwnSRLTphOkiUnTCfJ0ldhOkmWnDCdJEsP2jZxHQRBELz6ie1LgyAIgrpEJREEQRDUJSqJoGUkrdGX4ZYl/rLm0VckLbymbsm91zK3kk+TdFbLcQs6C0lDJW2zXDNtRV/21XAAawID0vlmwN7AoML1TcBf0vU2wJfrpLMpsFo63xU4Fli3FGY14EPAF4ETa0fBfw3gK8CP0vVYYK+KvN4GHJrOhwOblPybyp3ifRGflLqgdrSYxk0Vst1UON8B1zh7NF1vC3y/Ik5uuMpy58RvIY+GZcp5TjnPEbi3Ip8pvZR5FPCudL46sAj9TW4AACAASURBVHYr+eQ87zrpVLltBvwIuAG4uXD8d6Ojxf9ss/fuPmBGvaMQ7pKKdC4pXdf9PhTC7Aismc4/DJyBLzjLTifnvWvh/twKrAMMAx4G/gycUQqzUXq/dq4dyf34RkdO/v3RVPgfgJ0kDcVf7MnA/sBB+Mv+WeCHAGY2Q9JlwNcq0vk5MF7S6/EP76+Ay4D3FML8CjclMgV4qSKNC5PfW9P148BVwLW1AJJOAsbjK8YvBAYBP8Vf1Bo5cv8K+CPwe6BKXbhuGpIG4x/C9dN9U4qzDj3NoZwJ7EFa72Jm0yXtXJFX03BNyp2TT8MwLZQJmj+nuv6StsANUQ6R9N+FNNcBBvfivnwctyIwDG+ojAB+IOmTLeQDdZ63pB+n8q8uafvSfanq4VwF/CClV3yvjq4IW8OAX0hamM7r8RryntFe6feT6feS9HtQKb2tihfJKOibSmEafR9qnAtsK2lb4NPAj4GfALs0S0fSYTllanZvzGydwuUQM3tG0uHAT8zsJEkzCmmdnsowi+5nZEnGtdP15sAEuteqvY/MtWf9sZKQmT2fHtb3zewbkqYlvzXM7G63IdjF4jrpLDFf6/FfwNlmdrakqaUwI8xszwaybGpm+0s6ECDJpVKY/wK2B+5NYf4mae1SmBy51zCzzzeQpVEa/wN8Cngd/jGsBXoG+F4xgpk9Vkqjcv1KRriG5c7Jp0mY7DLR/Dk18t8c/4iti//xaiwEPt6izOAfwol4axEze0jSa1rNh/rPew/gELzyOaOUzhcr0llsZudWuB9a4dYDM1sbQNJXgb/jH3fhH+QN6fmM7i1E7fGMzOyRlM5uZrZ9IdwJku6VtCDJvrqkZ5KfgJdZWt2z0fehWGaTtA/wPTM7P4XPSSfrvcu4N0VWkbQhsB/wJZbm/cDm5gZRe2Bmp6R8/gCMM7OF6fpkYGkLjxX0y0pC0lvxm117sDUz4/+RtCmpBk/2pf5eJ51F6aNwMN1/ykGlMHdK2trM7quTxsuSVi/ktylL9zheTi9kLcyaFenkyH2tpPeYr02pom4aZnYWcJakY8zs7DrxAR6TtANgkgYBxwH39zJco3LnxG8YpoUyQfPnVNffzH4F/ErSW82smZXinHK9ZGYv1z7uklbxbFrKB+o8bzO7GLhY0gfM7OcZ6fxa0lHAL+l5Tw5pFMnMihXQ3ma2beH6XEnTk1vuM0rF0I5mdke62AEf8vk68HVJXzezL2SkUe/7UGOhpC/gQ007SxrA0v/9ynRafO+gzr3Bh65rnIovSr7dzCZLGgM8VPCfm+SrGs2osQFeadZ4Obk1pzdjZJ184F3CScDn0/UY4LuF898Dz+NDBrdTGmsspLMl8F3gwHS9SS3NQphZ6WY/iI+P3kfPMdLdgduAfwOXAvOAt5fS+Aw+JDAXbxHeBRxbCtNUbrwluAR4MZ0vBJ5pJY0Ubgd8nuWjtaPgt34qxz+Bf+HDQ8Mq0mgarlG5M+NnydKsTDnPKfM5NpwTaqFc30jpPADshn+c/6+VfHKeN94jOQO4Jx3fxoc1yuk8XHHMBU5qdJTSuBP/mA7ElWUOAu4s+K+Kz/ldnY6jKc0TpHBvAqan+/8IMA1vHRfDVI7NF/x3ps73oRDmtfiY/U7pemTFO1P3O5P73uXcm8xv3s+B2fj/6bu1oxTmS+nenZyOacAXc9Lvt4vpJK1hZs+X3DYxs4dTq3WAmS2sudVJY3VgpJmVrcvW/EdVuVvqHqcw6wFvwbuSfzKzpSw0StoN/xAJN4p447LIXUfWpmlIugQfB59GYWzTzI5N/l2tuEKcZXGrLHdO/BbyaFimQriGzynD/058TmhKIR+s0FrPLNcAvGVavC8/aiWfFK7h85b0c+AvwMUpykeAbc2sON/RJ8h3nDwLn28y4A7gU2Y2L/n/GG8JF2V5xZIF6Yr0hgCY2YKS+2m4aZ8eY/NmtnfflWYpWZb6ziT33PduNHXujaSzaTxvUftfHlzH/+LitaRxwE7p8g9mVh4+r5tRvzrwycVKDRIyNUOS+/vwHsLD6Xo7YFI6Xyf9Dqs6CmnkaNacXhHm9NJ1rkbL3sC30tEb7Zv7Savw69yTXI2Ypm6Nyp0ZPzffhmXKeU6Zz3FaxruZU67jKsIc10o+Oc+7Kp2iG/CO9NtQe4kWNAYbyDo9062ZNuGDJI3EBnlthvfCemhrJb+F+NxB+ejRK09h635nct+7jPtycDrOw3uCx6TjD8APepFeQy3Kekd/nJP4DktrkLxL0gfI1wwB75JNxNXPMLNpaSwQXMtpL7w1Z3RPToGPN29JvmbNbkB5wvndwOfVguZMakVNwIczAI6TtCPeOsvVivkL3tXuMd+Rxl53AIZLOr6UxsBWwzUo9wck/atR/BbzqFumlFZDDahm/qXk6s4JtSjzwXjLssghBbeGc08tvDMvSHqbmd2e4u0IvFDw3wX/gBYnyWsY8It03lTzTtJmuMbQBmb2Rrme/95mVgvziqRNzWxOCj+GaoWIZtqEOWPzNW2tH5fzsDSZnEnVd6aoqVb3vSvS6N5Y6glI+gTwNvON2ZD0A+CPkq40s/0k3UdFj8PMtinkcxLNtSgr6Y+VBLa0BskgWtMMAVhkZgtK6SxJ6e+VfustljoO13DYiDoaDunBHwWMUUGdDVdZqw0/tKLR8h5gOzNbktK/GJgK/KmFNNYHZkm6m55/tG8Da+HvS/GP9Axu4r3Gqs3CNSn3Qxn5NM0jp0zmQxDNNFFa0ZA6DviipJeARSmsmasy5tyXA/FW8iaSiib118YtJOfkA/nvzCfwCewh6fopus32Y2Ynpd9mWkw5mnfNKpLPArdImpvKM4pq7alm2oTPA9Mk3UTPZ10c4qmnrYWkYQ3SxsyeLF030lRr9N4VyVFvH4pX8rX810pu/5Wu96I5OVqUlfTHSqJKg+TPZnao8jVDAGZK+hAwUNJYfGLtzpyI1q3h8JiZjakT7DLgt8DXgRMK7gtrL6O1rtGyLt0vUm3ctpU0Tq5TntuA2yRdZIX5ll6Ga1ruRvFzZSlwcgN5czRRsjRVGrVCM2W+E291ro9XyjUW4koRTfNJ/rnP+358knxT/L1ZgKtSFituSj2fGgvwoatp5GneNaxIzOym9B/bPDk9aBXqnDTXJpzE0nvWlKnU1krvXtXIQFcQfHK6RjNNtZObyFEjp5I9DZgq6ZYk287AKWZW007M+R/kaFFW0u8mriWtj3fN34Xf0BtwrZkn0/DBYXh3vKvrbWYfq0hnDVwjYPfkdD3w1Tovbz1ZpuKTcFuW8vtJRdjXlMI8WvBrKndqiZ4GFF+kE8zsitw0MsozHPhcRRrv6E24euXOid9KHplleyMNnlOGf9WiQszsD30pc04+KVzD5y3pd8DTeMuyOAH+7VI6l+HDFL9OTnvhFclofOjmanzMfAe8N/Iw8GFLk9Ipjd/iGktXmdm4VJEcZmbvTv6D8J5NrWy3Aj80s0UlWWYBr095vER3L6o4rNJM2aRK2cMaNOYqqfOdOc7MniiEGQWMNbPfp+/JQEvrFAphGt6bQrjX4ls7gzd6/1Hwq1qYtwDXWvu0mc2V9BncUsBueAPtY8BlzRo/QL+cuN6xnhv+Un8VmIN3rW8AzqqTzgdz3JrI8jf8o/1PfBzwH8DVpTDvw4dZnsNf/iXAzFKYLLnxRTh7p+O1raZBz4m7F/GPR1GN9gb8w3M/PmZ9AdUT0E3DNSp3ZvxcWRqWKYU5qdFzauafwvy6cNyI/0lv7sV9eQu+evdZXL26/Aya5pPzvEkTzRnv8B+AtQrXa+HqwKsDswrua1IwH1JKo5k67o/xubN3pONC4McV6YyqOkrvVKWySYv/W+FrJL6SrkcCE1tM4+PpOc5J12OpVoCoujejK8Jtg/+vq5QHvooPja6ND0sdAdRWYd9aCLcb8E1csWW37LK0egM7/aCBBgkwNf3OSL+DcHXGltJpQZYXcN3n6el6A+DGUpjpwHoF2d4OnF8KU1duYIv0O67qyEmjjuzChx9OK7hNKaaRzidXxG0arlG5M+NnydKsTMn9vkbPqZl/nbw2Bn7ei/tyD95anopPah8KfL2VfHKeN9763zrjHX6AnjaJVgMeqOVBE42jFK5mk6urIqm51d6FinyXciv4vQb/cI/Eew1d9xcfZp1acPtLKe4gmqzJwCeSzwHuT9dDa88JOJvCWoTyUUhjGj4XVZTlvgZlalTJXpDei4vxCvRCetplq7p/05rdx9yj38xJZGqQ1LqvT6fhg3/gL1wxnXfjk8AbSfpuKZ16Jjzq8YqZLZG0WNI6+AKqjUthFpnZE5IGSBpgZrdI+k45TAO5j8dbDt9maQxvmTVLY+mI/oZdk7QianMHtTT+Lum9eE+parIvJ1yjcmfFz5SlWZkAXmjynJr5VzEfeENvZDaz2ZIGmm/Ze2Eatqy3krgqn2JePZ53QRNmFeDQNFlcOXSTuBT4s6RfpTB7AZelMe1ZNNc4Al/sNc7Mniu4XU23XaUs7SZJe+Pv+evwZzAK75XVbDbVVTYpcC5eUXw/XX8kuRXXZLzZfOhnKoCZPSVp1eR3T50ylqlcOV9RpuPwj/5C4EfytQwnmNkNhWBvMbMtG+T1vKT98HsKrgzxYjp/o7pNlfTImp4KD3XpN5UEeVov58lVGb+MT3CthVv3LPI3/EXYG3/xaywE/he6FqXUxcxqdmh+ImldXINhCj6EUJ5MfFrSWni3/lK5CuhzpTB15TazI1KYd5tvB9tFGpdumkYhfFFlcgA+Fl1M82tybZhP4y2qdUj3pEROuEblzomfJUtGmQDuafKcmvmjngufBuBDHUV7RLkyP58+SNMkfQOfBO4y6Z+ZD9R/3r+rCFsXM/tqGjevqUoeaWa1D+VBkv5idTSOlK+O+xnytJu+ig/H/d7Mtpf0dnxYqEaOsskE62kG42a5GYwii+TGAS2VYzjdmo0Xk8dtkmr2pHbDNfp+XRHuY2Z2lqQ98J71R3A7TsVK4i5JW5rZrDp5HYTPj9QqvruAD6f5mV0sqTn3lv44cb2Nmc2ocB8A7GtmV2amszdwrSWV0pLfLel0MP7RmY6/3NsA95jZW+VNiBFm9liKMxpfhFfWHlmT7mGpg/Du8qWWJsBy5ZZ0r5mNq3JrIY0LC5eLcfMHPzKzf6U/zbFmdmaTNHLDVZYbn0xtGD83j2ZlSv4Nn1MLz/Hgcj7WcyV17n0Zhc99rIpXIEPwRVqzc/JJYVp615sht4a6M/7R/KOZTS/4nYcbwFxK40huIO/9eIOrqHW0ELjczO6s3Rf8A9dQu0nSPWY2Pn3Ut0+9u5oNqLKyiehWNnmxkMa9+NxisddydfG/I+kgfDx/HD7Esy++QPAqSd8xs09J+jXVaxP2TmkstXIen2fpEUfSDDPbRtJZ+PzBLyVNtYIhQ0m7pPv3Dxr3+pZC0jrmFmQre9lWUuutZFnHqzrtwE0W3I3X3ENKfve0kM5P8Um/b5DG/SvC/ILCuC7wRnpOeNYdg0z+A4FbMmSpKze+YOdNeLd7e7rnI3YljR23WvYGed3dF+GalTsnn1xZMuVt9pwa+hfCrZregTdSbXso575cuqz59NXzTukchy8MOwU3NHcfvuq3tsfDLHxoq9J+WUrjrX30Xv0e7xGdDfwMbz23aufoncCjuAbVbVTY4UrhtsAt8h4NvKHgPi797lJ19OL+Xoj3Gh7CF26uzdKWEGbjFe0mVE/Yj8BVev+Vjp/jDRvwhi5029x6uHDMzZGx3/UkoGsV46HAB/EK40Izu1G+Kvk/wBUUhnSsTm2axp8PTGkZ/kB/Zt3mdmeaWdmGfZebfEHb98xscgNZb8I1FRY0CFNX7tSyPATv0RTHSxcCF5nZL5qlUchnBP4HrA0t/BFX65uf/M/Ex3PLafQY7sgJ16jcmfFzZWlYphSm4XPKfI674q3OeXhLb2PgYOupAptTrttxkxhFi50t5ZPCtfSuNyjXDPwj/1y6XhMfzqhaid2F9bRfNgb/oL8F/x/dBfyvmc1N/rnPck18qLBmUrvc6x6PT6CPpjCUbqUWt3wHvma9loG4gkIxnUcl3WRm75R0ujUwza/qVdA1tdSvWc+Rgu3wD/bTchthG1mhpyrpLjN7K3WQdCO+/qi2z8aHgYPMbLd6cVqhX1YS0PWQ349rHTyDv1hDcVWzImYNdKTTQ/sIvvL2flzz5Lvm+0v8DH+pf5qCH4SrCx6Y4j6Qwj+SwlXpdf8K7wHcSM8/yLGFMA9XiNZDbjUx/ZyZRsOXrTDMVk6jvE6iabhG5c6MnytL0z9Qs+eU+RynAB+ypJ+fGio/M7M3FcLklOsn+ET0pNJ9OSM3n+Te9HnnkD52EywN2cjnuSab2dbpelNgvpm9lCqwbfCNcZ4upPEnXFvoZ8npAOAYM3tz8s96lhmyPoivXr6PwoS1mT0i6R1mdrN6zo0UM6uZGUHSMbja8z/xCfSu5y1fq3E4cD6u1aVSOvemNL6R4l5WKPMa+JDR24DPmtkDqjO/WWo4fB9f8Phrei4ArDUAp5nZdqV7UeU2FFfFLa6b6dG4qKLfVRJy2yeHAu/FP0Dnm9m9kl4H3GVmoxrE3c26LZHundJ5Pb4r1cXmY/Nr4Prho9MfprgI6A/AuYU/VGVe1r2JylC8G1kVJneCrEtuucZMefHUqblpAN/MedkapHFwjtyqY7US8sqdk08tTM4fKOM5bY03NOr5DwVuq2ixzii7NZMZbwVXyVLbPGapNFvNJ8XpetebhDseX2fxy+T0fryH+p3kPw3vxY4GrsO1nbYys/cU0qiSuWsuIUOGF+jW1urhRUFDR9LtZva2OmmcYr6j24UV3mY9F6bOxjWcnigHVFrshn/oJ9Ozkuiq2NR4jvA+/Ft0RGbDoaHMqVd+Id2V8IG4Ib93FtI4HB86HIGr574lydC8IrY+GLfspAMfZ/wosHqF30eaxL23cH4xJVv0Bb93Fs5Xx3eF6o2sTdddUKEDX5UObrTsJ8BjeCvoPkrrLTLSuAlvaQ9Mx4dpYV/enPL0Rbkz49fWxixTmXLyS/fuAnxR2K7p+BEV+zz0wX2Ztaz5tPKsUthx+OTysfiEcdV9/hzeO4DC2oB0fTqucjwaH0//HL7qt4fV5D54r96Z7s2B1NlvG5a2fFp2wxdOrtIkr6808X+AwgI83PhmbZ3N1EZxe/EsR+E9z3/jcxLXUFg/ksLchzcea+sntgB+kZN+f1KBBcDMdpGrEG4ht1PyoKXxXTO7pHHs7laBmR0s6bWpR2F4F/sfye8m6OptfBOfSNxE0nbAqZZvv77KRkyZnOEBATuYd4dnmNkpkr6N20jKRfhS/bPxvZgNVx9sZuCtnEZfhWtU7pz4tTBVZTokI34r+QnvUX4S/5CCz318v26M3uUDPtwwaxnzaZqXemrDzEtHl591z23UdnD8KPV3cNwv/f5Pyf0AlraJ1LKsBQ7FP36D6B5uMrot1kJas1GKdzXwJnWvr5oL3CrpN/Qc3jmjcP7VDHnOl6t5g88RHpbmVb5eC5CGxd/L0vMoZxTCXIzPoz2drocC37bUkzDv0Tb75rxoZi9KQtJq5kNdmzeJA/SvdRIASHoPblFxDv5ybSLpf8ws54PZNfYm37v2JNxcsoCzJZ1qZhcUwp/E0ubEKy3DNsuvD8LU1PyeT0NrT7D0XrkN08h82ZrJ0VfhGoVpJf6p+MTuU9D18fsWXnnk0iw/w/9LZ1n33MFAfDVyK+SUS32QT05eZWN3tfCi54f9UOBIfPe8h9P736MxZnWsJfehrDUmmFnlh095azZq66seTceq6WiJ9EzWMrMRqt4gqaia/Gv8v9tjHqXENlaY4zFf3Le9MjclSsyXr/W5BrhR0lP4HFtT+l0lgW/J+Hbr1ivfFN/wu5VWNXiXeHvr1kJYD2+FFiuJqhWeK2qS59fpJfgmPvxh+FBENs1aLDlJ9HG4ZYlfC7NNrYKALo2w7evEWRZuwo29PZuuV8dVG3doIY2cco1NaS9LPk1p4cOu4sfIfOe707s8fQe8/WjSWm6WR2a4O1V/0VlTE+qW5n16ZOzaR2uZWdWq5UrM7JX03ylXDlWMsObzSQMkDS01dFYhY/W30m6EZlYzK35ymgcZQubCyv5YSSysVRCJufjG5gPw5e2NzH3PK5w/gb9AXekmtyK9NieeyPrYSdohQ+6b0sf955KuBQYXX9DUxeyh6ldym0edFksKm7NA647ccPU8avFpfG/uULfZimZ51PuDtULOcNNgM6t9uDGzZ5OSQ3egfJkb5tUsn5RXzvPuCy5h6eGbImPIay03Iue+gE/GTpNrdvVYdGYtmMyXW749EtdOmgysI+ksM/tmCzI/K+l7NFHrBX4raXfraYajzLfxVddXpesP4j23ZsPnAFfLt7zd0tJEtrnp+mz6jXZToQu5Gz6RcyXemv4gvsXgUSqtZKyTTm1ccjtga1xbw4B98EVChxTCNlzhKekSM/tIKf0ut/TB2hX4TfkPXQi/O24ptJncDcvWSNuicD0d2LX0Qb3NutUd7zGz8Y3kyA2XnldluSXdg2/SXvePI+lRvCV0BW4FtfJFlvRRXHe+8g+WhgZmmtkWdeIPxA29bdZAlmH4h/AY61aBfBO+tuKthXANZU55nW5mn2mQ10PA/o3ySe5Nn3dfkPPe4ZPAdVvLqQf8UZbuaRxbL06ddBpqqaUwOWb3p5nZdvKV1+PwSfcpGS3+oiwL6O7RQ3eFVVbR/i9chX4A1ZtI1cJtSbcdtpuLvaViI6hCjqm4jbYl+LxcD3J6c/2pJ1HsQv4TXwEJPuNfexlukm9j+ot6HxW6xyXnpKPGr8oBzTdA/1I6qigvtBtIt1Gz2tDH+4AzJf0B/3j8ztI2hSnMDZJ2z5C7smxyO/Qb4TZktqe7VbwOrrddpLLFUvD/vdwufbMFWjnhGpX798A2ku5vEH8LfPjgk/gE4bW4qYcedmrM7Cep0qn9wf67+AdLQwMPShpphT08Sv731/OvySXpU8BVkv6G3+PX4qYdijSUOeVVqcJZ4EON8mnxefcFOa3MZq3l6/AdFHvb06ixMMPtElzzaA98vuogem4WBDBIvsfF+/EKeJHSZj015Nu9TjOz5yR9GK9MzipUSFUf36p7dQa+X/Z9Df7bpHe2nu2mm6jfmzsAn1ct27TLpt/0JHKQb86xJt6NfIE6tXZmWmfjLZ+qVZXr4j2E1elevCd8f4DzzKyHRc/0Qr4b/7O/DTdDfXjBv6nchTCL6V6VargJhUNwXfaiXnePFdmFdBq1WB5h6T+xWWmBljIXctUrd278QjpD8RW9B5lZ1T7XDUkV1fb46vxipbR3jn+pPMWVvIsKfj3WJdSTWdK5+Ef+qlJexcVejfL5Bq5umfW8l5VmvZPUkj2VBq3lvurhSJqHr0B/KuWxLr547Z/Ax81sSq3no26bSYNwe1RvKaRzLL7/+nR8LmUk8FMz26kQZgawLb548CJc9XY/M9sl+X+6INpgvHFwv5Xm99K7tatV2Ihrodw5vbkvWQPlHTVYe7RSVRLQpy/kvbj64XC6F7Hsjy+6Mrzltqj8UjRIbxCwJ64tsrOZrb+sMpbSb7giOzONdgxX9LrccsNn+6f49wBX9KaMKZ2lqI3dNvPPzKO2kKqhzMpY7JWZzzI/78z85pnZ6JLbceZbw9aGS3+ID9dWtpYl/S8+EX8tS28p2oosP8Jtp11fyPsD+EKzs8zszZLuNrOJ6eN8FF6J3F2vAZLSEb6r3OJ0fTCu4DFO0onA42Z2fqP/h9wUyPVmtmvJ/SJ83ua31FG3zSh304q6USXSLI3+NNyUyxBJ30rnt5rZtcuQ1g5mNqFw/WtJk81sgqSZZHSd5ftX7I/3PG4ltUgqwu1NYXvHstxqvq3lCLktqoW41lOV3foMcXUszbeZbLodZaNyZ8afh296cyVu4qBsXj0bM7tN0gZ4Cxz8o/GvXP9MlCOzmbWyLqUyn/TbF88bJVtFDdyqPuSH4L2k2nDpY/jmP/VapC/jWnlfortnnrN+osxbzOzjtYuU97fM7H/SRxq6Tah/hW4T6ic2SjTJXdxL5jhcGeYL+OLMneUKF+X1IUXWwFc7l3k4Hb1St02sW3aQ9EEzqw0bn5WRRn3lDOvDlX+dfuB7QD+D68h/DDfbUXfXryZp3YuPZRZ3xhpJ925WU/FV2xOapHMZPva5WhO5b2okN022taR7teceuImFrWh9p73/kLfNZNPtKBuVu1l8fOX0ia3I3qRc++E64xfjq9YfxjW0svxbeF+aykwDi565+fTF88aHSIbhQy5D0/kwfIj1AXxV86/xoZ1JheMWSiva8eGYP+AbJx1fOwr+c4H1++A53oAPE42ie2X3jel9aeldb5LPVHwu6Hhgp+Q2EvhoIUzNSu4MYGZ6lkc3SHMtCtvEtijP8/Xeg1bfm0q/vrpxr4YjPbDidoIDKZk1bvFFeQ++6OYWvLX7CD6GuSZuEPABvAUyhwozyuSbCp8BDGhFbkrbWtK9jeVZwH/VytBimatexqytJ4tuzcqdkyZ9ayp8OvCawvXwkrwN/TPzuDdH5vRROxTv5a+Ct8obbpVazqcvnjfeWq6pkhZNTE/HzWePwnuBd9HTXPY4SiYt8EWnSx0F/xuANfrgOa6Pr66fmo7vpWe1KvD6FGYD3Djfb9P1lsBhvXiWa+JDUACb4YtQi9u8jiocG5XvSSHcG5Osj6RjCm77Cror5nrHu1N5/0XPbVQvavX/0ejd6HfDTZJ2tKU3YSm6FYdphixDVmeZ2XXy9RE19ckHrXuDk+9I+mWduECXNssSSUOs+aKbdenu2ufIXd7WcoqkG4BNgC9IWpteaJIoY5tJmmxHmVHunO0s71CeHnoOA6zn8NETFHaDy/DPYR7wSIbMw82sOC9xUdKcaiUfWMbnbT6n4uacbgAAIABJREFUcJakY8zs7DrBHgHeWhqKu98K2nkprZpxwjXMNQLLPIevb7iFnuPyLanAmtl/cEWNKmprpy7Ce6Y1jcS/4s/j/BayEt4z2ikNXd2AKwnsj2tLYQW12yach/eqbgGQW9L9Eb44srjqfSQ9J+QfxXviDXfRbIG6a1H6XSWB16zlCZia2/8DTpe0MX6zd6bnXsddyE1Mf9B6rj6+3Mz2ADCzi7S02eFN5frR95nZv8xNFG8L1LQieuzqlXgWuC/lV2kqPMk9Nf2JKuVW820tD6Pbbv3z8hXkh5bSaLiuAx9iyNlmMmc7ykblzolfs+JatHJrdGtmtcLvJF1PTwWE61rwr2m7XI5PRBdVp10ws/9Wt8XPRjI/IVepLFr07FrEmZNPOm36vDP5h6S1zWyhpC/j/6OvWfc6jQ/iZk5uhS7zNZ81s9p+y8j3nz8fH1IZmf4T/2NmR6Ug16RjmZBvM/o5ll4DUby/65vZlWk+ATNbLKnRAscq7sDnI5+Xm+/5vpl9Q0tvg5rDmrUKIslzq9y+E5ZWvacJ+V+a2XXp+t3A+9O3ZLqkyyzN16Xv1MZWWjehJmtRzOzoegL2m0oivYg7AMPVvSAOXMtoYJpYWoKvyqy1ej5vyWhfBevb0quPX1MKcxiu41yz77QrXqNvIulUvEv4cboNjP1U0nmlltkv6GmArFyuXLmLS/QX43sMdLUOzLd6HAF8yJU1uM3Mynvu1l3Xkc7H4GYh6m7YksJt2yxcvXLnxjezt5fj9oakufJd/N7W1iicZ2a/zPEv8D688rhS0hK8dXqlFdZWZMrczMhi03xSXjnPO4evmG/b+Tbc7Mg3gXOBNyf/L+PzbrXtYIfj61yuLqTxHXxuZFKSbXpR0cJaMIvfhEvx+7EXvmL6YHydVJHnUoVpSd634PN3XeR8UCVNTd+cg/DvALTeuwSYK+kr9NzvZG4pTHlC/rdyVecaNybFllXw78+/JN1pZsXeRO/XorQybtXJBz4eehK+cXxx7PN4YGwK08r2pVPoOSk9itLkDr7CeoPC9QbJbRi+5eMMvKVQ81+TBnMJ+AThNhXuWXLj46/D6/hVTX7/v+T3BbyLuhif2H+GbjMkXy+k0SfblzYrd058fMx8Hbxy/jHea9q9l+9On2xfWgg/Fp/gfqUVmcncvrRZPs2ed4tlmZp+v45vdtTlVnVv8A9l2e3PFfGKcz4P4x/GHkcvZJ2SfovzfpNLYcbhPYEF6fev5f8cXjGfgVfOB9eOUphd8Erv8+l6DL4ZWasyD8UbIffi35zvAENLYa7HK+PR6fgSrk5bfkaHA6eU70G67vXEfa8idfJBz71fB+Cb1hf/OJ/BJ3W7JoDqpLMHPu53Cb4Q6BFgj1KYWaVr1dzwyaj7cJs+Nf/BFX+gW/EPx7D0Z/kzcEYpTF25U54n45pHT+Ljlv+mpElDxuQ3TTS98Nbt9/Dhs9pe2uN6E65RuTPjL7O2ViGthlpozfyL7x4+3DEFX3j36VZlBm4HVl2WfHKfd+a9uRZf5zAXHwtfjZ4f+G/iH7FD0vFb3LRIMY2r8V7+vbia6Gfwodua/3qFYyNc6ePUXsj6p/R7Pa5Asj0wp3QP/hdvcW9F/b3IW9EC6/WEO/mKK8NwBYTahPxZFL5b+HdmQ3xuZELt+ZfS+F98VGNDmnz7lsq/twXs1ANXrVwHb7XPwidwP5v8Hq44lmqx4JXLfri2xF7pWEpFD7fjfy3drY1fJ7c1cY2n43FtkJPxjeSnAZ8qpZHTCqgrd8rjRujeOAVv1VyP7yNcc5tRerGGVeSzI6nng3d7z6BnpXtLxXFzxX1pGq5RuTPjL7O2ViGtZlpoDf1TmD/jH8ETgDF18mkqM94zmIzr8VepizbNJ/d5Z96bNfDNe2q98Q0p9diS/xnp+K+KNNbHh4L+iWvi/JQmHyhSr6BFWffClTremN6ZKcDepTA5vdSmH1R8mHkWbhcOfIj0+72Q+SZgSG/e20IaH0zP+9x0PYbSpl24KZinccWGut++yvSXRbhOPOjeeekg3BbRoHQDB+CG0XLTaTrEg7fiP4C3fM8kWS8thant6nUMpV29kn/DVkAzufGWRVUFNpye3fsD8d7QRXjL+OFyuuk+Kb3wU9OLdVvyG0ih0mkgT264ynK3EP/CFPch/EO2di8/LDVFgFHlI8e/8Iw+3xcy00BdNDef3Ofdwj16G74dZu29Ku/ktgE+V7IXBVXhgv+Ojdwo9BZxcyJH0rqKce57k9NLbfpBxSvrjUv/sb/04t7+Ch+xOJ+CGmspzGa4FtQN+PznzVQ0zprk0+u1KC1H6PQDX7gyCLd9s0tyq7XgWpmTyBqayviDjMPHoo8pv4zJv9YK+H66rmoF1JW70YtZ9sM/ynun47UV4Ws69ieSdMehx5aufTYn0ajcmfEHpHu7brpej57zGlu18KyXeU4i593KkHlrmsxJtPgON3zemWmchPeQ/5quXwfcUfBvutCQiuGb0nt1S+Hjd0P6IG7WC1lz3ptbKo5yL7XpB5Um8ywtyHxw1VEKMx23QDARVyR5E/Cmgn/TBZgsw1qUfme7SdIxeFd8KeNckk7Dx+6bWTEtG6mzQtgxhTD74WOyt+Itzp3woa2rk/+J+Mfw58n//cBVZva1FsrzBbwLXSm3GtuLKZsC3whvBRc1Nv5Q8L8NN2X9sVSWf+Evfs1U+Jl4BdxwbUJuuAZlPhPX5vpcb+KnNLLtTMk3W/qemU3ujX8Kk/1uNZIZNwj5Dktb7i5LPs2ed6ZM0/Cx/Xst2f9RMo6XzqcDu1lJu8nMti1oHH6Knmaq18GHpbZNcQbjPfLRBVnNzIqqwjmyLtN7V0jnBlzFtGpNRy3M1fjw2vdwTa/jgPFmdkAreaW0VsXXWhmF7ZYL/lPM7E2VkelS17+MnhpSB5nZboUwtTmwW2hxLUq/qSRKaq/FLRcH4C/ct9WCddFUAfzOzJ5JKmrj8L0i7i2EqfsHSdcPAv+/vXOPlqSo0v33IUjz6BYdRfEBIrpwAGlsaFBoRFFUXoLy0BEcQNTRO3oZHwgKgiNoTwuCPGYEdJAL4hIc3jAgAvK4vOkXDaIXH9yrKLNm5l6lBxBEv/vHjuyKyqrMjKzMrDp9zv6tVeucyoqKiKrMisjYsfe356qXX2IdmDksKbdseM8SmAfE0H4HH+9hukVZIpy1Qj2LYK6TsaaUFCmZ0mSm3w/zCLmN5MYwhcrzw+s/KuhHXiM/qVwR4f3bod+tN/n9oY5KUbOo7E9gnkKPwL7LVclqUl4PZZKvrbI+w8xwfwnznIkHuixdaeXNSyhXeb4T+5QJ4mXCgesBuDP6blZkNxHh+RoINxY0McM3w8xHZ0XVrgRwlaSHw3uug5l3lqA/6PJrNftaed2R/AqAr6o//unTko6NylQOqCRfCNtbehvsergeJvqXT0xW1eeBdMuwGJJrozJfhN2wXYYhAogM+S9y9fYdo4kSDqAE9+NpEyeBnlb65jCf9itgX/reMA8QqF6u3WNlQTcLYMFOJ6PfPxyojsT9DcyjKYvCXhvAozX6ANhEXthvpUtj7wtgcxUkNwp1PUbyQgDzSe4FW76fH72eFJuQWq7s/WGQb1JPnbufdzR8ve61VVgNenlM1sBw/f+jMOTmZUi5yvOdyMUkzwawAckPw1aZ3wRWxZDcy4JAQ5lK7i0kn5IU+/VnQXgPh6cvl/TOUTvInursF5TLJzKE3SV9Pnsii3/aA+ZimlEZ3CeL7j5o1D5HpKRbzgb4I+MuoCeAWBqAGfo7eizKKDaqqfyAhcvPjp7PBnBr+H9d2MVwTnj+GgB7FdRT6h8enpe6/8EutEfRkwL4NSyAbGBzquTzLKnT75J6rkWFgBiqhe6SdG9Sy5X048Uwk0ojfZ2a5as2Z6tePyC77sK5uhRDHBVS+4wC+zF6+2sLYHe6eyLYx+ue7xr92i1c6yfDVs7xaw+g2rupak/iHACva9C/ZUXtDPv+EIlKwnK+PDhCm/8DYW8pPH8+gHNHqCcfx8H8sYQ6NoGtPP8dtuK4HFGMVyjzS4wYizKdVhIZL4ZJD2c8E44BNlAvRi9p/KOwDe5hcuGPhjuo3WBSHmsjF1Ep6UhaNridwqF8JO5l4ZFxc+1PYxdNnX73v7kn1/EkTB/nRhTbJI9BefTseUjTvUktV8R5MJPES0d8P9B/DZRC8niYeWtzWL/Xgrlp7pTyemBYZPJZ6F95VvaZ1RIWmTlmTwDflHQNyVV7XDXPdxKyZEk/LHh5MYBfSfpU/gWafMQeAF5G8vTopTnol95eAOBQDslNndjFh2hpXV9Kky1Z1YUh9VwIy+L47fD8MNiAH/f7lxiyElW/Sa8wH3xN7iP5r+hPt3wvg+SPpEtpKXgHUFjly3SiqsyIcTrhWaGdF6R0cDpOEucDuIc9cb19YYMOAGwm6b0k/woAZNorRTrqB8ISw5ws6XckN0L/cg+hjktgG9MDqGSJR3IRyUtgQUXfLyoHmwzeU6PfeTK7/mIEWYQSqsxnqbo3heVILpJ0FPv17gfeDzNpHFDUTvj8B8FiBb4U9k9eIikzLb4B6bwbYXM2vPc3NEG81NeB/sH7nPzgndpnknejRMIC1Tcvdc53JbSMh9mA+VzYBPmEelkRdwBwEC1rYbyHsjXM3JoiQLd7kz5K+quwn/YDVAyWkhaFvcS3hUMnKCQpikgZUNdglFualud8lPF0FgbTLa8DM5MLtiKdnyv/Vti1mO0VvggW1/FK9DspfDD6P79X8nWSi1GRSwOYhpOEpC+TvBY9Ub3DJC0N/z8TNo9tXWf2v6E2W5lnw6XR89/CJD/yP5yYOulQ94DdTX0ONhEUfZ6vkNwrtd9D3p9kiwwTVpWQXaXuTUK5PUgejfLP/QRsM2//knb+CbYhuytMMG8lbLKej/o8I0kMeYzD5myd14GElWdqnyX9KncPEE+QpTcvdc63pP2qyklaNRmGSW4fmOdZRuF+jYYI0BWUS1VMLevnY7D4nkKyzyzpOpgX37Ayd0p6Y+7wsAG1Kh98ar8rRRcl9Snb0rSlvhcdugKWJfMGDFdlBsnY028N2ESYNP5Pu0kCWOXyNszt7XjYxfGKsEG7E2wvoW79IyUUz3Edwt0Uycej48Mmmlb6XcGrJO1XYT77FOzudDOSt8Ns8/sPqaus3HUw6ZD1Sz53Sjs7yLxtlgKrlvujZvaq2py9uuj1iNLBm+ZFk9LnX5HcEYBoGfqOgCW3QnhP4c1LTepmfYPMuH15ML8dHY6lDPCvJLkQtrcUq7PW7kNDUtqblTKgSjqf5H3oKfi+R1E++FRIbgqLoXol+lcBZSuiJ2BeUBnrSjqqoqmvoXdj+yzMU++AlD5Oy0liGDTXvOfDNtneABuUjpB5KUyCY2HLxl9J2qeo0Bj7LaDcfAZgM5hp4BUwv/YdMOQakrSE5v64eejvT6M7yWPDXs4VRZ+74v0guRuAP9IUY7O7+xdhtPwYhO13vBYmbLg5TPfqh6EvonnifGrY61GfqwbvGxP7/FGYa+XLYHtP18MigNsmyfuL/XL42YD5h4LiRXwbdqNzKoC3wPYBRlFMbUrKZxYSBtSwun1Q0pnh+RySO0i6u2afLoftQV2FguuX5FVRf54Dc5G+OCpyNck9FKTEC9gdg7Eo70O/bP1Qpk2cRAok75O0XXXJ7qHFPxAWFf2BirKd9zv050QAiwBsGPrWt6phCKKibc6eAPN2OU7SDrm6DoC5aQ7kIGDP334gd0XNvn4NZg6bB9t43B82AZXt7xTV1+frP+T1ymC6hDaWwr6vVvrcFCYGG7K3wQv0BsxvqkaOb4ZgsPh7ZkWAWBekfOZwbe2IiuC+cD7nhdVVdjN3X8p3mmvv7vzvZ0iZXaKnzwL435J+HZm9CdOLexrAHzHEGsEGsSgzZiURuIHkZ9AgKrYOwXb9lKQ/h+drwALcnoRtAl4F4EAOJi+CpDjXwjj6TQBfBbC3pIcKypR61kTEnj5vRX+MyXNJvh/Ajgmfu7Cvki4MduK3hr7vW9LvKpaQnF8yCZRtzqailD7T8gScCOApmGlua5gm0XdqtJVCkuNDis08gafDtf8wyY/DVkjrt1BvXVI+M2F399mAWrRqYjZBAIAsf8co4+lpwXx3Pfq90JZE/9/C/ux/D4fjSWZvkluiQSzKTFtJNI6KrdneXQDeJum/wvP1AVwvaccwgB4Es2XnvVCkyDOhSb9J3ijprQxeRSXl3g4TktuppMzVsB/4brC74adgAXdzc+WWSnp9sEOvkPTd6Fjy5y7pR7YaeT7M9BXbcmunL6VFVL8aFiMyLOJ6k2HvS7TH1+ozQ6QsyXfD9MA+BYvzKd2Ujd6ffL4lXZ9Q3+lDDv8edtd8RWKf5sP2VTaArUDnADhJ0l0p72+L7DOH8/kaSTfQHELWlLQylNkK5nG4VUVdl8Jc2r8RDv03WFDcvjX7tBDAB2ABlHFkfBwlXir/k9DGEpiX2RmSVtTpHzDDJokqSO6WtzU3rC8lXP5wSXX8/4e1U9hvkj+GyXH/M0xyo+9uKpiAsjv6XQC8BHYnFd/VXBrqWhe2ObtC0sNhc/Z1+cEmZTJp8rnDRX8NbPP+5+jZa/t+XDXqK50EGLk6jkowT1yNij6TfEDSViS/BeBfJF1HcnmNSaLyfNfs9zmw/ZrMJLYfLDDrL2DBWHXybxe1cYZyHjwj1rMCg/sOv4cNkCdK+k+a48FHYGKdm9Fy1J8l6a1RPeegYkClZak8HbZxLdie09/VMcOFen4GYAsVaHWFMqXyPwltLIWpPbwadu5qxaL4JBGRaqetUd/tAD6hXj7gbWG27TeS3FXSTcNMLkCy2SVrp0zkb39YesUFsDwF8aAhSbvm7M5DulJ9d59rs3AyIbkrTIZ8qORGyucOd3FbhjqTg+ZGJdGWXZofnOZHfycq+kwT8NsXNrFuD7v7vrrKbh29v/J8p9QT1XcXTNY7i3NZE+ZuuQB2freoU19BG6387oKp7k8wsTvANmbXBfAYgAWS9qYJFm4Pi1LPBAvz+lM/xogD6gh9vhzAR8omlyH9W6WPldjGEliszwApq+GZtidRRWqAWipHAPg+yd+Eul8C27gE7K79JljQTJ4siCaVsn7/VtLuJI9Tgapmqt2Z5OckLawqp3JPnzfB9iiWo7fpFv+9lBY1+z0AF0n6+ZD630OL69gAJkPQNSnXRWF+cMD2j0g+gIo+Szo6DHa/l/Qnkk/CYhOyeqtWu5XnuybPh+0fZHEq68Huwv9EsqkuVNu8LTfZrIjMfAeHY09LeoYhDiVMevk75crgPppy7eGw8x679da6oYJdDz8heS/6V++xC2xV/FIldUyjeXyS6Ke1ZVUYJHaGLdUz1ddVrpySjg9/29gYLOv36bDBal8kuLtVcABMy6oJK2HeUw+gNzkA/Z9hb9gP4WKSf4Zt2F8s6f9EZRYCWBoG3qIfV1sUfr+0qPLPA1iHvbgPwmRBzskVT+qzIocESU+gX+V3EYolMoB2zzdgzgzLSN4MrErA9BWaU8YNLdTfJs8hub1CBHvYC8kEMDMZkFtIZudrN9hewlVxJYkD6gWwjIXvgH3PByGKZ6nB8VUFZC7j74Gt3oDB+KUqGq223dwU0YG56R5J2xe8NqB1E6MgDZ3YTpm56S6YqNk+sME2306ylg9ryG+X1HE8gL+BbcINqPVKOjhX/jWwVJ4HKVK8JfkgLCp7BSL/cpnyaKskmpsWSvpcRZnGfa46B22e76jOjWAmGsDE534TvbalpAfr1pmrv/F1FeqZD+Bc2MqHsLiWD8Hk0veUScWsAVsBvD2U+QGAb6nmQMieI0bmFr4WgNtUTw4mqyv2XLpnmOmJJjuyA+y6uVcWYZ69thNM5PCJsGKaB+C0JquHGF9J9PNIy/XdTvJMDE+Ckpc2zzx9Vkmb1+CRktf2gunUvAP9+jmj0PiOQtLfk9wHlk1rnnpeJV+EbUYjPN8Etpp4L8zO/NlcVU9KGuZ50wUp5qarSa5X8UNto89V56DN820NmrmwyJPpAthnbcJpDd8PAJC5ML+O5PPC81jK5eJw7M+waPl8xHxdsuDO39E8oh6DrZBrwUHPpTNI9nkukfwQTBLkpqjMlySdG4p8A8BcmhjkpwF8C6brFMdXjMyMWklU2bo7aO9HQw7nvVluhd3lZIPlbADXSHpTVKZxv0nOlWnpjEyLd3xLYRuKWyvkO6BpHd0vaXOayN1asB/2xZJ+MaSOU2AmmytR4F9eoz+Vm86qiEkJ52guLK7hPNgP9UBJu0RlGvc5dbXbxvlO7E/lNRF+B8NUVWt7olW0szYGg+AgE1Mc5vkU96XWpnQYuC+BpZw9D7Z6+YKks2vWU+m5REtetqOCSB9NF+0OheRl0b7LcQAelfTPbVpFZtpKIsXW3RpKS5pTJm2eMXK/SX5WlvDlQwwCdbk+xtm2dpJ0e+798bGkyOCqQRcWTPYxDFHrDeaASyUtqmgmG5ji5b3Q09KpQ+Wmc0Idz0pSWCWdGX6oh3fQ50fKXqxzvlsi5S7zM9H/WarSZwvKNuEK2Ab7YgwKYO7VZkOSvhX+vRVDNKFIHqI0scUq5eXs2Mro+Ur0JxVaGfbGDgbwpvAbWiuh7SRm1EoipsjW3XIbR8B0a1bClrfzABytKK6A5DGwwLJ4sLyoyIuobr9J7i3pKiakLxx29zHKHUn+PWHQHXCXpAmpZWq9tyqo9XJM8inxpjMs/wIQbTpX7THk6irND16jnkarxjrnuw1GvWMt269r0JcHVBEENy5qrPhOgq0+Y8+lFZI+G+1bbgNbsVwBm5T3ga26Dw11lKYdbvxZZtokMcTWfZFq5tKt0dZyWUL4d8CE244FcMGQgXjoYBlee75MLbSzfjMxYX1CPa0MurRYgf9AhQwJyT0x6IJY26MnZdM5oY6kH2pVn3PnufPVblNI3lW1WUuLEcnIRAJPU41c74l9KQyCI/k/JS3goMx/HXn/On2pk2M99ly6TcFziebkUYRGudZHYUZNEim27pbbyzwfTgNws6TL6tr1aYEwf8SI/Wa/guQAkt7FxIT1NdpsNOiyX4Yk1sd5VVTmLNi+xltg9v/9YZ4heRNPSnuteIdUeanU7fMoq92U851ST1Rf4++GvUxvhF3LjwD4kqrzUdeCJUFwJF/V9e8915fUlcSmsNiWP4Tn6wB4saRHojIDCbpoIppHjGXiUws5cFeHB+wO5qgxt/ltmHDXw7DBYTaAxTXrWNqk3zAPh11gHiQXwfY39oZFpZ6aK7tJ7vuaM2KbOwFYL/x/MCz38SY13n9g1jZskLwM5gkVl7k/93d92F3YKP29P/yw5obv+28B3FKzjtL84HX6DMtZ/FmYbf0eAJ/u4nyP+bspPZ9tPML3NvAIry0Of29su92CvixNLHcfgOdGz5+LwbzXpTnCO/8s42poKjxgomTjbG8N2J3XBuH5X8A8erLXt0yoY0kb/R5WR/5YGEjmwKJqfwzg1zAhsbptNRpYokF0AYAfwVRn786VuSf8vQuWC3sWgJ+N+N0sCX+PA3B4fKxGHcsBbBg9fxFsTyIuc3euz2vn+wzg7nDOj4YlgursfI/xu6k8n209wjX38fCYGx1fCjOF/gommtj36KAfZyaWWzbsWgp/dwdwBiy96enR47zs+h/HYxKJPybJDSQ/Q/IVJF+QPbpqTNKfJS1RSJgu6T8lxYnaL0isqo1+r0cyNtdsCpsMYraQ9Dhs8/xaAJvCFCrr8qzsKs88ff4RvbiQFAYkyWF3WDFX0dI4ngQbVH+JnmZPXTLvkA8AuGZE75AUL5Wrc31+BL0NS0SeXfMk/YOamUdSzncKsefMqN9NyvlsTHAUuRAWr7AhgO+QzIQD3xf6sSbsWsw/6ra1Acn/TvIUkqdnj+x1SR9PrOrfSa4yAQbvuCyhWJYj/A+wVWX2uBIlaWPbZqbtSVTausdJyv4ELaZggyEvqU6/Sb4TJhPxC9hd/iYwYbHY0+pBmCfFd2GD+y3ZvkpqO6GeRp4+TFORjRMbfSGUO0GjxUk09g4p8FK5XwVy3TSf/lnqD/hqzbMr5Xwn1tPGd5MkMd+U4Bn2RpmUCWjSIXfG1y/J3SVdW1LHIUrwACN5B2xFmI+er+U9RstXfyFsZQnY6v0DijzbSK4pqQuX4TTGtWSZCg+MyTZaoz9LYN5O+eMXRP+/oMX21oYtx+cCWDv32m6wXLuPwsTDsoGltp0fJmT4KQA7h+cbA/jrGu9fF5au9TXh+UYA3p4r06oJAxabsld4bDhiHfvB9l9OgXmF5V8/AMDs8P+xMBHE1+fK/AMsruAVAF6QPbo4311d16Ocz5baWQGbeLPns2DupHXqSDKlpZar0e76ANYfcvwQ2Cr5F/nHuM7fTFtJJKXfHGN/lgCAKmIKODwd6AmKXGUb9iOzzZ4C9AnurWHdq+9qW+Xp0xSWJDYaoa5GSV1qtBNffyeGNvuuP44pMVaV9824XUbbIFzHh6A/5ug8SV+vUUfSNUTykwD+C5YjJI6ebzXLZRgjdosOzYLdbLxA0nFttlXETIu4Tk2/2TnBzjsXgFitHhqnA30bbHA5Cyb41QazYWqZ22GI6F7dyoYMugN6NC3wKMmzYT+gRcF8M+oe2zEA5isnjQAgub/B170wP3ggvv7OGXb9Sdp0xM9Ql1I9KkkLwt/a9vpJEPZK7oJdc1nMwWEj3Eil3jU/A7vGj4neIwyJvm4IFeQ4Ir5OS4M7lkliLMuVqfKAzfpnw5ZrG8CW48s7bI+wDb/jwvONAWyfK7MwoZ6lWVkA74+PtdjXJTCJgdnRsdmw4L66dVV6+rTQ39ZMGMiZJGCTTV0zxc8A/GXT6w8JJqm2znfbdU760cZvIrWOcA5fOIbPtARmOcge28HimTobt/KPmbaSOBCWMe1kSb/ZUaFzAAAQqklEQVSjSSAf2WF7/wTb1NoVpjm/EiYKNj8qk6Ie2uZdcxkpOlIppHj6NELliY3q0jipC4B/k1SVT6D0+qPlv+561TiduZHkfjAPsVHt6LdXFwFgNwVPVpZqDgHE5t5nYV5xB46hbevA6N+lUwV76oyr7JzM5StmmnpoaW5ptpOD+VKYe12yjlRJXbU8faYCYXDZKTxdJY2Q8L6k/OCJdS2B/SZb2WupaOtSSUNT566uhP2TdWFmvewcSJHJL7gg/zUGlWJrCR/ShCm3hDlNxOe7VQFFkmcq3Z22E3yS6BCaDMiOMPfBecHWfX38g2cLMr8p5ZkoHMcSHak6jDrorm6wxfzgwd35UbTgLpp6vqcTJL8DM5neVrSqa9F19ZBhx0eop3LSouXHOB6WFRAAboHJmvS5T3eFTxIdQvIg2F30PJhcw/4AjlWkw9I0piDUkRJvsQlWI+G4cZK46dy0jcr84GElsQAtrBpn4vkm+RbYb2hnAJvB7Pm3STotKtNq9smmpExatHzuD8DGEMCCPueOayXok0THkHwtLH8CYboxD+VebyNYqe7Ko1OZ9HEMum1C8mcA9k7YU2jSRspqr5UyQ97TuSz+VCG4kM+HiSh+FMBTkl4bvd6K6yp7ooV9qKarcuI5XyZpm6pjXTHTNq4nwb8BuA32Xa9Dcp6iqGBJj5G8EMB8knvBTAut6MDnYXVK0Lb4KjoedFsmZdO5KSkpUNsqYwXHd76nBCRvhEmP3An7za1ya45oy3U1jopfFbtQt88ALiD5YZRPWk+RXKCgmktT5n1qhLZGwlcSHULyBACHAvg5ogtS/elLGwdyJZqbxiaTTvJ2STtVl5wsbW46J7S1BMCDap4mNVWCeqyy+FMBkqfCMgo+DfNSuhUmy/FUVOYXMDf0/xheS6P2F0vatrpk33v+FsCXAfwO/WNErLs1F6Ys/Lxw6P8BOET9OnCd4SuJbjkQwGaSnikpUxnIxbR0oIUwPSVoI6JB9z6SF6HDQbcl9o7+fxLA26PnQuRi2wJEO2lSqxsa0/meakj6JADQ8sQfCpPqfwksHiWjFdfV4OCRkSVSGmU8/TSAV5dNWrJc5XNJzgnPH49fZ3qq1JHwSaJbHoAFTZVJUqTEFDQaXCT9mSbt0fWgMc5BtzGSDkspl7jpXJgfnBZdvwWA57A6ur6yO1UFxni+pxQkPw5biW8LiyU4F2Z2inkCwDKSTV1Xv4benX8Wu3BA7U7XmLTyk0PEEehtareOm5s6hOR2MImLB9B/QcbSwIUxBWw3B3NSStBxkDLoTiVG3VDOH2NCxr6qVWOKSSqUmzLne1yQ/AxsUlisAtXUFl1XZ8EEHV+J3s22VDOlaBvxFl3E0fTV75NEd9Ckt8/GoHvbLblypTEFKYNLQl/GIhyX2Jcp5YZYRdmPkDXygzMhFeiQiWVA8DGxz1NKFn+6QfI62D7CEvQ0uaCaYphtTFpd/57c3NQtT0o6vaqQpEtgch1FpEh3VLUxLuG4FJI9dKYIZXdSz4XJPGfJbDIeh8XFxHwDZlueC7NFfwu2IblLvGpswSQFAEfBlIMfZ5RvY4R6phVtua4CeLmkdzbtT0t7CZ3+nnyS6JbbaPIKV6J/KbnKBTYxpqBwcEntCDuWG6/J6rZ8LfwRhlXhLSTPyybtsHG8/hAb8rOSRMs+dqYsuv7wUM9CAAvbWDUGjpV0MU0DaleYLP434BpQbbmu3kHydZJWNOlMS5NWqt7USPgk0S2ZieIN0THBfrQZKTEFhYNLDaaScNyUWkmUbTqHp98f8rY8C0l+FGZ6uBfAHJKnSTopKhOnSd2Zw1OBNl41BqaMLP5UQu3Jbi8AcGgY5J9G7+auVhZHJExarJDuUMfaTr4nMWFSYgrYonQHOxaOC22VDrokPy/pK223Oyopm84JdSyTtA1NimUegKNhG6hx6szK6HomCD4m9mcsKUNXNwpcVz9W93uhBSoOMMJkPqzuvngLtqQ3NXJ/fJLoFpJ7wrwXZmXHJH2pTiBXyuCS0I+xDRptDLrjoM6mc0JdSfnBWZGxjy0IPoZ6SpWDZyrB9TXvunqypP81of5UTlqT/u24ualDSJ4Fky5+C+yOcH/0Mr0lxxSoHemOylwGai43ng26L6KlksyYA8t8N9Wos+lcxVmwAWc5gFvDnWafSifTMvalmKQqUbv5NqYTu2PQdfV9sHwvkyAl3iJFuqMzfCXRIezlNM7+rg/gWkk7V765V8fnYLIeneZgbuNuheQuAN4ME1Y7K3ppJYCrJD3cpP6uILlJwqZz0XvjybA0PzjJ5QB2Uy66PnfX2HjV6BTTlutqi/2pjLdggnRHl/hKolsyzZgnSb4UFk29Uc06DoDdhTfKwZxA483kmp4+U4mUTecishXI5qjOD14ZXd/SqtEpphXX1Ra5HL1J6w8FZSqlO7rEJ4luuTp4JpwEuwgEMzvVgRhDOlC065baZNCdBFuEeIKDAFyLsOkMO2+lSPp7ACB5K4B5klaG518EcE2ueGWa1ESTlDM6rbiutkjKpDWuVKlD8UmiQyRlwUuXhI3jWaqfTUpoJwfzOBl50J0Qa5FcC5au9UxJfyRrL6wq84NLOpL90fXnaDBjX6Xgo9OItlxX2yJl0mpLb2okfJLokDiADbZJPI9k3QA2Jg4ujbvbYl1tDLrjpHLTOYHzAdxD0+IB7LOfly+k6uj6cawaZzK7T7oDOVImrcvDYyL4xnWHRBvWCwCcCLuTPk7SDlGZscQUsCXhuMS2PgFbPSyHBXNtDOA7dTbsx0GdTefE+krzgzMhup4lgo91+uKsHnQZb9EWPkl0SEoAW0pMQcrgktCXVoTjKtpoddDtGpLHh3+HbjpLOrjl9pLSpLJC8NGZWbA9vamRcHNTtzxK8mxYANsikmsjmA5qxhSMnA6U7QvHlVHH02fi1Nx0boOkNKkJJilnZtGW3tRI+EqiQ8qiXkNMwTthGbRKYwrYQjpQticcl9LWrQD2jAbd2QCukfSmcbRfF5I/BbC1pKfD87VhJp7NW6q/TnR941WjM/3hCKlSR8VXEh1SFvUqk204FcAbimIK2G460LaE41Ko9PSZYiRtOjegTsa+kVeNzvSkQLpjbGO3ryQmCMmlAB6CRSiviimADd4nkfx2ydsl6YM12mpFOC6xrWNgMiDxoHuRpnA2uqpN5zH14XMA9mq6anSmF5yw3pRPEhOE5BKYy2OpemhCPSk5mFsRjqvRp4kPuqsTYdV4MoCrUGGScmYWKdIdXeLmpsnTRkzBAQCq7tJbEY5LRZZYaUllQSdjbwDPg60kq0xSzswiRbqjM3ySmCxEO4FcKbPKe2HCcR8M+kAbY+pGQM84JB0WVnaHlZVLWTU6046J6k15JGeHkLyg6Fhwe/0X2N38KQBuBPBJ2KqgbnRlpc1Q0mMALgTwvCAc9wcXjptypEz2eRlpZ/pzB8nkBGNt45NEt2wZPwkBbJnb2mzYSm472Mb1RgBeBuAj6KU9TaVycAnCcffABpkDAdxNsm7OBKcBJAc2pHPHUtKkTml9E6cTFgBYTPKnJO8nuSI4oowF37jugDiADT31xlUBbHG8QkpMQRvSHUzIZeB0S0p0/Sh1ONObSUt3+J5EBwSb8cLEALaUmIIzYJ5PQ48laju5cNyEqBldX1ldax1zVgsmrePkk0S3pASwFQZytTy4rG5y49OJ5DSpVatGpJmkHKc13NzUIakBbEUxBWw5HagLx00WJqRJbcMk5Tht4pNEh7QVwJYyuDhTH5LfRXF0fbZq/DsAp0ZvmwPg3b535EwKt0l3SxzAdk2DALaFJOeQXA/AAwB+TPLIOhWQfA/Jh0n+nuTjJFdGqrDOeNgiTO77wjL2bQq7NoBBk1T2GDBJOc448ZVEh5B8CSyA7V5Jt4UAtjfXjU8guawF6Y6kXAZOd5B8EMA2AL4Li66/hSExVVTGV43OlMJXEh3SYgBbLN1xpaQ/jlBHUi4Dp1Oy6Pr1UBxd33jV6Dht4iuJDgkBbCcBuBnmurgzgCMl1UpqzwbpQOvkMnC6gTUy9rWxanScNnEX2G45BsD8fAAbTI6jktzgckr4+0nY4JIq3VEnl4HTDXUy9rUh+Og4reGTRLc0DWBrnA60SjAuw4XjuqNmmtQ2BB8dpzXc3NQhJE+CxUjEAWz3SzqqZj2dpwN1X/zuYUma1DomKccZJ76S6BBJR+YC2M4ZMYBtHOlA3abRPWVpUhuvGh2nC3wlsRowjnSgvpIYD1UZ+8axanScOvgk0SHBs2gRgA1hd4WEmQ7mjFBXp+lASS6VVFei3GmZMpPUZHvmzFTc3NQtX0VLAWxN04G6cNxqQ5lJynHGjq8kOoTk7ZIGEs1MAheOW33oetXoOHXwlUQHRAFs95G8CBMMYGtZbtwZA01XjY7TJj5JdMNUCmBLzmXgOI6Tx81NE2ScAWwuHOc4zii4wN9kOWCMbblwnOM4tfFJYrKMM4CtLJeB4zjOUHySmCzjtPW1ITfuOM4MwyeJyTLOlURKLgPHcZw+fOO6Q6oC2Eh+XtJXOu6DC8c5jjMyvpLoljPKjnU9QQSyXMnbAfgogI0AvAzARwC4DIfjOKV4nEQHTKUAtpq5DBzHcfrwSaIbpmIA2zjkxh3HmWb4nkSHTKUAtnHIjTuOM/3wSaJDSH4Xtg/wJwD3wsxNp0k6aUL9ceE4x3Fq4ZNEh5BcJmkbkgcBmAfgaACLJW094a45juMk4d5N3eIBbI7jrNb4JNEtHsDmOM5qjZubOsAD2BzHmS64C2w3ZG6vmwOYD+AK2GSxN4B7JtUpx3GcuvhKokNCANueUQDbbADXSHrTZHvmOI6Thu9JdIsHsDmOs1rj5qZuOR/APSTjALbzJtcdx3Gceri5qWM8gM1xnNUZnyQcx3GcQnxPwnEcxynEJwnHcRynEJ8kHCcHyT+RXBY9XjlCHfuS3KL93jnOeHHvJscZ5ClJ2zSsY18AVwP4ceobSK4p6dmG7TpOq/hKwnESILktyVtILib5A5IbheMfJnkvyeUkLyG5LskdAbwLwElhJbIZyZtJbhfe80KSj4T/DyV5JcmbANxIcj2S55K8h+RSkvuEcluGY8tI3k/yNZP5JpyZhk8SjjPIOpGp6bKg5HsGgP0lbQvgXABfDmUvlTRf0lwADwE4XNIdAK4EcKSkbST9vKK9eaHuXQAcA+AmSdsDeAtsolkPlpfktLDC2Q7Ar1v+zI4zFDc3Oc4gfeYmklsB2ArAD0kClqf8t+HlrUieCGADWMraH4zQ3g8l/d/w/9sBvIvkZ8LzWQA2BnAngGNIvhw2MT08QjuOUxufJBynGgJ4UNIbh7x2HoB9JS0neSiANxfU8Sx6K/dZudeeyLW1n6Sf5so8RPJuAHsC+FeSfyPppvSP4Dij4eYmx6nmpwBeRPKNAEByLZJbhtdmA/htMEkdFL1nJXpqwIDlFdk2/L9/SVs/APAJhiULydeHv68C8AtJp8NUhT27oTMWfJJwnAokPQMb2BeRXA5gGYAdw8tfAHA3gNsB/CR62/cAHBk2nzcDcDKAj5FcCuCFJc2dAGAtAPeTfDA8B4ADATxAchnM9HV+Kx/OcSpwWQ7HcRynEF9JOI7jOIX4JOE4juMU4pOE4ziOU4hPEo7jOE4hPkk4juM4hfgk4TiO4xTik4TjOI5TiE8SjuM4TiH/HyNyHHgRi0GJAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                       Features  Importance\n",
      "2                     dst_bytes    0.292107\n",
      "24              srv_rerror_rate    0.145073\n",
      "1                     src_bytes    0.137521\n",
      "8                     logged_in    0.070587\n",
      "32  dst_host_same_src_port_rate    0.061325\n",
      "31       dst_host_diff_srv_rate    0.046960\n",
      "37     dst_host_srv_rerror_rate    0.040418\n",
      "0                      Duration    0.037696\n",
      "33  dst_host_srv_diff_host_rate    0.024021\n",
      "28               dst_host_count    0.015861\n",
      "21                  serror_rate    0.014469\n",
      "34         dst_host_serror_rate    0.014113\n",
      "26                diff_srv_rate    0.013917\n",
      "22               srv_error_rate    0.011445\n",
      "25                same_srv_rate    0.011157\n",
      "19                        count    0.011137\n",
      "36         dst_host_rerror_rate    0.010694\n",
      "23                  rerror_rate    0.010306\n",
      "29           dst_host_srv_count    0.009933\n",
      "30       dst_host_same_srv_rate    0.008326\n",
      "35     dst_host_srv_serror_rate    0.006300\n",
      "20                    srv_count    0.002605\n",
      "27           srv_diff_host_rate    0.002158\n",
      "6                           hot    0.001873\n",
      "18               is_guest_login    0.000000\n",
      "10                   root_shell    0.000000\n",
      "11                 su_attempted    0.000000\n",
      "12                     num_root    0.000000\n",
      "3                          land    0.000000\n",
      "4                wrong_fragment    0.000000\n",
      "13           num_file_creations    0.000000\n",
      "14                   num_shells    0.000000\n",
      "15             num_access_files    0.000000\n",
      "5                        urgent    0.000000\n",
      "9               num_compromised    0.000000\n",
      "16            num_outbound_cmds    0.000000\n",
      "17                is_host_login    0.000000\n",
      "7                    num_failed    0.000000\n",
      "['dst_bytes', 'srv_rerror_rate', 'src_bytes', 'logged_in', 'dst_host_same_src_port_rate', 'dst_host_diff_srv_rate', 'dst_host_srv_rerror_rate', 'Duration', 'dst_host_srv_diff_host_rate', 'dst_host_count', 'serror_rate', 'dst_host_serror_rate', 'diff_srv_rate', 'srv_error_rate', 'same_srv_rate', 'count', 'dst_host_rerror_rate', 'rerror_rate', 'dst_host_srv_count', 'dst_host_same_srv_rate', 'dst_host_srv_serror_rate', 'srv_count', 'srv_diff_host_rate', 'hot', 'is_guest_login', 'root_shell', 'su_attempted', 'num_root', 'land', 'wrong_fragment', 'num_file_creations', 'num_shells', 'num_access_files', 'urgent', 'num_compromised', 'num_outbound_cmds', 'is_host_login', 'num_failed']\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "XGBClassifier(colsample_bytree=0.6, gamma=0.2, max_depth=5, min_child_weight=3,\n",
       "              n_estimators=140, nthread=4, scale_pos_weight=20, seed=27,\n",
       "              subsample=0.6)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xgb_classifier(X_train, X_test, y_train, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bON9aNCsJ_Pk"
   },
   "source": [
    "## CatBoost"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KZW2zcoRCe3Q"
   },
   "source": [
    "CatBoost is another boosting algorithm created by Yandex. It has unique way of handling with categorical variables. It is shown that CatBoost performs extremely well with its default parameters. We can optimize the hyperparameters using cross validation, random search, grid search or Bayesian Optimization."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "xeNlk-qqTuLi",
    "outputId": "69f4e68b-8e9e-418b-9f95-0b9a4a4ab01b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting catboost\n",
      "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/20/37/bc4e0ddc30c07a96482abf1de7ed1ca54e59bba2026a33bca6d2ef286e5b/catboost-0.24.4-cp36-none-manylinux1_x86_64.whl (65.7MB)\n",
      "\u001b[K     |████████████████████████████████| 65.8MB 61kB/s \n",
      "\u001b[?25hRequirement already satisfied: numpy>=1.16.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.19.4)\n",
      "Requirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from catboost) (1.4.1)\n",
      "Requirement already satisfied: pandas>=0.24.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.1.5)\n",
      "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from catboost) (1.15.0)\n",
      "Requirement already satisfied: plotly in /usr/local/lib/python3.6/dist-packages (from catboost) (4.4.1)\n",
      "Requirement already satisfied: matplotlib in /usr/local/lib/python3.6/dist-packages (from catboost) (3.2.2)\n",
      "Requirement already satisfied: graphviz in /usr/local/lib/python3.6/dist-packages (from catboost) (0.10.1)\n",
      "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.24.0->catboost) (2018.9)\n",
      "Requirement already satisfied: python-dateutil>=2.7.3 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.24.0->catboost) (2.8.1)\n",
      "Requirement already satisfied: retrying>=1.3.3 in /usr/local/lib/python3.6/dist-packages (from plotly->catboost) (1.3.3)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (1.3.1)\n",
      "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (0.10.0)\n",
      "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (2.4.7)\n",
      "Installing collected packages: catboost\n",
      "Successfully installed catboost-0.24.4\n"
     ]
    }
   ],
   "source": [
    "!pip install catboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "zQmb6gzCKAOQ"
   },
   "outputs": [],
   "source": [
    "from catboost import Pool, CatBoostClassifier, cv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "1r_hxOEITp5D"
   },
   "outputs": [],
   "source": [
    "#model = CatBoostClassifier(eval_metric='Accuracy',use_best_model=True,random_seed=42)\n",
    "model = CatBoostClassifier(eval_metric='Accuracy',random_seed=42)\n",
    "\n",
    "#now just to make the model to fit the data\n",
    "model.fit(X_train,y_train,eval_set=(X_test,y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "LPZW8d9EUKqP",
    "outputId": "dfdc42c1-7100-4cb7-a4bd-47648672b7f6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start Predicting\n",
      "\n",
      "Resulting Metrics\n",
      "Accuracy : 0.988\n",
      "AUC : 0.982139\n",
      "F1 Score : 0.902759\n"
     ]
    }
   ],
   "source": [
    "print(\"Start Predicting\")\n",
    "predictions = model.predict(X_test)\n",
    "pred_proba = model.predict_proba(X_test)[:, 1]\n",
    "\n",
    "print(\"\\nResulting Metrics\")\n",
    "print(\"Accuracy : %.4g\" % metrics.accuracy_score(y_test, predictions))\n",
    "print(\"AUC : %f\" % metrics.roc_auc_score(y_test, pred_proba))\n",
    "print(\"F1 Score : %f\" % metrics.f1_score(y_test, predictions))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JRHz3nOo5Lek"
   },
   "source": [
    "## Hyperparameter Optimization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "2VI8wYsJDHne"
   },
   "source": [
    "We will apply RandomSearch and Bayesian Optimization to tune the hyperparameters. Since it is a time consuming process, we only show how to use the libraries for hyperparameter optimization task. For this reason, it is unlikely to beat the performance obtained by default parameters using RandomSearch or GridSearch. \n",
    "\n",
    "However, we can obtain better model if we use Bayesian Optimization."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "6KSi9W0aWHIr"
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "from scipy.stats import uniform as sp_randFloat\n",
    "from scipy.stats import randint as sp_randInt  \n",
    "\n",
    "parameters = {'depth'         : sp_randInt(1, 10),\n",
    "              'learning_rate' : sp_randFloat(),\n",
    "              'iterations'    : sp_randInt(10, 100)\n",
    "}\n",
    "    \n",
    "randm = RandomizedSearchCV(estimator=model, param_distributions = parameters, \n",
    "                               cv = 2, n_iter = 10, n_jobs=-1)\n",
    "randm.fit(X_train, y_train, eval_set=(X_test,y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4pSR8Fjk7BIu",
    "outputId": "d85670cb-156c-481c-c4f7-fdb5a7565ac3"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<catboost.core.CatBoostClassifier at 0x7fbd143f3c88>"
      ]
     },
     "execution_count": 58,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "randm.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "0DO1YzGh8VlF",
    "outputId": "539fae0c-1583-4aa0-cef3-a1c7bcd7d1e4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'depth': 8, 'iterations': 63, 'learning_rate': 0.017860792872981812}"
      ]
     },
     "execution_count": 59,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "randm.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "cQZBnttG8Z4r",
    "outputId": "178dc2b9-44d4-4b44-aadd-d835d3583955"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start Predicting\n",
      "\n",
      "Resulting Metrics\n",
      "Accuracy : 0.9892\n",
      "AUC : 0.997010\n",
      "F1 Score : 0.916126\n"
     ]
    }
   ],
   "source": [
    "print(\"Start Predicting\")\n",
    "predictions = randm.best_estimator_.predict(X_test)\n",
    "pred_proba = randm.best_estimator_.predict_proba(X_test)[:, 1]\n",
    "\n",
    "print(\"\\nResulting Metrics\")\n",
    "print(\"Accuracy : %.4g\" % metrics.accuracy_score(y_test, predictions))\n",
    "print(\"AUC : %f\" % metrics.roc_auc_score(y_test, pred_proba))\n",
    "print(\"F1 Score : %f\" % metrics.f1_score(y_test, predictions))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ulqF9iCf9o9S",
    "outputId": "966f72c4-2bde-44bd-eef4-ba5a8e469bab"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting scikit-optimize\n",
      "  Downloading scikit_optimize-0.9.0-py2.py3-none-any.whl (100 kB)\n",
      "\u001b[?25l\r",
      "\u001b[K     |███▎                            | 10 kB 18.7 MB/s eta 0:00:01\r",
      "\u001b[K     |██████▌                         | 20 kB 21.3 MB/s eta 0:00:01\r",
      "\u001b[K     |█████████▉                      | 30 kB 25.2 MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████                   | 40 kB 27.0 MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████▍               | 51 kB 29.6 MB/s eta 0:00:01\r",
      "\u001b[K     |███████████████████▋            | 61 kB 31.8 MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████▉         | 71 kB 24.8 MB/s eta 0:00:01\r",
      "\u001b[K     |██████████████████████████▏     | 81 kB 25.6 MB/s eta 0:00:01\r",
      "\u001b[K     |█████████████████████████████▍  | 92 kB 27.2 MB/s eta 0:00:01\r",
      "\u001b[K     |████████████████████████████████| 100 kB 7.3 MB/s \n",
      "\u001b[?25hRequirement already satisfied: numpy>=1.13.3 in /usr/local/lib/python3.7/dist-packages (from scikit-optimize) (1.19.5)\n",
      "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-optimize) (1.1.0)\n",
      "Requirement already satisfied: scipy>=0.19.1 in /usr/local/lib/python3.7/dist-packages (from scikit-optimize) (1.4.1)\n",
      "Collecting pyaml>=16.9\n",
      "  Downloading pyaml-21.10.1-py2.py3-none-any.whl (24 kB)\n",
      "Requirement already satisfied: scikit-learn>=0.20.0 in /usr/local/lib/python3.7/dist-packages (from scikit-optimize) (1.0.1)\n",
      "Requirement already satisfied: PyYAML in /usr/local/lib/python3.7/dist-packages (from pyaml>=16.9->scikit-optimize) (3.13)\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn>=0.20.0->scikit-optimize) (3.0.0)\n",
      "Installing collected packages: pyaml, scikit-optimize\n",
      "Successfully installed pyaml-21.10.1 scikit-optimize-0.9.0\n"
     ]
    }
   ],
   "source": [
    "!pip install scikit-optimize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 374
    },
    "id": "DJJg5nbw9nh4",
    "outputId": "d53fa7ac-22f4-4dff-b8f2-f453e7d691f0"
   },
   "outputs": [
    {
     "ename": "ModuleNotFoundError",
     "errorDetails": {
      "actions": [
       {
        "action": "open_url",
        "actionText": "Open Examples",
        "url": "/notebooks/snippets/importing_libraries.ipynb"
       }
      ]
     },
     "evalue": "ignored",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-23-66df6b8fc58c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0mskopt\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mBayesSearchCV\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mskopt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcallbacks\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mDeadlineStopper\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mVerboseCallback\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mDeltaXStopper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mskopt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mspace\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mReal\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mCategorical\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mInteger\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0msklearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel_selection\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mStratifiedKFold\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mModuleNotFoundError\u001b[0m: No module named 'skopt'",
      "",
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0;32m\nNOTE: If your import is failing due to a missing package, you can\nmanually install dependencies using either !pip or !apt.\n\nTo view examples of installing some common dependencies, click the\n\"Open Examples\" button below.\n\u001b[0;31m---------------------------------------------------------------------------\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "from skopt import BayesSearchCV\n",
    "from skopt.callbacks import DeadlineStopper, VerboseCallback, DeltaXStopper\n",
    "from skopt.space import Real, Categorical, Integer\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import make_scorer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "M1t6h3_R8yps"
   },
   "outputs": [],
   "source": [
    "# Converting average precision score into a scorer suitable for model selection\n",
    "roc_auc = make_scorer(roc_auc_score, greater_is_better=True, needs_threshold=True)\n",
    "\n",
    "# Defining your search space\n",
    "search_spaces = {'iterations': Integer(10, 300),\n",
    "                 'depth': Integer(1, 8),\n",
    "                 'learning_rate': Real(0.01, 1.0, 'log-uniform'),\n",
    "                 #'random_strength': Real(1e-9, 10, 'log-uniform'),\n",
    "                 #'bagging_temperature': Real(0.0, 1.0),\n",
    "                 #'border_count': Integer(1, 255),\n",
    "                 #'l2_leaf_reg': Integer(2, 30),\n",
    "                 #'scale_pos_weight':Real(0.01, 1.0, 'uniform')\n",
    "                 }\n",
    "# Setting up BayesSearchCV\n",
    "opt = BayesSearchCV(model,\n",
    "                    search_spaces,\n",
    "                    scoring=roc_auc,\n",
    "                    n_iter=2,\n",
    "                    n_jobs=-1,  \n",
    "                    return_train_score=False,\n",
    "                    refit=True, #use best model\n",
    "                    optimizer_kwargs={'base_estimator': 'GP'},\n",
    "                    random_state=42)\n",
    "# Running the optimization\n",
    "opt.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "cE1pAWjnDgvC",
    "outputId": "c4b0c07b-790a-4717-93de-2d614e70ca7d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start Predicting\n",
      "\n",
      "Resulting Metrics\n",
      "Accuracy : 0.9559\n",
      "AUC : 0.989242\n",
      "F1 Score : 0.714186\n"
     ]
    }
   ],
   "source": [
    "print(\"Start Predicting\")\n",
    "predictions = opt.best_estimator_.predict(X_test)\n",
    "pred_proba = opt.best_estimator_.predict_proba(X_test)[:, 1]\n",
    "\n",
    "print(\"\\nResulting Metrics\")\n",
    "print(\"Accuracy : %.4g\" % metrics.accuracy_score(y_test, predictions))\n",
    "print(\"AUC : %f\" % metrics.roc_auc_score(y_test, pred_proba))\n",
    "print(\"F1 Score : %f\" % metrics.f1_score(y_test, predictions))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4hbJmha32Zz7"
   },
   "source": [
    "# LightGBM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Mwl2xjBL2b7g"
   },
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: lightgbm in /Users/onurboyar/opt/anaconda3/lib/python3.9/site-packages (3.3.1)\n",
      "Requirement already satisfied: scipy in /Users/onurboyar/opt/anaconda3/lib/python3.9/site-packages (from lightgbm) (1.7.1)\n",
      "Requirement already satisfied: wheel in /Users/onurboyar/opt/anaconda3/lib/python3.9/site-packages (from lightgbm) (0.37.0)\n",
      "Requirement already satisfied: scikit-learn!=0.22.0 in /Users/onurboyar/opt/anaconda3/lib/python3.9/site-packages (from lightgbm) (0.24.2)\n",
      "Requirement already satisfied: numpy in /Users/onurboyar/opt/anaconda3/lib/python3.9/site-packages (from lightgbm) (1.20.3)\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /Users/onurboyar/opt/anaconda3/lib/python3.9/site-packages (from scikit-learn!=0.22.0->lightgbm) (2.2.0)\n",
      "Requirement already satisfied: joblib>=0.11 in /Users/onurboyar/opt/anaconda3/lib/python3.9/site-packages (from scikit-learn!=0.22.0->lightgbm) (1.1.0)\n"
     ]
    }
   ],
   "source": [
    "!pip install lightgbm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "NX7X6r4269-O"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn import metrics\n",
    "import lightgbm as lgb\n",
    "import sklearn\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import (confusion_matrix, precision_recall_curve, auc,\n",
    "                             roc_curve, recall_score, classification_report, f1_score,\n",
    "                             precision_recall_fscore_support, accuracy_score)\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "from scipy.stats import uniform as sp_randFloat\n",
    "from scipy.stats import randint as sp_randInt \n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import make_scorer\n",
    "from sklearn.metrics import precision_recall_fscore_support as score\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "Pveyd2InZVM0",
    "outputId": "f628d923-a755-4c51-8e6e-f1a866747d05",
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'boosting_type': 'gbdt',\n",
       " 'class_weight': None,\n",
       " 'colsample_bytree': 1.0,\n",
       " 'importance_type': 'split',\n",
       " 'learning_rate': 0.1,\n",
       " 'max_depth': -1,\n",
       " 'min_child_samples': 20,\n",
       " 'min_child_weight': 0.001,\n",
       " 'min_split_gain': 0.0,\n",
       " 'n_estimators': 100,\n",
       " 'n_jobs': -1,\n",
       " 'num_leaves': 31,\n",
       " 'objective': None,\n",
       " 'random_state': None,\n",
       " 'reg_alpha': 0.0,\n",
       " 'reg_lambda': 0.0,\n",
       " 'silent': True,\n",
       " 'subsample': 1.0,\n",
       " 'subsample_for_bin': 200000,\n",
       " 'subsample_freq': 0}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf = lgb.LGBMClassifier()\n",
    "clf.get_params()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "id": "YWAwflg9AC-m"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Type_Groups\n",
       "0    94630\n",
       "1    15611\n",
       "Name: Type_Groups, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_kdd = kdd.drop('Type_Groups', axis = 1)\n",
    "Y_kdd = pd.DataFrame(kdd.Type_Groups)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_kdd, Y_kdd, test_size=0.2, random_state=1)\n",
    "\n",
    "X_train, X_valid, y_train, y_valid = train_test_split(X_train, y_train, test_size=0.25, random_state=1) # 0.25 x 0.8 = 0.2\n",
    "y_train['Type_Groups'] = np.where(y_train['Type_Groups'] == 'Normal', 0, 1)\n",
    "y_test['Type_Groups'] = np.where(y_test['Type_Groups'] == 'Normal', 0, 1)\n",
    "y_valid['Type_Groups'] = np.where(y_valid['Type_Groups'] == 'Normal', 0, 1)\n",
    "y_train.groupby(['Type_Groups'])['Type_Groups'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "id": "ILBPmvDTDp1Z"
   },
   "outputs": [],
   "source": [
    "def lightgbm_classifier(X_train, X_test, X_valid, y_train, y_test, y_valid, useTrainCV=True, cv_folds=3, \n",
    "                        early_stopping_rounds=50, categorical_features=None, class_weight=None):\n",
    "    \"\"\"\n",
    "    # {'learning_rate': 0.1, 'max_depth': 5, 'min_child_weight': 3} 0.862920874517388\n",
    "    # {'colsample_bytree': 1.0, 'gamma': 0.2} 0.871\n",
    "    # {'gamma': 0.2, 'scale_pos_weight': 1} 0.8702009952422571\n",
    "    # {'subsample': 0.6} 0.864310306628855\n",
    "    \"\"\"\n",
    "    for c in categorical_features:\n",
    "        X_train[c] = X_train[c].astype('category')\n",
    "        X_test[c] = X_test[c].astype('category')\n",
    "        X_valid[c] = X_valid[c].astype('category')\n",
    "\n",
    "    alg = lgb.LGBMClassifier(#bagging_fraction=None,\n",
    "                                #max_bin=None,\n",
    "                                learning_rate=0.01,\n",
    "                                #num_iterations=None,\n",
    "                                #boosting='dart', #  Dropouts meet Multiple Additive Regression Trees\n",
    "                                #lambda_l1=None, \n",
    "                                #lambda_l2=None,\n",
    "                                #min_gain_to_split=None,\n",
    "                                objective='binary', #or multiclass,\n",
    "                                categorical_feature=categorical_features,\n",
    "                                scale_pos_weight = class_weight\n",
    "                                )\n",
    "    \n",
    "    if useTrainCV:\n",
    "        print(\"Start Feeding Data\")\n",
    "        rs_params = {\n",
    "        'bagging_fraction': (0.5, 0.6, 0.7, 0.8),\n",
    "        'bagging_freq': sp_randInt(5, 8),\n",
    "        'min_child_samples': (20, 25, 30, 35, 40),\n",
    "         #'min_child_weight': sp_randFloat(0.0005, 0.002),\n",
    "        'feature_fraction': sp_randFloat(0.5, 0.7),\n",
    "        'max_depth': sp_randInt(10, 13),\n",
    "        'learning_rate': (0.0001, 0.001, 0.01, 0.1),\n",
    "        'min_data_in_leaf': sp_randInt(90, 120),\n",
    "        'num_leaves': sp_randInt(1200, 1550)\n",
    "        }\n",
    "\n",
    "        # Initialize a RandomizedSearchCV object using 5-fold CV-\n",
    "        rs_cv = RandomizedSearchCV(estimator=alg, param_distributions=rs_params, cv = cv_folds, n_iter=10,verbose=0)\n",
    "\n",
    "        # Train on training data-\n",
    "        rs_cv.fit(X_train, y_train, eval_set=(X_test,y_test),verbose=0)\n",
    "        #print(rs_cv.best_params_)\n",
    "        alg.set_params(**rs_cv.best_params_)\n",
    "\n",
    "    print('OPTIMAL PARAMS OF THE MODEL')\n",
    "    print(alg.get_params())\n",
    "    print('Start Training')\n",
    "    alg.fit(X_train,\n",
    "            y_train,\n",
    "            eval_set=(X_valid,y_valid),\n",
    "            early_stopping_rounds=100,\n",
    "            eval_metric=\"binary_logloss\")\n",
    "\n",
    "\n",
    "    print(\"Start Predicting\")\n",
    "    predictions = alg.predict(X_valid)\n",
    "    pred_proba = alg.predict_proba(X_valid)[:, 1]\n",
    "    print('Confusion Matrix')\n",
    "    print(confusion_matrix(y_valid, predictions))\n",
    "    print(\"\\nResulting Metrics\")\n",
    "    print(\"Accuracy : %.4g\" % metrics.accuracy_score(y_valid, predictions))\n",
    "    print(\"AUC : %f\" % metrics.roc_auc_score(y_valid, pred_proba))\n",
    "    #print(\"F1 Score : %f\" % metrics.f1_score(y_valid, predictions))\n",
    "    precision, recall, fscore, support = score(y_valid, predictions)\n",
    "    print('precision: {}'.format(precision))\n",
    "    print('recall: {}'.format(recall))\n",
    "    print('fscore: {}'.format(fscore))\n",
    "    print('support: {}'.format(support))\n",
    "\n",
    "    feat_imp = alg.feature_importances_\n",
    "    feat = X_train.columns.tolist()\n",
    "    # clf.best_estimator_.booster().get_fscore()\n",
    "    res_df = pd.DataFrame({'Features': feat, 'Importance': feat_imp}).sort_values(by='Importance', ascending=False)\n",
    "    res_df.plot('Features', 'Importance', kind='bar', title='Feature Importance Plot Based on Split')\n",
    "    plt.ylabel('Feature Importance Score')\n",
    "    plt.show()\n",
    "\n",
    "    feat_imp = alg.booster_.feature_importance(importance_type = 'gain')\n",
    "    feat = X_train.columns.tolist()\n",
    "    # clf.best_estimator_.booster().get_fscore()\n",
    "    res_df = pd.DataFrame({'Features': feat, 'Importance': feat_imp}).sort_values(by='Importance', ascending=False)\n",
    "    res_df.plot('Features', 'Importance', kind='bar', title='Feature Importance Plot Based on Gain')\n",
    "    plt.ylabel('Feature Importance Score')\n",
    "    plt.show()\n",
    "    \n",
    "    #print(res_df)\n",
    "    #print(res_df[\"Features\"].tolist())\n",
    "    print('ROC CURVE')\n",
    "    fpr, tpr, thresholds = roc_curve(y_valid, pred_proba)\n",
    "    roc_auc = auc(fpr, tpr)\n",
    "    plt.title('Receiver Operating Characteristic')\n",
    "    plt.plot(fpr, tpr, label='AUC = %0.4f'% roc_auc)\n",
    "    plt.legend(loc='lower right')\n",
    "    plt.plot([0,1],[0,1],'r--')\n",
    "    plt.xlim([-0.001, 1])\n",
    "    plt.ylim([0, 1.001])\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.show();\n",
    "    return alg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "L0XjgbJh7KPI",
    "outputId": "f84f9f00-ad52-45e9-8005-e6f9fe46016f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start Feeding Data\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=98, min_child_samples=35 will be ignored. Current value: min_data_in_leaf=98\n",
      "[LightGBM] [Warning] feature_fraction is set=0.6048252104839297, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.6048252104839297\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=98, min_child_samples=35 will be ignored. Current value: min_data_in_leaf=98\n",
      "[LightGBM] [Warning] feature_fraction is set=0.6048252104839297, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.6048252104839297\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=98, min_child_samples=35 will be ignored. Current value: min_data_in_leaf=98\n",
      "[LightGBM] [Warning] feature_fraction is set=0.6048252104839297, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.6048252104839297\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=110, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=110\n",
      "[LightGBM] [Warning] feature_fraction is set=0.8435769421326961, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.8435769421326961\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.8, subsample=1.0 will be ignored. Current value: bagging_fraction=0.8\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=110, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=110\n",
      "[LightGBM] [Warning] feature_fraction is set=0.8435769421326961, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.8435769421326961\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.8, subsample=1.0 will be ignored. Current value: bagging_fraction=0.8\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=110, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=110\n",
      "[LightGBM] [Warning] feature_fraction is set=0.8435769421326961, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.8435769421326961\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.8, subsample=1.0 will be ignored. Current value: bagging_fraction=0.8\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=102, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=102\n",
      "[LightGBM] [Warning] feature_fraction is set=0.958684443290947, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.958684443290947\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=102, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=102\n",
      "[LightGBM] [Warning] feature_fraction is set=0.958684443290947, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.958684443290947\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=102, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=102\n",
      "[LightGBM] [Warning] feature_fraction is set=0.958684443290947, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.958684443290947\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=91, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=91\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9558185180119705, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9558185180119705\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=91, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=91\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9558185180119705, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9558185180119705\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=91, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=91\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9558185180119705, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9558185180119705\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "[LightGBM] [Fatal] Check failed: (feature_fraction) <= (1.0) at /tmp/pip-req-build-u3eu75cl/compile/src/io/config_auto.cpp, line 363 .\n",
      "\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "[LightGBM] [Fatal] Check failed: (feature_fraction) <= (1.0) at /tmp/pip-req-build-u3eu75cl/compile/src/io/config_auto.cpp, line 363 .\n",
      "\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "[LightGBM] [Fatal] Check failed: (feature_fraction) <= (1.0) at /tmp/pip-req-build-u3eu75cl/compile/src/io/config_auto.cpp, line 363 .\n",
      "\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=92, min_child_samples=40 will be ignored. Current value: min_data_in_leaf=92\n",
      "[LightGBM] [Warning] feature_fraction is set=1.076976370840204, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=1.076976370840204\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=92, min_child_samples=40 will be ignored. Current value: min_data_in_leaf=92\n",
      "[LightGBM] [Warning] feature_fraction is set=1.076976370840204, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=1.076976370840204\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=92, min_child_samples=40 will be ignored. Current value: min_data_in_leaf=92\n",
      "[LightGBM] [Warning] feature_fraction is set=1.076976370840204, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=1.076976370840204\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=119, min_child_samples=25 will be ignored. Current value: min_data_in_leaf=119\n",
      "[LightGBM] [Warning] feature_fraction is set=0.8205665175291902, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.8205665175291902\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.8, subsample=1.0 will be ignored. Current value: bagging_fraction=0.8\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=119, min_child_samples=25 will be ignored. Current value: min_data_in_leaf=119\n",
      "[LightGBM] [Warning] feature_fraction is set=0.8205665175291902, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.8205665175291902\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.8, subsample=1.0 will be ignored. Current value: bagging_fraction=0.8\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=119, min_child_samples=25 will be ignored. Current value: min_data_in_leaf=119\n",
      "[LightGBM] [Warning] feature_fraction is set=0.8205665175291902, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.8205665175291902\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.8, subsample=1.0 will be ignored. Current value: bagging_fraction=0.8\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=118, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=118\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9223983359313086, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9223983359313086\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=118, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=118\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9223983359313086, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9223983359313086\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=118, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=118\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9223983359313086, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9223983359313086\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=119, min_child_samples=25 will be ignored. Current value: min_data_in_leaf=119\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9816003388941108, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9816003388941108\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.5, subsample=1.0 will be ignored. Current value: bagging_fraction=0.5\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=119, min_child_samples=25 will be ignored. Current value: min_data_in_leaf=119\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9816003388941108, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9816003388941108\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.5, subsample=1.0 will be ignored. Current value: bagging_fraction=0.5\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=119, min_child_samples=25 will be ignored. Current value: min_data_in_leaf=119\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9816003388941108, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9816003388941108\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.5, subsample=1.0 will be ignored. Current value: bagging_fraction=0.5\n",
      "[LightGBM] [Warning] bagging_freq is set=5, subsample_freq=0 will be ignored. Current value: bagging_freq=5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=118, min_child_samples=35 will be ignored. Current value: min_data_in_leaf=118\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9233236722259122, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9233236722259122\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=118, min_child_samples=35 will be ignored. Current value: min_data_in_leaf=118\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9233236722259122, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9233236722259122\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=118, min_child_samples=35 will be ignored. Current value: min_data_in_leaf=118\n",
      "[LightGBM] [Warning] feature_fraction is set=0.9233236722259122, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.9233236722259122\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=6, subsample_freq=0 will be ignored. Current value: bagging_freq=6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "[LightGBM] [Fatal] Check failed: (feature_fraction) <= (1.0) at /tmp/pip-req-build-u3eu75cl/compile/src/io/config_auto.cpp, line 363 .\n",
      "\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "[LightGBM] [Fatal] Check failed: (feature_fraction) <= (1.0) at /tmp/pip-req-build-u3eu75cl/compile/src/io/config_auto.cpp, line 363 .\n",
      "\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "[LightGBM] [Fatal] Check failed: (feature_fraction) <= (1.0) at /tmp/pip-req-build-u3eu75cl/compile/src/io/config_auto.cpp, line 363 .\n",
      "\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/sklearn/model_selection/_validation.py:372: FitFailedWarning: \n",
      "6 fits failed out of a total of 30.\n",
      "The score on these train-test partitions for these parameters will be set to nan.\n",
      "If these failures are not expected, you can try to debug them by setting error_score='raise'.\n",
      "\n",
      "Below are more details about the failures:\n",
      "--------------------------------------------------------------------------------\n",
      "6 fits failed with the following error:\n",
      "Traceback (most recent call last):\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/sklearn/model_selection/_validation.py\", line 681, in _fit_and_score\n",
      "    estimator.fit(X_train, y_train, **fit_params)\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/sklearn.py\", line 895, in fit\n",
      "    callbacks=callbacks, init_model=init_model)\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/sklearn.py\", line 688, in fit\n",
      "    callbacks=callbacks, init_model=init_model)\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/engine.py\", line 228, in train\n",
      "    booster = Booster(params=params, train_set=train_set)\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py\", line 2229, in __init__\n",
      "    train_set.construct()\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py\", line 1472, in construct\n",
      "    categorical_feature=self.categorical_feature, params=self.params)\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py\", line 1270, in _lazy_init\n",
      "    self.__init_from_np2d(data, params_str, ref_dataset)\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py\", line 1320, in __init_from_np2d\n",
      "    ctypes.byref(self.handle)))\n",
      "  File \"/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py\", line 110, in _safe_call\n",
      "    raise LightGBMError(_LIB.LGBM_GetLastError().decode('utf-8'))\n",
      "lightgbm.basic.LightGBMError: Check failed: (feature_fraction) <= (1.0) at /tmp/pip-req-build-u3eu75cl/compile/src/io/config_auto.cpp, line 363 .\n",
      "\n",
      "\n",
      "  warnings.warn(some_fits_failed_message, FitFailedWarning)\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/sklearn/model_selection/_search.py:972: UserWarning: One or more of the test scores are non-finite: [0.85839207 0.95119783 0.95800111 0.85839207        nan 0.85839207\n",
      " 0.94877586 0.95752941 0.85839207        nan]\n",
      "  category=UserWarning,\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] min_data_in_leaf is set=91, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=91\n",
      "[LightGBM] [Warning] feature_fraction is set=1.0925216030437248, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=1.0925216030437248\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=91, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=91\n",
      "[LightGBM] [Warning] feature_fraction is set=1.0925216030437248, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=1.0925216030437248\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=91, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=91\n",
      "[LightGBM] [Warning] feature_fraction is set=1.0925216030437248, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=1.0925216030437248\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.6, subsample=1.0 will be ignored. Current value: bagging_fraction=0.6\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n",
      "[LightGBM] [Warning] min_data_in_leaf is set=102, min_child_samples=20 will be ignored. Current value: min_data_in_leaf=102\n",
      "[LightGBM] [Warning] feature_fraction is set=0.958684443290947, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.958684443290947\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.7, subsample=1.0 will be ignored. Current value: bagging_fraction=0.7\n",
      "[LightGBM] [Warning] bagging_freq is set=7, subsample_freq=0 will be ignored. Current value: bagging_freq=7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "OPTIMAL PARAMS OF THE MODEL\n",
      "{'boosting_type': 'gbdt', 'class_weight': None, 'colsample_bytree': 1.0, 'importance_type': 'split', 'learning_rate': 0.1, 'max_depth': 11, 'min_child_samples': 20, 'min_child_weight': 0.0020456191751726163, 'min_split_gain': 0.0, 'n_estimators': 100, 'n_jobs': -1, 'num_leaves': 1503, 'objective': 'binary', 'random_state': None, 'reg_alpha': 0.0, 'reg_lambda': 0.0, 'silent': True, 'subsample': 1.0, 'subsample_for_bin': 200000, 'subsample_freq': 0, 'categorical_feature': ['wrong_fragment', 'is_host_login', 'is_guest_login'], 'scale_pos_weight': 20, 'bagging_fraction': 0.7, 'bagging_freq': 7, 'feature_fraction': 0.958684443290947, 'min_data_in_leaf': 102}\n",
      "Start Training\n",
      "[1]\tvalid_0's binary_logloss: 0.322729\n",
      "Training until validation scores don't improve for 100 rounds\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1224: UserWarning: categorical_feature keyword has been found in `params` and will be ignored.\n",
      "Please use categorical_feature argument of the Dataset constructor to pass this parameter.\n",
      "  .format(key))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_feature in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1433: UserWarning: Overriding the parameters from Reference Dataset.\n",
      "  _log_warning('Overriding the parameters from Reference Dataset.')\n",
      "/Users/onurboyar/opt/anaconda3/envs/chemical-vae/lib/python3.7/site-packages/lightgbm/basic.py:1245: UserWarning: categorical_column in param dict is overridden.\n",
      "  _log_warning('{} in param dict is overridden.'.format(cat_alias))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2]\tvalid_0's binary_logloss: 0.279461\n",
      "[3]\tvalid_0's binary_logloss: 0.248944\n",
      "[4]\tvalid_0's binary_logloss: 0.226445\n",
      "[5]\tvalid_0's binary_logloss: 0.208238\n",
      "[6]\tvalid_0's binary_logloss: 0.19364\n",
      "[7]\tvalid_0's binary_logloss: 0.182277\n",
      "[8]\tvalid_0's binary_logloss: 0.172074\n",
      "[9]\tvalid_0's binary_logloss: 0.163742\n",
      "[10]\tvalid_0's binary_logloss: 0.156812\n",
      "[11]\tvalid_0's binary_logloss: 0.151366\n",
      "[12]\tvalid_0's binary_logloss: 0.146876\n",
      "[13]\tvalid_0's binary_logloss: 0.142753\n",
      "[14]\tvalid_0's binary_logloss: 0.139481\n",
      "[15]\tvalid_0's binary_logloss: 0.136974\n",
      "[16]\tvalid_0's binary_logloss: 0.134955\n",
      "[17]\tvalid_0's binary_logloss: 0.133346\n",
      "[18]\tvalid_0's binary_logloss: 0.132046\n",
      "[19]\tvalid_0's binary_logloss: 0.131051\n",
      "[20]\tvalid_0's binary_logloss: 0.130285\n",
      "[21]\tvalid_0's binary_logloss: 0.129776\n",
      "[22]\tvalid_0's binary_logloss: 0.129187\n",
      "[23]\tvalid_0's binary_logloss: 0.128903\n",
      "[24]\tvalid_0's binary_logloss: 0.128773\n",
      "[25]\tvalid_0's binary_logloss: 0.128569\n",
      "[26]\tvalid_0's binary_logloss: 0.12833\n",
      "[27]\tvalid_0's binary_logloss: 0.128394\n",
      "[28]\tvalid_0's binary_logloss: 0.128331\n",
      "[29]\tvalid_0's binary_logloss: 0.12798\n",
      "[30]\tvalid_0's binary_logloss: 0.127879\n",
      "[31]\tvalid_0's binary_logloss: 0.127962\n",
      "[32]\tvalid_0's binary_logloss: 0.128099\n",
      "[33]\tvalid_0's binary_logloss: 0.128129\n",
      "[34]\tvalid_0's binary_logloss: 0.128448\n",
      "[35]\tvalid_0's binary_logloss: 0.12882\n",
      "[36]\tvalid_0's binary_logloss: 0.129237\n",
      "[37]\tvalid_0's binary_logloss: 0.129647\n",
      "[38]\tvalid_0's binary_logloss: 0.130094\n",
      "[39]\tvalid_0's binary_logloss: 0.130565\n",
      "[40]\tvalid_0's binary_logloss: 0.131014\n",
      "[41]\tvalid_0's binary_logloss: 0.131374\n",
      "[42]\tvalid_0's binary_logloss: 0.131551\n",
      "[43]\tvalid_0's binary_logloss: 0.131763\n",
      "[44]\tvalid_0's binary_logloss: 0.131981\n",
      "[45]\tvalid_0's binary_logloss: 0.13197\n",
      "[46]\tvalid_0's binary_logloss: 0.132046\n",
      "[47]\tvalid_0's binary_logloss: 0.132171\n",
      "[48]\tvalid_0's binary_logloss: 0.132303\n",
      "[49]\tvalid_0's binary_logloss: 0.132396\n",
      "[50]\tvalid_0's binary_logloss: 0.13204\n",
      "[51]\tvalid_0's binary_logloss: 0.131787\n",
      "[52]\tvalid_0's binary_logloss: 0.131704\n",
      "[53]\tvalid_0's binary_logloss: 0.131501\n",
      "[54]\tvalid_0's binary_logloss: 0.131268\n",
      "[55]\tvalid_0's binary_logloss: 0.131087\n",
      "[56]\tvalid_0's binary_logloss: 0.130892\n",
      "[57]\tvalid_0's binary_logloss: 0.130727\n",
      "[58]\tvalid_0's binary_logloss: 0.130651\n",
      "[59]\tvalid_0's binary_logloss: 0.130449\n",
      "[60]\tvalid_0's binary_logloss: 0.130325\n",
      "[61]\tvalid_0's binary_logloss: 0.130242\n",
      "[62]\tvalid_0's binary_logloss: 0.130177\n",
      "[63]\tvalid_0's binary_logloss: 0.130114\n",
      "[64]\tvalid_0's binary_logloss: 0.130052\n",
      "[65]\tvalid_0's binary_logloss: 0.130049\n",
      "[66]\tvalid_0's binary_logloss: 0.130012\n",
      "[67]\tvalid_0's binary_logloss: 0.129999\n",
      "[68]\tvalid_0's binary_logloss: 0.129979\n",
      "[69]\tvalid_0's binary_logloss: 0.129961\n",
      "[70]\tvalid_0's binary_logloss: 0.129983\n",
      "[71]\tvalid_0's binary_logloss: 0.129953\n",
      "[72]\tvalid_0's binary_logloss: 0.129901\n",
      "[73]\tvalid_0's binary_logloss: 0.129842\n",
      "[74]\tvalid_0's binary_logloss: 0.129759\n",
      "[75]\tvalid_0's binary_logloss: 0.129584\n",
      "[76]\tvalid_0's binary_logloss: 0.129479\n",
      "[77]\tvalid_0's binary_logloss: 0.129418\n",
      "[78]\tvalid_0's binary_logloss: 0.129293\n",
      "[79]\tvalid_0's binary_logloss: 0.129155\n",
      "[80]\tvalid_0's binary_logloss: 0.128917\n",
      "[81]\tvalid_0's binary_logloss: 0.128725\n",
      "[82]\tvalid_0's binary_logloss: 0.128606\n",
      "[83]\tvalid_0's binary_logloss: 0.128501\n",
      "[84]\tvalid_0's binary_logloss: 0.128422\n",
      "[85]\tvalid_0's binary_logloss: 0.128071\n",
      "[86]\tvalid_0's binary_logloss: 0.127812\n",
      "[87]\tvalid_0's binary_logloss: 0.127578\n",
      "[88]\tvalid_0's binary_logloss: 0.12748\n",
      "[89]\tvalid_0's binary_logloss: 0.127396\n",
      "[90]\tvalid_0's binary_logloss: 0.127209\n",
      "[91]\tvalid_0's binary_logloss: 0.127063\n",
      "[92]\tvalid_0's binary_logloss: 0.126858\n",
      "[93]\tvalid_0's binary_logloss: 0.12665\n",
      "[94]\tvalid_0's binary_logloss: 0.126605\n",
      "[95]\tvalid_0's binary_logloss: 0.126416\n",
      "[96]\tvalid_0's binary_logloss: 0.126372\n",
      "[97]\tvalid_0's binary_logloss: 0.126278\n",
      "[98]\tvalid_0's binary_logloss: 0.126256\n",
      "[99]\tvalid_0's binary_logloss: 0.126104\n",
      "[100]\tvalid_0's binary_logloss: 0.125977\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[100]\tvalid_0's binary_logloss: 0.125977\n",
      "Start Predicting\n",
      "Confusion Matrix\n",
      "[[30121  1514]\n",
      " [   17  5096]]\n",
      "\n",
      "Resulting Metrics\n",
      "Accuracy : 0.9583\n",
      "AUC : 0.994558\n",
      "precision: [0.99943593 0.7709531 ]\n",
      "recall: [0.95214162 0.99667514]\n",
      "fscore: [0.97521571 0.86940203]\n",
      "support: [31635  5113]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC CURVE\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lgm=lightgbm_classifier(X_train, X_test, X_valid, y_train.values[:,0], y_test.values[:,0], y_valid.values[:,0],categorical_features=['wrong_fragment','is_host_login','is_guest_login'], class_weight=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "id": "E80OIjHIp-Iu"
   },
   "outputs": [],
   "source": [
    "class LGBMClassifier_GainFE(lgb.LGBMClassifier):\n",
    "    @property\n",
    "    def feature_importances_(self):\n",
    "        if self._n_features is None:\n",
    "            raise LGBMNotFittedError('No feature_importances found. Need to call fit beforehand.')\n",
    "        return self.booster_.feature_importance(importance_type='gain')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "id": "3wRi55m1qB3S"
   },
   "outputs": [],
   "source": [
    "clf2 = LGBMClassifier_GainFE(num_leaves= 15, max_depth=-1, \n",
    "                         random_state=314, \n",
    "                         silent=True, \n",
    "                         metric='None', \n",
    "                         n_jobs=4, \n",
    "                         n_estimators=1000,\n",
    "                         colsample_bytree=0.9,\n",
    "                         subsample=0.9,\n",
    "                         learning_rate=0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "3KhAmUe1qGyz",
    "outputId": "8e8c77e7-a4d2-41b1-8005-ddff562b53e7"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/sklearn/preprocessing/_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "/usr/local/lib/python3.7/dist-packages/sklearn/preprocessing/_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "LGBMClassifier_GainFE(colsample_bytree=0.9, metric='None', n_estimators=1000,\n",
       "                      n_jobs=4, num_leaves=15, random_state=314, subsample=0.9)"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf2.fit(X_train, y_train)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 312
    },
    "id": "n9GecHj6qem0",
    "outputId": "daffcf91-bbde-47de-9d2a-4f6b0f2a7e3f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fcf7abbc590>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from lightgbm import plot_importance\n",
    "plot_importance(lgm, importance_type='split')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "t2DIfzbDwt2T",
    "outputId": "f9f03b21-6000-41cb-da35-fa06166af434"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([3.90054541e+02, 8.80575069e+04, 1.14325519e+05, 0.00000000e+00,\n",
       "       0.00000000e+00, 5.25393974e-04, 7.82429062e-02, 0.00000000e+00,\n",
       "       6.12673932e+04, 0.00000000e+00, 0.00000000e+00, 7.05105010e-02,\n",
       "       5.79198937e+00, 8.89278017e-03, 1.07161002e-04, 8.91803029e-06,\n",
       "       0.00000000e+00, 0.00000000e+00, 1.66371241e-01, 2.17877023e+03,\n",
       "       1.65834631e+03, 2.63367189e+02, 1.38179920e+03, 2.06523890e+04,\n",
       "       7.08741984e+03, 1.09304856e+04, 2.25130631e+03, 1.94579584e+03,\n",
       "       2.79831328e+03, 5.85775506e+05, 3.06545442e+04, 7.52813725e+03,\n",
       "       3.56715270e+04, 2.76064867e+05, 2.38836527e+04, 4.37254228e+02,\n",
       "       9.38432260e+04, 1.03467943e+03])"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lgm.booster_.feature_importance(importance_type = 'gain')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 312
    },
    "id": "Wxy5NAaVq44A",
    "outputId": "3fb34d0a-ebef-4a8f-8877-47bc0f2e7bb4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fcf876c9fd0>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_importance(lgm, importance_type='gain')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kbNqiio6DAR-"
   },
   "source": [
    "## TabNet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "yIiiOAkWDLaI",
    "outputId": "2ecfb344-83cf-4a29-ed89-08a1f2b398b4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting pytorch_tabnet\n",
      "  Downloading https://files.pythonhosted.org/packages/44/dd/e24309700d20dfc75643d231a3093179652d309cefcadc338e1a20ff288e/pytorch_tabnet-3.0.0-py3-none-any.whl\n",
      "Requirement already satisfied: tqdm<5.0,>=4.36 in /usr/local/lib/python3.6/dist-packages (from pytorch_tabnet) (4.41.1)\n",
      "Requirement already satisfied: scikit_learn>0.21 in /usr/local/lib/python3.6/dist-packages (from pytorch_tabnet) (0.22.2.post1)\n",
      "Requirement already satisfied: numpy<2.0,>=1.17 in /usr/local/lib/python3.6/dist-packages (from pytorch_tabnet) (1.19.4)\n",
      "Requirement already satisfied: torch<2.0,>=1.2 in /usr/local/lib/python3.6/dist-packages (from pytorch_tabnet) (1.7.0+cu101)\n",
      "Requirement already satisfied: scipy>1.4 in /usr/local/lib/python3.6/dist-packages (from pytorch_tabnet) (1.4.1)\n",
      "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.6/dist-packages (from scikit_learn>0.21->pytorch_tabnet) (1.0.0)\n",
      "Requirement already satisfied: dataclasses in /usr/local/lib/python3.6/dist-packages (from torch<2.0,>=1.2->pytorch_tabnet) (0.8)\n",
      "Requirement already satisfied: future in /usr/local/lib/python3.6/dist-packages (from torch<2.0,>=1.2->pytorch_tabnet) (0.16.0)\n",
      "Requirement already satisfied: typing-extensions in /usr/local/lib/python3.6/dist-packages (from torch<2.0,>=1.2->pytorch_tabnet) (3.7.4.3)\n",
      "Installing collected packages: pytorch-tabnet\n",
      "Successfully installed pytorch-tabnet-3.0.0\n"
     ]
    }
   ],
   "source": [
    "!pip install pytorch_tabnet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "9mdtUD2dDBqE"
   },
   "outputs": [],
   "source": [
    "from pytorch_tabnet.tab_model import TabNetClassifier\n",
    "\n",
    "classifier = TabNetClassifier()\n",
    "classifier.fit(X_train=X_train, y_train=y_train)\n",
    "\n",
    "#predictions = classifier.predict_proba(X_test)[:,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "xWCl7CwwDD3o"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "inzva_week2_anomaly_detection_plus_lightgbm.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}