MalwareDetection_ExploratoryTerritory.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "MalwareDetection_ExploratoryTerritory.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nwNwJvqeNrvc",
        "colab_type": "text"
      },
      "source": [
        "# Microsoft Malware Detection\n",
        "## ExploratoryTerritory "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SyXGsfsoStzE",
        "colab_type": "code",
        "outputId": "7a8fe752-8a91-4329-de80-4badaef73632",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        }
      },
      "source": [
        "import _pickle as cPickle\n",
        "import calendar\n",
        "import gc\n",
        "import gzip\n",
        "import lightgbm as lgb\n",
        "import math\n",
        "import matplotlib.lines as mlines\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import pickle\n",
        "import plotly as py\n",
        "import plotly.graph_objs as go\n",
        "import seaborn as sns\n",
        "\n",
        "from datetime import datetime\n",
        "from keras import callbacks\n",
        "from keras.callbacks import LearningRateScheduler\n",
        "from keras.layers.convolutional import Conv1D\n",
        "from keras.layers.convolutional import MaxPooling1D\n",
        "from keras.layers import Dense\n",
        "from keras.layers import LSTM\n",
        "from keras.models import Sequential\n",
        "from keras.optimizers import Adam\n",
        "from numba import jit\n",
        "from pathlib import Path\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import OneHotEncoder, LabelEncoder\n",
        "from sklearn.impute import SimpleImputer\n",
        "from sklearn.metrics import roc_auc_score\n",
        "from tqdm import tqdm, tqdm_notebook"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k618izylUrMK",
        "colab_type": "text"
      },
      "source": [
        "Functions to facilitate pickling."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tHGG3UmUFNT3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def dumpPickle(fileName, content):\n",
        "    with gzip.open(fileName, 'wb') as f:\n",
        "      cPickle.dump(content, f, protocol=1)\n",
        "\n",
        "def loadPickle(fileName):    \n",
        "    with gzip.open(fileName, 'rb') as f:\n",
        "      loaded_object = cPickle.load(f)\n",
        "      return loaded_object\n",
        "    \n",
        "def pickleExists(fileName):\n",
        "    file = Path(fileName)\n",
        "    if file.is_file():\n",
        "        return True\n",
        "    return False"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RhlRSBRMPeA8",
        "colab_type": "text"
      },
      "source": [
        "**Setting up the Kaggle API.** \n",
        "<br>\n",
        "When prompted upload your kaggle API key. [Getting your kaggle API key.](https://medium.com/@yvettewu.dw/tutorial-kaggle-api-google-colaboratory-1a054a382de0) "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "biFEt8xsQRD4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Run this cell and select the kaggle.json file downloaded\n",
        "# from the Kaggle account settings page.\n",
        "from google.colab import files\n",
        "files.upload()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n1kZYgD-QUuk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Next, install the Kaggle API client.\n",
        "!pip install -q kaggle\n",
        "\n",
        "# The Kaggle API client expects this file to be in ~/.kaggle,\n",
        "# so move it there.\n",
        "!mkdir -p ~/.kaggle\n",
        "!cp kaggle.json ~/.kaggle/\n",
        "\n",
        "# This permissions change avoids a warning on Kaggle tool startup.\n",
        "!chmod 600 ~/.kaggle/kaggle.json"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7HodaOXheph2",
        "colab_type": "text"
      },
      "source": [
        "Set pickled = 1 if you want to download the relavent pickles and skip time consuming data cleaning and training."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LQY1yZGNelC9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "pickled = 1 # We have set pickled to 1 after capturing the ouputs of the notebook cells."
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "X4RJbjJxe85b",
        "colab_type": "text"
      },
      "source": [
        "**Pickled Objects** :  [Kaggle link](https://www.kaggle.com/rachanaj/da-pickles)\n",
        "\n",
        "<br>\n",
        "train_df.pkl : Pickle of training data after preprocessing. <br>\n",
        "test_df.pkl : Pickle of testing data after preprocessing. <br>\n",
        "LGBMModel.pkl : Pickle of LGBM model after training. <br>\n",
        "Downloads Pickles from kaggle if pickling is set to 1 : "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bFO_e0nMe8NA",
        "colab_type": "code",
        "outputId": "846ef3ae-7998-4b5e-cecb-4f2019fba3cf",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 133
        }
      },
      "source": [
        "if pickled:\n",
        "  !kaggle datasets download -d rachanaj/da-pickles\n",
        "  !unzip 'da-pickles.zip' "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading da-pickles.zip to /content\n",
            "100% 833M/836M [00:10<00:00, 71.2MB/s]\n",
            "100% 836M/836M [00:10<00:00, 85.0MB/s]\n",
            "Archive:  da-pickles.zip\n",
            "  inflating: LGBMModel.pkl           \n",
            "  inflating: test_df.pkl             \n",
            "  inflating: train_df.pkl            \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oIv1-UEXGMp8",
        "colab_type": "text"
      },
      "source": [
        "**Loading Dataframe from pickle if present. Takes about 1 minute.**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HawowfkXFm7q",
        "colab_type": "code",
        "outputId": "764c2588-7d8d-432f-9b6f-88413d5de7e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "source": [
        "%%time\n",
        "trainPickleName = 'train_df.pkl'\n",
        "testPickleName = 'test_df.pkl'\n",
        "pickled = pickleExists(trainPickleName) and pickleExists(testPickleName)\n",
        "if pickled:trainPickleName = 'train_df.pkl'\n",
        "testPickleName = 'test_df.pkl'\n",
        "pickled = pickleExists(trainPickleName) and pickleExists(testPickleName)\n",
        "if pickled:\n",
        "  train = loadPickle(trainPickleName)\n",
        "  test = loadPickle(testPickleName)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 31.8 s, sys: 5.62 s, total: 37.4 s\n",
            "Wall time: 43.2 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rCmvWEbFKqMn",
        "colab_type": "code",
        "outputId": "b9a4eb58-bf66-4dee-85eb-36577cc7c5be",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        }
      },
      "source": [
        "pickled"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cjosLM1mQigY",
        "colab_type": "text"
      },
      "source": [
        "##Loading the Data<br>\n",
        "**Data source for train and test data**: [Kaggle link for Microsoft Malware Prediction](https://www.kaggle.com/c/microsoft-malware-prediction/*data*)\n",
        "<br>\n",
        "\n",
        "\n",
        "*   Each row in this dataset corresponds to a system, uniquely identified by a MachineIdentifier.\n",
        "*   HasDetections is the target and indicates whether Malware was detected on the system.\n",
        "*   HasDetections is missing in the test dataset and must be predicted using the train dataset.\n",
        "\n",
        "<hr>\n",
        "\n",
        "**Data source for Antivirus Signature vs Timestamp** : [Kaggle link](https://www.kaggle.com/cdeotte/malware-timestamps)\n",
        "<br>\n",
        "This dataset consists of mappings from antivirus signature versions to timestamps. Antivirus signature version ('AvSigVersion') is updated approximately every 2 hours. 95\\% of user antiviruses regularly update their antivirus signature version making them a trustworthy timestamp for each dataset observation. This means that the antivirus signature version of a system when it was sampled can be mapped to the time at which the system was sampled. The timestamps from this dataset are provided by Microsoft. Microsoft has derived these timestamps in the manner explained above, by approximating sampling time from 'AvSigVersion' of the observation.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uFuQtB7bQXJv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  !kaggle competitions download -c microsoft-malware-prediction\n",
        "  !unzip 'train.csv.zip' \n",
        "  !unzip 'test.csv.zip' \n",
        "  !kaggle datasets download -d cdeotte/malware-timestamps\n",
        "  !unzip 'malware-timestamps.zip' "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iS3HqpbWRxA1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  Av_datedict = np.load('AvSigVersionTimestamps.npy', allow_pickle=True)\n",
        "  Av_datedict = Av_datedict[()]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O9uIQXvYwXdI",
        "colab_type": "text"
      },
      "source": [
        "**Optimized datatypes for loading a dataframe from train/test.csv.**\n",
        "<br>\n",
        "This also gives a good idea of the attributes of the data. A brief description of each attribute is [here](https://www.kaggle.com/c/microsoft-malware-prediction/data)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g5FHpBzKwD8-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "data_types = {\n",
        "        'MachineIdentifier':                                    'object',\n",
        "        'ProductName':                                          'category',\n",
        "        'EngineVersion':                                        'category',\n",
        "        'AppVersion':                                           'category',\n",
        "        'AvSigVersion':                                         'category',\n",
        "        'IsBeta':                                               'int8',\n",
        "        'RtpStateBitfield':                                     'float16',\n",
        "        'IsSxsPassiveMode':                                     'int8',\n",
        "        'DefaultBrowsersIdentifier':                            'float16',\n",
        "        'AVProductStatesIdentifier':                            'float32',\n",
        "        'AVProductsInstalled':                                  'float16',\n",
        "        'AVProductsEnabled':                                    'float16',\n",
        "        'HasTpm':                                               'int8',\n",
        "        'CountryIdentifier':                                    'int16',\n",
        "        'CityIdentifier':                                       'float32',\n",
        "        'OrganizationIdentifier':                               'float16',\n",
        "        'GeoNameIdentifier':                                    'float16',\n",
        "        'LocaleEnglishNameIdentifier':                          'int8',\n",
        "        'Platform':                                             'category',\n",
        "        'Processor':                                            'category',\n",
        "        'OsVer':                                                'category',\n",
        "        'OsBuild':                                              'int16',\n",
        "        'OsSuite':                                              'int16',\n",
        "        'OsPlatformSubRelease':                                 'category',\n",
        "        'OsBuildLab':                                           'category',\n",
        "        'SkuEdition':                                           'category',\n",
        "        'IsProtected':                                          'float16',\n",
        "        'AutoSampleOptIn':                                      'int8',\n",
        "        'PuaMode':                                              'category',\n",
        "        'SMode':                                                'float16',\n",
        "        'IeVerIdentifier':                                      'float16',\n",
        "        'SmartScreen':                                          'category',\n",
        "        'Firewall':                                             'float16',\n",
        "        'UacLuaenable':                                         'float32',\n",
        "        'Census_MDC2FormFactor':                                'category',\n",
        "        'Census_DeviceFamily':                                  'category',\n",
        "        'Census_OEMNameIdentifier':                             'float16',\n",
        "        'Census_OEMModelIdentifier':                            'float32',\n",
        "        'Census_ProcessorCoreCount':                            'float16',\n",
        "        'Census_ProcessorManufacturerIdentifier':               'float16',\n",
        "        'Census_ProcessorModelIdentifier':                      'float16',\n",
        "        'Census_ProcessorClass':                                'category',\n",
        "        'Census_PrimaryDiskTotalCapacity':                      'float32',\n",
        "        'Census_PrimaryDiskTypeName':                           'category',\n",
        "        'Census_SystemVolumeTotalCapacity':                     'float32',\n",
        "        'Census_HasOpticalDiskDrive':                           'int8',\n",
        "        'Census_TotalPhysicalRAM':                              'float32',\n",
        "        'Census_ChassisTypeName':                               'category',\n",
        "        'Census_InternalPrimaryDiagonalDisplaySizeInInches':    'float16',\n",
        "        'Census_InternalPrimaryDisplayResolutionHorizontal':    'float16',\n",
        "        'Census_InternalPrimaryDisplayResolutionVertical':      'float16',\n",
        "        'Census_PowerPlatformRoleName':                         'category',\n",
        "        'Census_InternalBatteryType':                           'category',\n",
        "        'Census_InternalBatteryNumberOfCharges':                'float32',\n",
        "        'Census_OSVersion':                                     'category',\n",
        "        'Census_OSArchitecture':                                'category',\n",
        "        'Census_OSBranch':                                      'category',\n",
        "        'Census_OSBuildNumber':                                 'int16',\n",
        "        'Census_OSBuildRevision':                               'int32',\n",
        "        'Census_OSEdition':                                     'category',\n",
        "        'Census_OSSkuName':                                     'category',\n",
        "        'Census_OSInstallTypeName':                             'category',\n",
        "        'Census_OSInstallLanguageIdentifier':                   'float16',\n",
        "        'Census_OSUILocaleIdentifier':                          'int16',\n",
        "        'Census_OSWUAutoUpdateOptionsName':                     'category',\n",
        "        'Census_IsPortableOperatingSystem':                     'int8',\n",
        "        'Census_GenuineStateName':                              'category',\n",
        "        'Census_ActivationChannel':                             'category',\n",
        "        'Census_IsFlightingInternal':                           'float16',\n",
        "        'Census_IsFlightsDisabled':                             'float16',\n",
        "        'Census_FlightRing':                                    'category',\n",
        "        'Census_ThresholdOptIn':                                'float16',\n",
        "        'Census_FirmwareManufacturerIdentifier':                'float16',\n",
        "        'Census_FirmwareVersionIdentifier':                     'float32',\n",
        "        'Census_IsSecureBootEnabled':                           'int8',\n",
        "        'Census_IsWIMBootEnabled':                              'float16',\n",
        "        'Census_IsVirtualDevice':                               'float16',\n",
        "        'Census_IsTouchEnabled':                                'int8',\n",
        "        'Census_IsPenCapable':                                  'int8',\n",
        "        'Census_IsAlwaysOnAlwaysConnectedCapable':              'float16',\n",
        "        'Wdft_IsGamer':                                         'float16',\n",
        "        'Wdft_RegionIdentifier':                                'float16',\n",
        "        'HasDetections':                                        'int8'\n",
        "}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XyJhvZqyR2mC",
        "colab_type": "text"
      },
      "source": [
        "**Loading the dataset into a dataframe:** <br>\n",
        "Takes about 5 minutes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DKqxWufDwpi9",
        "colab_type": "code",
        "outputId": "4f9e050d-2c52-4f52-a5f4-6d81ef3a6114",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "source": [
        "%%time\n",
        "if not(pickled): \n",
        "  train = pd.read_csv('train.csv', dtype = data_types, low_memory = True)\n",
        "  test  = pd.read_csv('test.csv',  dtype = data_types, low_memory = True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 2 µs, sys: 1 µs, total: 3 µs\n",
            "Wall time: 5.25 µs\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6NmXoF_uyTUk",
        "colab_type": "text"
      },
      "source": [
        "**Adding a new attribute 'Date'** that indicates the sampling timestamp of each observation. The timestamp of each observation is derived from its Antivirus Signature Version mapping."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_q1jS9PsyR51",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled): \n",
        "  train['Date'] = train['AvSigVersion'].map(Av_datedict)\n",
        "  test['Date'] = test['AvSigVersion'].map(Av_datedict)\n",
        "  train_date = train['Date']\n",
        "  test_date = test['Date']"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gezcg1405M7r",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Storing the target for the train dataset. \n",
        "target_df = train['HasDetections']\n",
        "\n",
        "# Storing the Machine Identifiers for the test data set\n",
        "test_machine_id = test['MachineIdentifier']\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IKsB_mGoxvq_",
        "colab_type": "text"
      },
      "source": [
        "# Data Exploration"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KiBjHXyDxx5N",
        "colab_type": "code",
        "outputId": "260555d8-703d-4419-87bb-0f79751bb53c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 384
        }
      },
      "source": [
        "train.head()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>ProductName</th>\n",
              "      <th>EngineVersion</th>\n",
              "      <th>AppVersion</th>\n",
              "      <th>AvSigVersion</th>\n",
              "      <th>IsBeta</th>\n",
              "      <th>RtpStateBitfield</th>\n",
              "      <th>IsSxsPassiveMode</th>\n",
              "      <th>AVProductStatesIdentifier</th>\n",
              "      <th>AVProductsInstalled</th>\n",
              "      <th>AVProductsEnabled</th>\n",
              "      <th>HasTpm</th>\n",
              "      <th>CountryIdentifier</th>\n",
              "      <th>CityIdentifier</th>\n",
              "      <th>OrganizationIdentifier</th>\n",
              "      <th>GeoNameIdentifier</th>\n",
              "      <th>LocaleEnglishNameIdentifier</th>\n",
              "      <th>Platform</th>\n",
              "      <th>Processor</th>\n",
              "      <th>OsVer</th>\n",
              "      <th>OsBuild</th>\n",
              "      <th>OsSuite</th>\n",
              "      <th>OsPlatformSubRelease</th>\n",
              "      <th>OsBuildLab</th>\n",
              "      <th>SkuEdition</th>\n",
              "      <th>IsProtected</th>\n",
              "      <th>AutoSampleOptIn</th>\n",
              "      <th>SMode</th>\n",
              "      <th>IeVerIdentifier</th>\n",
              "      <th>SmartScreen</th>\n",
              "      <th>Firewall</th>\n",
              "      <th>UacLuaenable</th>\n",
              "      <th>Census_MDC2FormFactor</th>\n",
              "      <th>Census_DeviceFamily</th>\n",
              "      <th>Census_OEMNameIdentifier</th>\n",
              "      <th>Census_OEMModelIdentifier</th>\n",
              "      <th>Census_ProcessorCoreCount</th>\n",
              "      <th>Census_ProcessorManufacturerIdentifier</th>\n",
              "      <th>Census_ProcessorModelIdentifier</th>\n",
              "      <th>Census_PrimaryDiskTotalCapacity</th>\n",
              "      <th>Census_PrimaryDiskTypeName</th>\n",
              "      <th>Census_SystemVolumeTotalCapacity</th>\n",
              "      <th>Census_HasOpticalDiskDrive</th>\n",
              "      <th>Census_TotalPhysicalRAM</th>\n",
              "      <th>Census_ChassisTypeName</th>\n",
              "      <th>Census_InternalPrimaryDiagonalDisplaySizeInInches</th>\n",
              "      <th>Census_InternalPrimaryDisplayResolutionHorizontal</th>\n",
              "      <th>Census_InternalPrimaryDisplayResolutionVertical</th>\n",
              "      <th>Census_PowerPlatformRoleName</th>\n",
              "      <th>Census_InternalBatteryNumberOfCharges</th>\n",
              "      <th>Census_OSVersion</th>\n",
              "      <th>Census_OSArchitecture</th>\n",
              "      <th>Census_OSBranch</th>\n",
              "      <th>Census_OSBuildNumber</th>\n",
              "      <th>Census_OSBuildRevision</th>\n",
              "      <th>Census_OSEdition</th>\n",
              "      <th>Census_OSSkuName</th>\n",
              "      <th>Census_OSInstallTypeName</th>\n",
              "      <th>Census_OSInstallLanguageIdentifier</th>\n",
              "      <th>Census_OSUILocaleIdentifier</th>\n",
              "      <th>Census_OSWUAutoUpdateOptionsName</th>\n",
              "      <th>Census_IsPortableOperatingSystem</th>\n",
              "      <th>Census_GenuineStateName</th>\n",
              "      <th>Census_ActivationChannel</th>\n",
              "      <th>Census_IsFlightsDisabled</th>\n",
              "      <th>Census_FlightRing</th>\n",
              "      <th>Census_ThresholdOptIn</th>\n",
              "      <th>Census_FirmwareManufacturerIdentifier</th>\n",
              "      <th>Census_FirmwareVersionIdentifier</th>\n",
              "      <th>Census_IsSecureBootEnabled</th>\n",
              "      <th>Census_IsWIMBootEnabled</th>\n",
              "      <th>Census_IsVirtualDevice</th>\n",
              "      <th>Census_IsTouchEnabled</th>\n",
              "      <th>Census_IsPenCapable</th>\n",
              "      <th>Census_IsAlwaysOnAlwaysConnectedCapable</th>\n",
              "      <th>Wdft_IsGamer</th>\n",
              "      <th>Wdft_RegionIdentifier</th>\n",
              "      <th>WeekNumber</th>\n",
              "      <th>Date</th>\n",
              "      <th>HasDetections</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>134.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1.0</td>\n",
              "      <td>206.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>21705.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>136.0</td>\n",
              "      <td>476940.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>299451.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>18.906250</td>\n",
              "      <td>1440.0</td>\n",
              "      <td>900.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.294967e+09</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2573.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>2018-08-20 21:11:00</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>6.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>4.0</td>\n",
              "      <td>172.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>108360.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>135.0</td>\n",
              "      <td>476940.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>102385.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>13.898438</td>\n",
              "      <td>1366.0</td>\n",
              "      <td>768.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.000000e+00</td>\n",
              "      <td>19.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>16.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>NOT_SET</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1416.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>6.0</td>\n",
              "      <td>18.0</td>\n",
              "      <td>2018-03-01 22:10:00</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>336.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>55.0</td>\n",
              "      <td>674.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>46.0</td>\n",
              "      <td>48.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>192.0</td>\n",
              "      <td>114473.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>113907.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>21.500000</td>\n",
              "      <td>1920.0</td>\n",
              "      <td>1080.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.294967e+09</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>185.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2018-08-13 23:35:00</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>25.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>31.0</td>\n",
              "      <td>21.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>26.0</td>\n",
              "      <td>22.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>159.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>63.0</td>\n",
              "      <td>238475.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>227116.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>18.500000</td>\n",
              "      <td>1366.0</td>\n",
              "      <td>768.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.294967e+09</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>17.0</td>\n",
              "      <td>17.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>NaN</td>\n",
              "      <td>3.0</td>\n",
              "      <td>31.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2018-08-17 03:01:00</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>18.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>40.0</td>\n",
              "      <td>14.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>2269.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>52.0</td>\n",
              "      <td>476940.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>101900.0</td>\n",
              "      <td>0</td>\n",
              "      <td>6144.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>14.000000</td>\n",
              "      <td>1366.0</td>\n",
              "      <td>768.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.000000e+00</td>\n",
              "      <td>15.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>17.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>130.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2018-08-14 12:33:00</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   ProductName  EngineVersion  ...                Date  HasDetections\n",
              "0          0.0            1.0  ... 2018-08-20 21:11:00              0\n",
              "1          0.0            6.0  ... 2018-03-01 22:10:00              0\n",
              "2          0.0            1.0  ... 2018-08-13 23:35:00              0\n",
              "3          0.0            1.0  ... 2018-08-17 03:01:00              1\n",
              "4          0.0            1.0  ... 2018-08-14 12:33:00              1\n",
              "\n",
              "[5 rows x 79 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zzJ5IqnTxohw",
        "colab_type": "code",
        "outputId": "fab154c4-5434-4f05-99f2-e558fa2298e7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        }
      },
      "source": [
        "print(train.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(8921483, 79)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pCx2BrdiyRLy",
        "colab_type": "text"
      },
      "source": [
        "### Checking the target."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BX26ELJ3yDwQ",
        "colab_type": "code",
        "outputId": "758fd18f-3765-442b-e3a1-dad0ef6e2794",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        }
      },
      "source": [
        "target_df.value_counts().plot.bar(color=['green','red'])\n",
        "plt.title('HasDetections(target)')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'HasDetections(target)')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEFCAYAAADNFLE8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HFzxRiPzya-4",
        "colab_type": "text"
      },
      "source": [
        "Shows that target is balanced."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "moO8uXPK0prh",
        "colab_type": "text"
      },
      "source": [
        "### Checking the attributes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WRcR45RDyJc0",
        "colab_type": "code",
        "outputId": "1b6efee5-75ec-4a24-b829-f1972f55b4b4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 407
        }
      },
      "source": [
        "stats = []\n",
        "for col in train.columns:\n",
        "    stats.append((col, train[col].nunique(), train[col].isnull().sum() * 100 / train.shape[0], train[col].value_counts(normalize=True, dropna=False).values[0] * 100, train[col].dtype))\n",
        "    \n",
        "stats_df = pd.DataFrame(stats, columns=['Feature', 'Unique_values', 'Percentage of missing values', 'Percentage of values in the biggest category', 'type'])\n",
        "stats_df.sort_values('Percentage of missing values', ascending=False)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>Feature</th>\n",
              "      <th>Unique_values</th>\n",
              "      <th>Percentage of missing values</th>\n",
              "      <th>Percentage of values in the biggest category</th>\n",
              "      <th>type</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>65</th>\n",
              "      <td>Census_ThresholdOptIn</td>\n",
              "      <td>2</td>\n",
              "      <td>63.524472</td>\n",
              "      <td>63.524472</td>\n",
              "      <td>float16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>69</th>\n",
              "      <td>Census_IsWIMBootEnabled</td>\n",
              "      <td>2</td>\n",
              "      <td>63.439038</td>\n",
              "      <td>63.439038</td>\n",
              "      <td>float16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>28</th>\n",
              "      <td>SmartScreen</td>\n",
              "      <td>10</td>\n",
              "      <td>35.610840</td>\n",
              "      <td>48.379782</td>\n",
              "      <td>float32</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>26</th>\n",
              "      <td>SMode</td>\n",
              "      <td>2</td>\n",
              "      <td>6.027686</td>\n",
              "      <td>93.928812</td>\n",
              "      <td>float16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>74</th>\n",
              "      <td>Wdft_IsGamer</td>\n",
              "      <td>2</td>\n",
              "      <td>3.401352</td>\n",
              "      <td>69.205344</td>\n",
              "      <td>float16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>27</th>\n",
              "      <td>IeVerIdentifier</td>\n",
              "      <td>248</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>43.556010</td>\n",
              "      <td>float32</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25</th>\n",
              "      <td>AutoSampleOptIn</td>\n",
              "      <td>2</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>99.997108</td>\n",
              "      <td>int8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23</th>\n",
              "      <td>SkuEdition</td>\n",
              "      <td>8</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>61.809690</td>\n",
              "      <td>float32</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>OsPlatformSubRelease</td>\n",
              "      <td>9</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>43.888735</td>\n",
              "      <td>float32</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>78</th>\n",
              "      <td>HasDetections</td>\n",
              "      <td>2</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>50.020731</td>\n",
              "      <td>int8</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>79 rows × 5 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                    Feature  ...     type\n",
              "65    Census_ThresholdOptIn  ...  float16\n",
              "69  Census_IsWIMBootEnabled  ...  float16\n",
              "28              SmartScreen  ...  float32\n",
              "26                    SMode  ...  float16\n",
              "74             Wdft_IsGamer  ...  float16\n",
              "..                      ...  ...      ...\n",
              "27          IeVerIdentifier  ...  float32\n",
              "25          AutoSampleOptIn  ...     int8\n",
              "23               SkuEdition  ...  float32\n",
              "21     OsPlatformSubRelease  ...  float32\n",
              "78            HasDetections  ...     int8\n",
              "\n",
              "[79 rows x 5 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EBPkKVEe03rd",
        "colab_type": "text"
      },
      "source": [
        "Two things to note\n",
        "\n",
        "\n",
        "*  PuaMode, Census_ProcessorClass, Census_IsFlightingInternal and Census_InternalBatteryType have 70%+ missing values, which means that these columns are useless and should be definitely dropped\n",
        "*  In DefaultBrowsersIdentifier column 95% values belong to one category, so this attribute might not be useful.\n",
        "<br>\n",
        "(These columns would have already been dropped if the dataframes were loaded from the pickle)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0PPUgriF13GS",
        "colab_type": "text"
      },
      "source": [
        "# Data Visualization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7HSsSw6n2GWe",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# function to plot data\n",
        "def plot_categorical_feature(col, only_bars=False, top_n=10, by_touch=False):\n",
        "    top_n = top_n if train[col].nunique() > top_n else train[col].nunique()\n",
        "    print(f\"{col} has {train[col].nunique()} unique values and type: {train[col].dtype}.\")\n",
        "    print(train[col].value_counts(normalize=True, dropna=False).head())\n",
        "    if not by_touch:\n",
        "        if not only_bars:\n",
        "            df = train.groupby([col]).agg({'HasDetections': ['count', 'mean']})\n",
        "            df = df.sort_values(('HasDetections', 'count'), ascending=False).head(top_n).sort_index()\n",
        "            data = [go.Bar(x=df.index, y=df['HasDetections']['count'].values, name='counts'),\n",
        "                    go.Scatter(x=df.index, y=df['HasDetections']['mean'], name='Detections rate', yaxis='y2')]\n",
        "\n",
        "            layout = go.Layout(dict(title = f\"Counts of {col} by top-{top_n} categories and mean target value\",\n",
        "                                xaxis = dict(title = f'{col}',\n",
        "                                             showgrid=False,\n",
        "                                             zeroline=False,\n",
        "                                             showline=False,),\n",
        "                                yaxis = dict(title = 'Counts',\n",
        "                                             showgrid=False,\n",
        "                                             zeroline=False,\n",
        "                                             showline=False,),\n",
        "                                yaxis2=dict(title='Detections rate', overlaying='y', side='right')),\n",
        "                           legend=dict(orientation=\"v\"))\n",
        "\n",
        "        else:\n",
        "            top_cat = list(train[col].value_counts(dropna=False).index[:top_n])\n",
        "            df0 = train.loc[(train[col].isin(top_cat)) & (train['HasDetections'] == 1), col].value_counts().head(10).sort_index()\n",
        "            df1 = train.loc[(train[col].isin(top_cat)) & (train['HasDetections'] == 0), col].value_counts().head(10).sort_index()\n",
        "            data = [go.Bar(x=df0.index, y=df0.values, name='Has Detections'),\n",
        "                    go.Bar(x=df1.index, y=df1.values, name='No Detections')]\n",
        "\n",
        "            layout = go.Layout(dict(title = f\"Counts of {col} by top-{top_n} categories\",\n",
        "                                xaxis = dict(title = f'{col}',\n",
        "                                             showgrid=False,\n",
        "                                             zeroline=False,\n",
        "                                             showline=False,),\n",
        "                                yaxis = dict(title = 'Counts',\n",
        "                                             showgrid=False,\n",
        "                                             zeroline=False,\n",
        "                                             showline=False,),\n",
        "                                ),\n",
        "                           legend=dict(orientation=\"v\"), barmode='group')\n",
        "        \n",
        "        py.offline.iplot(dict(data=data, layout=layout))\n",
        "        \n",
        "    else:\n",
        "        top_n = 10\n",
        "        top_cat = list(train[col].value_counts(dropna=False).index[:top_n])\n",
        "        df = train.loc[train[col].isin(top_cat)]\n",
        "\n",
        "        df1 = train.loc[train['Census_IsTouchEnabled'] == 1]\n",
        "        df0 = train.loc[train['Census_IsTouchEnabled'] == 0]\n",
        "\n",
        "        df0_ = df0.groupby([col]).agg({'HasDetections': ['count', 'mean']})\n",
        "        df0_ = df0_.sort_values(('HasDetections', 'count'), ascending=False).head(top_n).sort_index()\n",
        "        df1_ = df1.groupby([col]).agg({'HasDetections': ['count', 'mean']})\n",
        "        df1_ = df1_.sort_values(('HasDetections', 'count'), ascending=False).head(top_n).sort_index()\n",
        "        data1 = [go.Bar(x=df0_.index, y=df0_['HasDetections']['count'].values, name='Nontouch device counts'),\n",
        "                go.Scatter(x=df0_.index, y=df0_['HasDetections']['mean'], name='Detections rate for nontouch devices', yaxis='y2')]\n",
        "        data2 = [go.Bar(x=df1_.index, y=df1_['HasDetections']['count'].values, name='Touch device counts'),\n",
        "                go.Scatter(x=df1_.index, y=df1_['HasDetections']['mean'], name='Detections rate for touch devices', yaxis='y2')]\n",
        "\n",
        "        layout = go.Layout(dict(title = f\"Counts of {col} by top-{top_n} categories for nontouch devices\",\n",
        "                            xaxis = dict(title = f'{col}',\n",
        "                                         showgrid=False,\n",
        "                                         zeroline=False,\n",
        "                                         showline=False,\n",
        "                                         type='category'),\n",
        "                            yaxis = dict(title = 'Counts',\n",
        "                                         showgrid=False,\n",
        "                                         zeroline=False,\n",
        "                                         showline=False,),\n",
        "                                    yaxis2=dict(title='Detections rate', overlaying='y', side='right'),\n",
        "                            ),\n",
        "                       legend=dict(orientation=\"v\"), barmode='group')\n",
        "\n",
        "        py.offline.iplot(dict(data=data1, layout=layout))\n",
        "        layout['title'] = f\"Counts of {col} by top-{top_n} categories for touch devices\"\n",
        "        py.offline.iplot(dict(data=data2, layout=layout))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MW7KjvFa3lOf",
        "colab_type": "text"
      },
      "source": [
        "## Variables with high amount of unique values"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nuzcX4cW3riL",
        "colab_type": "text"
      },
      "source": [
        "### Engine Version"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wKJM__1X3wHF",
        "colab_type": "code",
        "outputId": "752ad13e-e115-4acf-c595-432a590b99ee",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "plot_categorical_feature('EngineVersion', by_touch=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "EngineVersion has 67 unique values and type: float32.\n",
            "0.0    0.430990\n",
            "1.0    0.412030\n",
            "5.0    0.029728\n",
            "7.0    0.023809\n",
            "6.0    0.018000\n",
            "Name: EngineVersion, dtype: float64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"268318d1-774d-4ce6-8f52-0da40671b8b2\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"268318d1-774d-4ce6-8f52-0da40671b8b2\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '268318d1-774d-4ce6-8f52-0da40671b8b2',\n",
              "                        [{\"name\": \"Nontouch device counts\", \"type\": \"bar\", \"x\": [0.0, 1.0, 2.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0], \"y\": [3357684, 3218411, 105560, 229047, 144420, 182968, 117522, 83962, 61917, 62107]}, {\"name\": \"Detections rate for nontouch devices\", \"type\": \"scatter\", \"x\": [0.0, 1.0, 2.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0], \"y\": [0.5006349614794007, 0.5597880444728781, 0.5033819628647215, 0.3485441852545547, 0.44295111480404376, 0.3076220978531765, 0.314111400418645, 0.4433076868107001, 0.44050906859182454, 0.529650441979165], \"yaxis\": \"y2\"}],\n",
              "                        {\"barmode\": \"group\", \"legend\": {\"orientation\": \"v\"}, \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Counts of EngineVersion by top-10 categories for nontouch devices\"}, \"xaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"EngineVersion\"}, \"type\": \"category\", \"zeroline\": false}, \"yaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"Counts\"}, \"zeroline\": false}, \"yaxis2\": {\"overlaying\": \"y\", \"side\": \"right\", \"title\": {\"text\": \"Detections rate\"}}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('268318d1-774d-4ce6-8f52-0da40671b8b2');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"0fb5af85-76af-4c93-b3ac-735a6d56f4ae\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"0fb5af85-76af-4c93-b3ac-735a6d56f4ae\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '0fb5af85-76af-4c93-b3ac-735a6d56f4ae',\n",
              "                        [{\"name\": \"Touch device counts\", \"type\": \"bar\", \"x\": [0.0, 1.0, 2.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0], \"y\": [487383, 457504, 14735, 36171, 16165, 29440, 18954, 9964, 8728, 6609]}, {\"name\": \"Detections rate for touch devices\", \"type\": \"scatter\", \"x\": [0.0, 1.0, 2.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0], \"y\": [0.4303986802986563, 0.5027781177869484, 0.4141839158466237, 0.3137872881590224, 0.3854005567584287, 0.2703804347826087, 0.2826843937955049, 0.4088719389803292, 0.4046746104491292, 0.48479346345891966], \"yaxis\": \"y2\"}],\n",
              "                        {\"barmode\": \"group\", \"legend\": {\"orientation\": \"v\"}, \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Counts of EngineVersion by top-10 categories for touch devices\"}, \"xaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"EngineVersion\"}, \"type\": \"category\", \"zeroline\": false}, \"yaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"Counts\"}, \"zeroline\": false}, \"yaxis2\": {\"overlaying\": \"y\", \"side\": \"right\", \"title\": {\"text\": \"Detections rate\"}}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('0fb5af85-76af-4c93-b3ac-735a6d56f4ae');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tQ9tv5oh33Uo",
        "colab_type": "text"
      },
      "source": [
        "**We** can see that two categories take 84% of all values. The difference in detection rates is noticable. Other categories have different detection rate, but it is simply due to the low number of samples in them. Patterns for touch and non-touch devices are quite similar."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "llqJ4EUv37v2",
        "colab_type": "text"
      },
      "source": [
        "### App Version"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OGG3z5ce3_C1",
        "colab_type": "code",
        "outputId": "a2a913bd-ce96-4463-ad0b-f52a93c94989",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 658
        }
      },
      "source": [
        "plot_categorical_feature('AppVersion')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "AppVersion has 107 unique values and type: float32.\n",
            "0.0    0.576050\n",
            "3.0    0.095380\n",
            "4.0    0.040338\n",
            "6.0    0.030539\n",
            "5.0    0.028837\n",
            "Name: AppVersion, dtype: float64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"8adba7cc-6d1c-49aa-99e6-6cedd842d264\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"8adba7cc-6d1c-49aa-99e6-6cedd842d264\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '8adba7cc-6d1c-49aa-99e6-6cedd842d264',\n",
              "                        [{\"name\": \"counts\", \"type\": \"bar\", \"x\": [0.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0], \"y\": [5139224, 850929, 359871, 257270, 272455, 205480, 203525, 226501, 235032, 194699]}, {\"name\": \"Detections rate\", \"type\": \"scatter\", \"x\": [0.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0], \"y\": [0.5303851320744144, 0.4852143950905422, 0.4573222071242195, 0.5217942239670386, 0.5018480115982455, 0.4863052365193693, 0.444495762191377, 0.42864711414077644, 0.3241048027502638, 0.3169867333679166], \"yaxis\": \"y2\"}],\n",
              "                        {\"legend\": {\"orientation\": \"v\"}, \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Counts of AppVersion by top-10 categories and mean target value\"}, \"xaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"AppVersion\"}, \"zeroline\": false}, \"yaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"Counts\"}, \"zeroline\": false}, \"yaxis2\": {\"overlaying\": \"y\", \"side\": \"right\", \"title\": {\"text\": \"Detections rate\"}}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('8adba7cc-6d1c-49aa-99e6-6cedd842d264');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9cDjRNuIlTAy",
        "colab_type": "text"
      },
      "source": [
        "###Census_InternalPrimaryDiagonalDisplaySizeInInches"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XF2N93LKlg5r",
        "colab_type": "code",
        "outputId": "fdc89576-f7e2-4416-e0cf-288faa58c112",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "plot_categorical_feature('Census_InternalPrimaryDiagonalDisplaySizeInInches', True, by_touch=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Census_InternalPrimaryDiagonalDisplaySizeInInches has 785 unique values and type: float16.\n",
            "15.500000    0.346867\n",
            "13.898438    0.106717\n",
            "14.000000    0.060803\n",
            "11.601562    0.035799\n",
            "21.500000    0.030862\n",
            "Name: Census_InternalPrimaryDiagonalDisplaySizeInInches, dtype: float64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"37b005ab-1de0-44ae-9ee2-f4413d653f53\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"37b005ab-1de0-44ae-9ee2-f4413d653f53\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '37b005ab-1de0-44ae-9ee2-f4413d653f53',\n",
              "                        [{\"name\": \"Nontouch device counts\", \"type\": \"bar\", \"x\": [11.6015625, 13.296875, 13.8984375, 14.0, 15.5, 15.6015625, 17.296875, 18.5, 21.5, 23.0], \"y\": [170745, 207644, 867319, 495582, 2899500, 155365, 169003, 211832, 262232, 170008]}, {\"name\": \"Detections rate for nontouch devices\", \"type\": \"scatter\", \"x\": [11.6015625, 13.296875, 13.8984375, 14.0, 15.5, 15.6015625, 17.296875, 18.5, 21.5, 23.0], \"y\": [0.4517965386980585, 0.44932673229180714, 0.5239686897208524, 0.507262168521052, 0.510435937230557, 0.5125414346860618, 0.5034585184878375, 0.5377563352090335, 0.5414861649226639, 0.5293397957743166], \"yaxis\": \"y2\"}],\n",
              "                        {\"barmode\": \"group\", \"legend\": {\"orientation\": \"v\"}, \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Counts of Census_InternalPrimaryDiagonalDisplaySizeInInches by top-10 categories for nontouch devices\"}, \"xaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"Census_InternalPrimaryDiagonalDisplaySizeInInches\"}, \"type\": \"category\", \"zeroline\": false}, \"yaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"Counts\"}, \"zeroline\": false}, \"yaxis2\": {\"overlaying\": \"y\", \"side\": \"right\", \"title\": {\"text\": \"Detections rate\"}}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('37b005ab-1de0-44ae-9ee2-f4413d653f53');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"7077195e-9d03-4073-a8a6-a20b050b7780\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"7077195e-9d03-4073-a8a6-a20b050b7780\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '7077195e-9d03-4073-a8a6-a20b050b7780',\n",
              "                        [{\"name\": \"Touch device counts\", \"type\": \"bar\", \"x\": [10.0, 10.1015625, 11.6015625, 12.0, 12.296875, 13.203125, 13.296875, 13.8984375, 14.0, 15.5], \"y\": [28908, 109868, 148631, 29168, 58831, 70543, 36150, 84759, 46868, 195065]}, {\"name\": \"Detections rate for touch devices\", \"type\": \"scatter\", \"x\": [10.0, 10.1015625, 11.6015625, 12.0, 12.296875, 13.203125, 13.296875, 13.8984375, 14.0, 15.5], \"y\": [0.3284557907845579, 0.29774820693923615, 0.4474100288634269, 0.4355458036204059, 0.4910506365691557, 0.517514140311583, 0.4799170124481328, 0.5062825186705836, 0.5048647264658189, 0.5210211980621844], \"yaxis\": \"y2\"}],\n",
              "                        {\"barmode\": \"group\", \"legend\": {\"orientation\": \"v\"}, \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Counts of Census_InternalPrimaryDiagonalDisplaySizeInInches by top-10 categories for touch devices\"}, \"xaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"Census_InternalPrimaryDiagonalDisplaySizeInInches\"}, \"type\": \"category\", \"zeroline\": false}, \"yaxis\": {\"showgrid\": false, \"showline\": false, \"title\": {\"text\": \"Counts\"}, \"zeroline\": false}, \"yaxis2\": {\"overlaying\": \"y\", \"side\": \"right\", \"title\": {\"text\": \"Detections rate\"}}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('7077195e-9d03-4073-a8a6-a20b050b7780');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Mg_EIKfa5417",
        "colab_type": "text"
      },
      "source": [
        "## Time Series Aspect"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vR17hui-1vZ9",
        "colab_type": "text"
      },
      "source": [
        "Malware risk detection is in essence a time series problem, with the sampling date of each data point greatly influencing the some of the system's properties.\n",
        "<br>\n",
        "Excerpt from the [Data Description](https://www.kaggle.com/c/microsoft-malware-prediction/data):  \n",
        "  \n",
        "\n",
        "> *Malware detection is inherently a time-series problem, but it is made complicated by the introduction of new machines, machines that come online and offline, machines that receive patches, machines that receive new operating systems, etc*.\n",
        "\n",
        "\n",
        "Thus, it would make sense to explore the relationship between sampling time and malware detection rate."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VaInayhW7NUs",
        "colab_type": "text"
      },
      "source": [
        "Defined below is a function to plot density of a attribute value vs time.\n",
        "<br>\n",
        "For the training data : We will be using this to plot detection density (i.e. malware detection rate vs time) as well as sampling density.\n",
        "<br>\n",
        "For the test data we will just be plotting sampling density.\n",
        "\n",
        "<br>\n",
        "\n",
        "A new Dummy attribute with value 1 for all observations is added. This dummy attribute will helpful in calculating sampling density.\n",
        "The number of observations with 'dummy' = 1 and 'Date' belonging to a certain interval of time will estimate sampling density"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kXdWP3NZ7Hzd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def TimeSeriesPlot(\n",
        "    data_frame,              # Data frame for the plot.\n",
        "    column,                  # Column for rate, left y-axis.\n",
        "    target='HasDetections',  # Malware detection rate will be plotted in the right y axis.\n",
        "    title = ''):\n",
        "    \n",
        "    start = datetime(2018,1,1) # Leftmost x-axis value. Start date of observations.\n",
        "    end = datetime(2018,12,1)  # Rightmost x-axis value. End date of observations.\n",
        "    \n",
        "    top_left=5\n",
        "    top_right=4\n",
        "\n",
        "    value_counts = data_frame[column].value_counts(dropna = False)\n",
        "    value_counts_dict = value_counts.to_dict()\n",
        "\n",
        "    nm = value_counts.index.values\n",
        "    threshold = 0.99 * len(data_frame)\n",
        "    sum_values = 0\n",
        "    lnn2 = 0\n",
        "\n",
        "    for x in nm:\n",
        "        lnn2 += 1\n",
        "        try:\n",
        "           sum_values += value_counts_dict[x]\n",
        "        except:\n",
        "           sum_values += cv[x]\n",
        "        if sum_values > threshold:\n",
        "            break\n",
        "    \n",
        "    top_left = min(top_left, len(nm))\n",
        "    top_right = min(top_right, len(nm), lnn2, top_left)\n",
        "\n",
        "    # Calculate rate within each time interval\n",
        "    timeInterval = (end - start).days * 24 * 3600 + (end - start).seconds\n",
        "    size = timeInterval // (3600 * (7 * 24)) + 5\n",
        "    data_counts = np.zeros([size, 2 * top_left + 1], dtype = float)\n",
        "    idx = 0\n",
        "    idx2 = {}\n",
        "\n",
        "    for i in range(top_left):\n",
        "        idx2[nm[i]] = i + 1\n",
        "    low = start\n",
        "    high = add_time(start)\n",
        "    data_times = [low + (high - low) / 2]\n",
        "    while low < end:\n",
        "        data_slice = data_frame[(data_frame['Date'] < high) & (data_frame['Date'] >= low)]\n",
        "        data_counts[idx, 0] = len(data_slice)\n",
        "        for key in idx2:\n",
        "            if nan_check(key): \n",
        "              data_slice2 = data_slice[data_slice[column].isna()]\n",
        "            else: \n",
        "              data_slice2 = data_slice[data_slice[column] == key]\n",
        "            data_counts[idx, idx2[key]] = len(data_slice2)\n",
        "            if target in data_frame:\n",
        "                data_counts[idx, top_left + idx2[key]] = data_slice2['HasDetections'].mean()\n",
        "        low = high\n",
        "        high = add_time(high)\n",
        "        data_times.append(low + (high - low) / 2)\n",
        "        idx += 1\n",
        "\n",
        "    fig = plt.figure(1,figsize=(15,3))\n",
        "    ax3 = fig.add_subplot(1,1,1)\n",
        "    lines = []; labels = []\n",
        "    for i in range(top_left):\n",
        "        tmp, = ax3.plot(data_times,data_counts[0 : idx+1, i + 1], 'green')\n",
        "        lines.append(tmp)\n",
        "        labels.append(str(nm[i]))\n",
        "\n",
        "    ax3.spines['left'].set_color('green')\n",
        "    ax3.yaxis.label.set_color('green')\n",
        "    ax3.tick_params(axis = 'y', colors = 'green')\n",
        "\n",
        "    if column != 'dummy': \n",
        "      ax3.set_ylabel('Category Density', color = 'green')\n",
        "    else: \n",
        "      ax3.set_ylabel('Sampling rate', color = 'green')\n",
        "    ax3.set_yticklabels([])\n",
        "\n",
        "    if target in data_frame:\n",
        "        ax4 = ax3.twinx()\n",
        "        for i in range(top_right):\n",
        "            tmp = ax4.plot(data_times, data_counts[0 : idx + 1, i + 1 + top_left],'black')\n",
        "            lines.append(tmp)\n",
        "        ax4.spines['left'].set_color('green')\n",
        "        ax4.set_ylabel('Detection Rate', color='k')\n",
        "    black_line = mlines.Line2D([], [], color='black')\n",
        "    green_line = mlines.Line2D([], [], color='green')\n",
        "    plt.legend([green_line,black_line], ['Sampling rate', 'Detection Rate'])\n",
        "    if title!='':\n",
        "      plt.title(title)\n",
        "    plt.show()\n",
        "\n",
        "# Function to increment datetime   \n",
        "def add_time(sdate):\n",
        "    months=0\n",
        "    days=7\n",
        "    hours=0\n",
        "    month = sdate.month -1 + months\n",
        "    year = sdate.year + month // 12\n",
        "    month = month % 12 + 1\n",
        "    day = sdate.day + days\n",
        "    if day>calendar.monthrange(year,month)[1]:\n",
        "        day -= calendar.monthrange(year,month)[1]\n",
        "        month += 1\n",
        "        if month > 12:\n",
        "            month = 1\n",
        "            year += 1\n",
        "    hour = sdate.hour + hours\n",
        "    if hour > 23:\n",
        "        hour = 0\n",
        "        day += 1\n",
        "        if day > calendar.monthrange(year,month)[1]:\n",
        "            day -= calendar.monthrange(year,month)[1]\n",
        "            month += 1\n",
        "            if month > 12:\n",
        "                month = 1\n",
        "                year += 1\n",
        "    return datetime(year, month, day, hour, sdate.minute)\n",
        "\n",
        "def nan_check(x):\n",
        "    if isinstance(x,float):\n",
        "        if math.isnan(x):\n",
        "            return True\n",
        "    return False\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_mbaCn4xA3aD",
        "colab_type": "code",
        "outputId": "c5ed39ba-3cee-4d4d-98be-1c6122af0a86",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 437
        }
      },
      "source": [
        "# Dummy attribute will helpful in calculating sampling density\n",
        "# The number of observations with dummy = 1 will be measured in an interval of time to estimate sampling density.\n",
        "train['dummy'] = 1\n",
        "test['dummy'] = 1\n",
        "TimeSeriesPlot(train, 'dummy', title = 'Train Data - Sampling density versus Time and Detection density versus Time')\n",
        "TimeSeriesPlot(test, 'dummy', title = 'Test Data - Sampling density versus Time')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA40AAADSCAYAAADuZCWGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nOzdeXhNV/fA8e8SQiKaVFDUPNUYMcTc\nitb4tobQolRbv9JXddRSdCIdUapFJx1eVE1VNQ+tORRFzWOVIMQsiCQisX9/nJP0JhK5iSQ30fV5\nnvvImfZe59xzrrvu3mcfMcaglFJKKaWUUkqlJI+rA1BKKaWUUkoplXNp0qiUUkoppZRSKlWaNCql\nlFJKKaWUSpUmjUoppZRSSimlUqVJo1JKKaWUUkqpVGnSqJRSSimllFIqVZo0KpVDiYibiESKSBlX\nx3KnE5GpIjLc/jtQRPZkU73vi8ikLK7jVxHpmZV15EYi8raIfOXqONJLRNaJyNOujiMjsuuYi0gl\nEcnS54mJyFMisiQr68iNRKSCiES6Og6lVObTpFGpTGIneAmvGyIS7TCd7i/txph4Y4yXMeZYBmKp\nJCLGof5TIrJARB5KRxl9RGR1eutODxEJEpEdInJZRM6JyApXJ8nGmNXGmBqujCEzGWNaG2N+hOx5\nT3MChx9cUrseuxlj3jPG9HN1rJnJ/hHiuohcsV8HRGSciBRPRxlhIhKYCbG0FJFQx3l30jE3xkw2\nxrQDEJG89udtOddGlfVE5FuH6yjWPt8SphcYYw4bY7xcHadSKvNp0qhUJrETPC/7P8xjQHuHeT8m\nX19E8mZXTEAdYCUwX0SeyOp6nSEi9wH/A14GvIHywJfADVfGpbJeVp/7Dj+4JJz/J4F2DvNmZmX9\nLvajMaYQ4At0AUoDW0TkHteGpbKaiOQRkSz9XmeM6eNwXY3COt8Srqv2WVm3Usq1NGlUKpvYrQAz\nRWS6iFwBnhCRxiKyUUQiRCTcbhXIZ6+f5NdruwvlOBFZYrcibBCR8s7UbYwJN8aMBd4DRomI2GW+\nJSKH7fL2iEgHe34tYAJwv/0L8jl7fgcR2W63DB4Tkbdv45DUAQ7ZLXvGGHPFGDPbGBNm1+XMsXlO\nRP624x8mIpXtbS7bxzlh/ZYiEioi74jIeRE5IiLdUwoqeQuJ3fLyqojsEpFLdrn5HZYPtVtyT4hI\n31u1OIjVdSvEjncZ1hd7x+VNHfZ5u4g84LBsnYgEi8jv9vZLRaSwvcxTRKbZ+xYhIn+ISBGH7Z5O\n6T21j/FJxy+aItJVRLamEHtTex8d131MRP60/84jIm/Y78c5EZkhInfbyxJavnuLyDHg1zRiTtLa\nJQ7deG+1XXokKzMhvqftui/Y72VD+32PEJHPkm3fR0T2i8hFsa7J0qnUk0dEZtvnSISIrBaRag7L\nb3ldi0hbsVoML9kxiDP7Z4yJNcbsBh4DIoABDmV2EKuFP8I+P2ra86cDJYEl9jnyqj3/Vuelr4hM\nEusavSgiP4uIN7AAKCP/tEIVk2TdscXqabDHLnelWD8kJSy75XWX7Bi7ichY+5w4DLRNttxHRP5n\nxxgmIu8mnMf2+7jG3j5CrM/D1g7bPiPWZ8cVe1l3h+1W26uttf/dY+9rF/vcaOdQTn77+NRKIf6/\nRKStw7S7fQ76OXH814nIeyKyAbhqH/PUYk5+/JN0401tu/RIocx19vHeKCJXRWSufc5MF+tzepM4\n9C4Rkeoistze//0i0iW9MSilsogxRl/60lcmv4BQoGWyee8DsUB7rB9sPIAAoCGQF6gAHAResNfP\nCxignD09FTgH1AfyATOBqanUX8m6vG+aX8Uus7I93RUoYcfTA4gE7rGX9QFWJ9v+QaCGvX5tO55H\nMniMKgPXgDFAC6BgsuXOHJs5QCHAzz62vwHlgLuB/UBPe/2WQBzwMZDf3o8ooJLDsR3usG6oQxxh\nwEagOFaSdxDoYy97BKsVqxpQEJju+J6lsM+bHWIItI/3JHtZaeA80MY+vm3t4+trL18H/GUfN08g\nBHjfXvY8MNc+p9zsc8TLYbunb/GeHgBaOUwvAF5OIXbBOq9bOMz7BRho//0asB64FygAfAv84Hg+\nYrUse9px3irmMCAw2bUzKa19vcW5lqS8FMpMiG+C/d78B4i2968oUMp+b5ra63exj9t9WOficCAk\nlbrzAE9jnacF7Dq2OCxP9boGitnnSJC9bBDWefx0KnUl7lOy+R8C6x2uq9P2v27A/wF/A+6pHPu0\nzstlwDSsay4f8EBK11EKx7yavW8P2tu9YR/TfGlddyns3wvAHvt98sVK4ozD8gXAF1jn3j3AVuAZ\nh2viun0c3IAXgeP2sruAS/zzeVkCqJ78WiLZZ7U97w2sVriE6S7AtlTifxeY7DDdEdidjs+FUPt4\n5ksj5iTnBw7/T9xqu1tcVzedbyT7v8eO7wDWZ3jC5/IBrM/8vPa58429rhdwAnjSXlbP3vf7bhWH\nvvSlr+x5aUujUtlrnTFmgTHmhjEm2hiz2RizyRgTZ4w5DEwEmt9i+9nGmC3GmOvAj4B/Ous/af9b\nGMAYM8tYrZA3jDHTsL581E9tY2PMSmPMHnv9HcCMNOJNlTHmL6wvDmWAn4DzIvK9iHjay505NiON\n1UK5E9gHLDXGhBpjLmJ9ma3jsO4NYJgx5poxZiWwFKsVxhmfGmNOGWPOAwv557h3Bb4zxuwzxlwF\nglMrQEQqYCXaCTGsBhY7rPIkMN8Ys8w+vkuBHSRtNfnOGPOXMSbKPmYJcVwHimAlwfH2OeLsYBRT\ngCfsGIsAD2Elv0kYYwzW+/24va4P1hfZGfYq/YA3jDEnjDEx9rF4TJJ2lxtmjIkyxkTfRsy3s69p\nec9+bxZj/Qgx1Rhz1lit3+v453zqB3xojDlgjInD+vLcQETuTV6g/V5Oss/TGKwEs56IFHRYLbXr\n+hFguzHmF3vZGOBsBvbrJPY1DzwLfGFfX/HGmO/t+QGpbJvqeSlW6+pDwHPGmIvGmOvGmLWplJNc\nd7vclfa+jcDqpt7QYZ3UrrvkugJjjTFh9rojEhbY70lLYIB97p0GPrXrT/C3MeZ7Y0w8MBkoJf+0\nXhugpogUsD8r9zq5fz8A7R3e5172vJRMAzqJSAF7uoc9D5z7XPje/gy6fpsxZ3S7tHxvrHsdEz6X\nDxpjVtnXzk/8c111tJdNsT/3t2L9QPRoJsWhlLoNmjQqlb2OO06ISFURWSRW17XLWL8436qr3SmH\nv6OwfplNj4QvtRfs+p+Wf7qpRQBVb1W/WN0ZV4vIWRG5hPVre4rrizVqZ+LAIymtY4z53RjzmDGm\nCFZC+CAw1N7emWNz2uHv6BSmHY/PeTvZSnAUqyueM1I77iVJ+p4meX+TKZlKDAnKAo8nvBf2+9Eo\nWYypxTEJWA7MEqsL6Qhx/r7BH4COIuKB9UV6lTHmTCrrTgO6iNXttwuwyU6owEr+FzjEvsueX8xh\ne8fjk9GYM7pdmuyEIsGtzqeywOcO+3oO60eJUsnLFKvr5Ci7u99l4JC9yPFcdur8MsbcwGqBS697\nsa95O/bByc6zEvzz2ZDcrc7L0sA5Y8ylDMRUEofz32HfHONw9vMu+XWY/LrKD5x2iP9zrBbH1OoB\nq/X6MtaPJM8Dp0RkoYhUSWvH7P05DvwBdBarG3lr/kkEk6+7H6u192ER8cL6sSBhXWc+FxzPkQzF\nfDv76oT0XFdNk+1rN6zzUynlYpo0KpW9kg8D/zWwG6vV5C7gHZy8ZymDgrC+IB2yW76+BJ7D6urk\ng9V1KKH+lIasnwH8DJQ2xnhjdUFMMV5jjdrp9MAjxphNWL8q17RnZfax8bUTowRl+KflNaPCSZoo\npHhfm8O6KcWQ4DjwP2OMj8OroDHm47SCMNb9a8ONMdWAZljvc0oj9t70nhprdN6tQCdu3RqC3aJ7\nCquF0bE1BKwv/K2SxV/AGHPKYXvj8PetYr6K1ZUwQXEnt8sux7G6Nzruq4d9Dif3JFZ31wexWtIq\n2fOdOZfDcTin7FbbmxLTWxERN6wu8SEOsQcni93TGDPLXp78HLnVeXkcKCIid6VQdVqPvDiJlSQk\n37cT6dk/W5LjxM3XVRRQ2CH+u4wxfs4UbIxZYoxpiZW4HML6XLpptVQ2n4zVit8NWOt4LaRgOlbS\nFoTVuhzqEH9anwtJ6r9FzKleV+nY16x0HFiRbF+9jDEvZHMcSqkUaNKolGsVwrqP5KpYg2P8Nysq\nEZF7ROQl4C1gsP3l3Qvry8ZZaxXpi9XSmOA0VjetfMnivWCMiRGRRiTt4pXemJqLNZhEMXu6GtaX\n240OdWXmsckDDBdrkIlAoB0w+zbLnAU8IyL32d1qUx0YyBjzN7DTIYYHgIcdVvkBCBKRVnbrVAER\naSEiabaGisiDIlLT/uJ9GasLZ0qj0Kb0noLVRXUo1vs/L43qpmENqtKYpMfvK+DDhEEtxBr4pEMG\nY94OdBdrwKMGQOcM7GtW+gp40z4vEwZaSa0LXSGse3fPY31h/yAd9SwE/EWko/2eDcC6xzJNIpJP\nRKpj/dBTGKtLJsA3wPMiEiAWLxFx7EZ5Guv+swSpnpd2a9pyrFZXH7vOBxzKKSIihVIJcRbQQazn\noibcr3kFSCnxTsss4BURuVdEfIHBCQvsGNcAo0XkLrEGJqrkEGeqRKSEfWw8sborXyWFc83u1nqe\npMcNrHuuG2LdczkljeqmY30mPUvSH2PS9bmQRszbgeYiUlqs7uVD0ruvWWw+UENEetjnUj4RaSAO\nAyQppVxHk0alXOs14CmsL0tfYw2CkWnE7h6Klay0ATobY6ZAYqvReKwuVOFYg3o4fmH7DWvgldMi\nkvAL+XPAR2KN/voG1pe1jLqI9av6bjvGxXZ5Y+zlmX1swrC+CIVjtQD0MdZ9lRlmjFmA1Vq7FutY\nrbcXXUtlk+5AU6yugm/i0KpntywEYSWeZ7Ee2/Iazn1Ol8T6gnoZa0CQ5aTcFS6l9xSs1uMKWPfW\nRadR1zSsVrPf7HuUEnyCdZ/oCvv8+J3U75NLK+Y3sRLYCKzjMc3J7bKFMeYnrP39SawupwnXV0r+\nh9WqdhIr3t/TUc9prFaqj7G6wJYh7aSqp338L2L9AHAaqJ/QymWM2Yh1HX9pr3MQ+55W24dAsN09\n8BUnzsuEbQ/adb1o17Mb67wKtcty7KaMMWYP1vX9pV1uW6CDw3156fElsAKrS/Rmbv4x6Amsgar2\n2vv8E8la2VLhhpXMhmMlhU2wum+mZBgwzd7XzgDGus95Ltb7NvdWFdndvLdgdT2d5TA/lPR9Ltwq\n5qVYgzvtwvrcn5/Bfc0SdjfnNljvVzhWr4aPsLoXK6VcTBx6Cyml1B1JRFoC3xpjymVxPbWAP4H8\n9j1auYKICHAEa1TO1S4OR6k7hoi8C5Qxxjzt6liUUup2aEujUkrdBrGeNecu1mAXI4B5uSlhtHXF\nah1d4+pAlLpT2F1le2ON/KyUUrmaJo1KKXV7nsfqOngIiCGbu3TdLhFZB4wDnjfa9USpTCEiz2F1\nJZ1njHG6S7JSSuVU2j1VKaWUUkoppVSqtKVRKaWUUkoppVSqNGlUSimllFJKKZWqvNldYZ4yeYzH\neY+0V1RKKaWUUkqpO1BUVJQxxuSaBrxsTxo9+nlw9Y2r2V2tUkoppZRSSuUIIpLWc5ERkbbAZ1jP\nUv3WGDMihXW6AsMBA+wwxvTI5FABFySNSimllFJKKaVSJyJuwOdAKyAM2Cwi840xex3WqQwMBZoa\nYy6KSLGsiifXNIkqpZRSSiml1L9EA+CQMeawMSYWmAF0TLZOX+BzY8xFAGPMmawKRpNGpZRSSiml\nlMpZ7gWOO0yH2fMcVQGqiMh6Edlod2fNEto9VSmllFJKqVzg+vXrhIWFERMT4+pQlJMKFChAqVKl\nyJcvX/JFeUVki8P0RGPMxHQWnxeoDAQCpYC1IlLLGBOR4YBvUZFSSiml1B3tSNgRnnr9Kd7p/w4t\nm7V0dThKZUhYWBiFChWiXLlyiIirw1FpMMZw/vx5wsLCKF++fPLFccaY+rfY/ARQ2mG6lD3PURiw\nyRhzHTgiIgexksjNtxn6TbR7qlJKKaXuaKdOnaLR/Y0ImR5Cq/tb8fjjj3P48GFXh6VUusXExODr\n66sJYy4hIvj6+ma0ZXgzUFlEyouIO9AdmJ9snblYrYyISBGs7qpZ8uGmSaNSSiml7ljh4eG0aNGC\nsyfP4t3bG+6H2b/MpmrVqrz88sucPXvW1SEqlS6aMOYuGX2/jDFxwAvAMmAfMMsYs0dE3hWRDvZq\ny4DzIrIXWAUMMsacz4Swb6JJo1JKKaXuSCdOnCAwMJDQY6GYnoZJr0+i14BexD0fR6vOrZgwYQIV\nK1bk/fff5+pVfYa0Us744IMPqFGjBn5+fvj7+7Np06YsrS8wMJAtW6xb//7zn/8QEZHpt+vd0ocf\nfpit9Tkyxiw2xlQxxlQ0xnxgz3vHGDPf/tsYY141xlQ3xtQyxszIqlg0aVRKKaXUHScsLIzAwEDC\nw8Op8nIVytcuT/sq7fny4S+pXrE6W+pvYeWmlTz00EO8/fbbVKpUia+//pq4uDhXh65UjrVhwwYW\nLlzIn3/+yc6dO1m+fDmlS5dOe8NMsnjxYnx8fDK1zLSueVcmjTmJJo1KKaWUuqMcO3aM5s2bc/r0\nacb9OI6d+XfyQoMXcMvjRkH3gvz02E9Exkbyzu53+Onnn1i/fj0VK1akX79+1KxZk19++QVjjKt3\nQ6kcJzw8nCJFipA/f34AihQpQsmSJQF49913CQgIoGbNmjz77LOJ11BgYCADBgygfv36VKtWjc2b\nN9O5c2cqV67MW2+9BUBoaChVq1alZ8+eVKtWjUcffZSoqKib6i9Xrhznzp0jNDSUatWq0bdvX2rU\nqEHr1q2Jjo4GYPPmzYmtoIMGDaJmzZo3lbN69Wruv/9+OnToQPXq1QHo1KkT9erVo0aNGkycaA1i\nOmTIEKKjo/H396dnz54ATJ06lQYNGuDv789///tf4uPjM/MQ51iSHR+KEizPAs8CuLu517v21rUs\nr1MppZRS/z5Hjx6lRYsWXLhwgWXLlvFl+JfM3jubsFfD8CnwTwvF1J1T6fVLL4Y2G8qHD32IMYYF\nCxYwZMgQ9u3bR+PGjRk1ahTNmjVz4d4oldS+ffuoVq0aAK8sfYXtp7Znavn+xf35tO2nqS6PjIyk\nWbNmREVF0bJlS7p160bz5s0BuHDhAoULFwagV69edO3alfbt2xMYGEjDhg0ZOXIkn332GSNHjmTr\n1q0ULlyYihUrsmPHDq5cuUL58uVZt24dTZs25f/+7/+oXr06AwcOJDAwkNGjR1O/fn3KlSvHli1b\niIyMpFKlSmzZsgV/f3+6du1Khw4deOKJJ6hZsybffPMNjRs3ZsiQISxcuJDdu3cn2Y/Vq1fz8MMP\ns3v37sRRTRPij46OJiAggDVr1uDr64uXlxeRkZGJx//1119nzpw55MuXj/79+9OoUSOefPLJWx5X\nx/ctgYhEGWMKpu8dcp1saWk0w8xEM8zUN8NM/bx59CkfSimllMp8R44coXnz5ly8eJHly5dTrkY5\npu+eztP+TydJGAGe8HuCvnX78tG6j1j812JEhA4dOrBz506+/fZbjh49mtgSsXfvXhftkVI5i5eX\nF1u3bmXixIkULVqUbt26MWnSJABWrVpFw4YNqVWrFitXrmTPnj2J23XoYI3bUqtWLWrUqEGJEiXI\nnz8/FSpU4Phx6/n1pUuXpmnTpgA88cQTrFu37paxlC9fHn9/fwDq1atHaGgoERERXLlyhcaNGwPQ\no0ePVLdv0KBBksdgjBs3jtq1a9OoUSOOHz/OX3/9ddM2K1asYOvWrQQEBODv78+KFSv+NSMxawan\nlFJKqVzv77//pkWLFkRGRrJixQrq1q3Le2veIzY+lhcbvJjiNp+1/Yw/TvxBr196se2/2yjjXYa8\nefPyzDPP8PjjjzNu3DhGjBhBrVq1ePrppwkODqZUqVLZvGdKpexWLYJZyc3NjcDAQAIDA6lVqxaT\nJ0+me/fu9O/fny1btlC6dGmGDx+e5DETCd1Z8+TJk/h3wnTCPYXJRxlNa9RRx3Lc3NwSu6c6q2DB\nfxr5Vq9ezfLly9mwYQOenp4EBgam+JgMYwxPPfUUH330UbrquhPoPY1KKaWUytUOHTpEYGAgUVFR\nrFy5krp16xIbH8sXW76gbaW23FfkvhS388jnwU+P/cT1+Ot0m92N2PjYxGWenp4MGTKEv//+m1de\neYWpU6dSpUqVxFYVpf6NDhw4kKQFbvv27ZQtWzYxwSpSpAiRkZHMnj073WUfO3aMDRs2ADBt2rQM\ndQ338fGhUKFCiSO6zpjh3GCily5d4u6778bT05P9+/ezcePGxGX58uXj+vXrADz00EPMnj2bM2fO\nAFaX1qNHj6Y7ztxIk0allFJKZbmwsDBGjhzJpEmTuHz5cqaVe/DgQZo3b05MTAwrV65M7K72056f\nOBV5ipcbvnzL7Sv7Vua7Dt+xMWwjQ5YPuWm5r68vY8aM4cCBAzRu3JjevXvTu3fvFAfpUOpOFxkZ\nyVNPPUX16tXx8/Nj7969DB8+HB8fH/r27UvNmjVp06YNAQEB6S77vvvu4/PPP6datWpcvHiR5557\nLkMxfvfdd/Tt2xd/f3+uXr2Kt7d3mtu0bduWuLg4qlWrxpAhQ2jUqFHismeffRY/Pz969uxJ9erV\nef/992ndujV+fn60atWK8PDwDMWZ22TLQDiOCn5Y0Fx9Q5+FpJRSSt3pbty4wcqVK/niiy+YP39+\n4iiD+fPnp3379vTs2ZN27dol6WaWHvv37+fBBx8kLi6OlStXJo6SaIyh4bcNuXTtEvue30ceSfs3\n8peWvMT4P8Yzp+scgqoFpbhOfHw87777Lu+99x7Vq1fnp59+umlwC6WyUkoDqtwJQkNDeeSRR24a\nsCYjIiMj8fLyAmDEiBGEh4fz2Wef3Xa5t0MHwlFKKaWUSubixYuMHTuWqlWr0qpVK0JCQhg4cCB/\n//03Gzdu5Nlnn2Xt2rUEBQVRvHhx+vbty6pVq7hx44bTdezdu5fAwEDi4+NZtWpVkmH1N53YxOaT\nm3mpwUtOJYwAH7f6mICSAfSe15vDF1Me2MLNzY3g4GCWLVvGmTNnCAgIYOrUqU7HrJTKeosWLcLf\n35+aNWsSEhKS+FgPdXu0pVEppZRSmWLr1q188cUXTJ8+nejoaJo0aUL//v159NFHb2pNjIuLY8WK\nFUybNo05c+YQGRnJvffeS/fu3enZsyf+/v6pDoSxe/duHnroIfLkycPKlStv+gX/8Z8fZ8lfSwh7\nNQwvdy+n4w+NCKXO13WocHcF1v/fegrkLZDquidPnuTxxx9n7dq19OnTh3HjxuHh4eF0XUplxJ3a\n0nin05ZGpZRSSv2rRUdHM2nSJBo0aED9+vWZOXMmvXr1Ytu2baxfv56ePXum2P00b968tGnThsmT\nJ3P69GlmzJhBvXr1GDduHHXr1k28d+jvv/9Ost2uXbto0aIFbm5urF69+qYvYicun2D23tk8U+eZ\ndCWMAOV8yjG502T+DP+T15a9dst1S5YsyYoVKxg6dCjffvstjRo14uDBg+mqTymlcgtNGpVSSimV\nbocOHWLgwIHce++99O7dm8jISMaPH8+JEyf4+uuvEwekcYanpyfdunVj3rx5hIeH8/XXX1OsWDHe\nfvttKlWqROPGjRk/fjwrV66kRYsW5M+fnzVr1nDffTePivrlli+JvxHP8w2ez9B+dbivA4OaDOKL\nLV8wY/etR17MmzcvH374IYsXL+bEiRPUq1ePmTNnZqhepZTKybR7qlJKKaWcEhcXx6JFi/jiiy/4\n9ddfyZs3L0FBQfTv35/mzZun+Vy19Dp+/DjTp0/nxx9/ZOfOnQCUKlWKVatWUalSpZvWj4mLofTY\n0jQt3ZS53edmuN7r8ddpMbkFO07vYEvfLak+siN5rN27d+f333/nueee45NPPqFAgdS7tyqVEdo9\nNXfS7qlKKaWU+tfo2LEjnTp1Ys+ePbz77rscO3aMWbNmERgYmOkJI0Dp0qV5/fXX2bFjB7t27WLE\niBGEhISkmDACTNs1jXNR59J8zEZa8rnlY8ajMyiQtwCP/fQYUdfTfrxG6dKlWb16NYMGDeLLL7+k\nSZMmN3WtVUqp3EqTRqWUUkqladOmTSxevJg333yT0NBQ3n77bUqUKJFt9desWZPBgwdTrly5FJcb\nY/hs02fUKlaLwHKBt11fqbtK8UPQD+w+s5sXF7/o1Db58uVj1KhRzJ8/n9DQUOrWrcvPP/9827Eo\nlZO4ubnh7+9PjRo1qF27NmPGjElz5OPQ0FCmTZuW4TonTZrEyZMnE6f79OnD3r17M1yeY7lFixbF\n39+fqlWrMnbs2DS3Wb16Nb///vtt153baNKolFJKqTSNGTMGb29vBg8eTN68eV0dzk3WHl3LztM7\neanhS5nW6tm2UlvevP9Nvt/+PZO2T3J6u/bt27Nt2zaqVq3Ko48+yksvvcS1a9cyJSalXM3Dw4Pt\n27ezZ88efvvtN5YsWUJwcPAtt8nspPHbb7+levXqGS7PUbdu3di+fTvr16/ngw8+4Pjx47dcX5NG\npZRSSqkUHDlyhJ9//pl+/fpRqFAhV4eTos82fYavhy89a/XM1HKHBw6nRbkW9F/Un91nnH/weNmy\nZQkJCeGVV15h/Pjx3H///Rw5ciRTY1PK1YoVK8bEiROZMGECxhji4+MZNGgQAQEB+Pn58fXXXwMw\nZMgQQkJC8Pf3Z+zYsamuBzBy5Ehq1apF7dq1GTJkCLNnz2bLli2Jj+KJjo4mMDCQLVu2ADB9+nRq\n1aqV2BshgZeXF2+++Sa1a9emUaNGnD59+pb74uvrS6VKlQgPDwdgwYIFNGzYkDp16tCyZUtOnz5N\naGgoX331FWPHjsXf35+QkBDOnj1Lly5dCAgIICAggPXr12f2Yc4Rct5PhUoppZTKUT799FPy5MnD\niy86100zu4VGhDLvwDwGNx2MR77MfVaiWx43pnWZhv9X/nSe2ZmvH/mawHLO3cPp7u7O2LFjeeCB\nB+jduzeVKlXC39+fZs2acbgdrF8AACAASURBVP/999OsWTOKFy+eqfGqf49XXnmF7du3Z2qZ/v7+\nfPrpp+napkKFCsTHx3PmzBnmzZuHt7c3mzdv5tq1azRt2pTWrVszYsQIRo8ezcKFCwGYOHFiiuvt\n37+fefPmsWnTJjw9Pblw4QKFCxdmwoQJjB49mvr16yep++TJkwwePJitW7dy991307p1a+bOnUun\nTp24evUqjRo14oMPPuD111/nm2++4a233kp1P44dO0ZMTAx+fn4ANGvWjI0bNyIifPvtt4waNYox\nY8bQr18/vLy8GDhwIAA9evRgwIABNGvWjGPHjtGmTRv27duXrmOYGhFpC3wGuAHfGmNGJFv+NPAx\ncMKeNcEY822mVJ6MJo1KKaWUStXFixf57rvv6NGjB/fee6+rw0nR5398jiA8V/+5LCm/uFdxfnrs\nJ7rM6sKDUx6kbom6DGw8kMdqPEbePGl/lQoKCqJOnTp8//33rFu3jm+++YZx48YBUKlSpSRJZOXK\nlbNkUCGlssOvv/7Kzp07mT17NgCXLl3ir7/+wt3d3an1li9fTu/evfH09ASgcOHCt6xv8+bNBAYG\nUrRoUQB69uzJ2rVr6dSpE+7u7jzyyCMA1KtXj99++y3FMmbOnMnatWvZv38/EyZMSBz1OCwsjG7d\nuhEeHk5sbCzly5dPcfvly5cnub/y8uXLREZG4uWVvufEJicibsDnQCsgDNgsIvONMclv5pxpjHnh\ntipzgiaNSimllErV119/zdWrV3n11VddHUqKrsZe5dtt39KlehdKe5fOsnruL3s/R185ytSdUxmz\nYQw95vRgyIohDGg0gGfqPEOh/LfutluuXDneffddAGJjY9m2bRshISGsW7eOBQsWMGnSJMDq7ueY\nRPr7++fIe0iV66W3RTCrHD58GDc3N4oVK4YxhvHjx9OmTZsk66xevTrJdGrrLVu2LNPiypcvX+IP\nMG5ubsTFxaW4Xrdu3ZgwYQJbtmyhdevWdOjQgeLFi/Piiy/y6quv0qFDB1avXs3w4cNT3P7GjRts\n3LgxKx6x0wA4ZIw5DCAiM4COwO2PAJQBek+jUkoppVIUGxvLuHHjaNWqFbVr13Z1OCmasmMKETER\nt/2YDWd45POgb72+7H1+L/O6z6Osd1kGLBtA6bGlGbJ8CCevnEy7EKxuqw0bNmTgwIHMnTuXs2fP\nsnfvXiZOnEibNm3Ytm0bAwYMICAgAB8fH1q1akVwcDCbN2/OcOy7z+zmxOUTaa+oVDqcPXuWfv36\n8cILLyAitGnThi+//JLr168DcPDgQa5evUqhQoW4cuVK4naprdeqVSv+97//ERVlPebmwoULADdt\nn6BBgwasWbOGc+fOER8fz/Tp02nevHmG9qV+/fr06tWLzz77DLBaPxN6V0yePDlxveSxtG7dmvHj\nxydOp6PLcF4R2eLwejbZ8nsBx1F5wux5yXURkZ0iMltEsuyXM00alVJKKZWi6dOnEx4ennjvTk5z\nw9xg3B/jqF+yPo1LNc62evNIHjrc14G1vdey8ZmNtK7Ymo9//5hyn5aj97ze6RowB0BEqFatGn37\n9mXKlCkcPnyYsLAwZsyYQe/evTl79izBwcE0aNCABx54gIULF6b5iAOwWnPWHl1L26ltqfVlLepO\nrMv2U5l7D5z694mOjk585EbLli1p3bo1w4YNA6xHYVSvXp26detSs2ZN/vvf/xIXF4efnx9ubm7U\nrl2bsWPHprpe27Zt6dChA/Xr18ff35/Ro0cD8PTTT9OvX7/EgXASlChRghEjRtCiRQtq165NvXr1\n6NixY4b3bfDgwfzvf//jypUrDB8+nMcee4x69epRpEiRxHXat2/PL7/8kjgQzrhx49iyZQt+fn5U\nr16dr776ytnq4owx9R1eEzMQ8gKgnDHGD/gNmJzG+hkmxpisKjtFBT8saK6+cTVb61RKKaVU+hhj\n8PPzQ0TYsWNHjrzP7te/f6XN1DZM6TSFXrV7uTSWwxcPM3bDWL7f/j1R16NoW6ktAxsP5MHyD2bK\nsYuIiGDy5Ml88sknHDt2jOrVqzNo0CB69Ohx0/1ixhiWHFrChyEfsv74eooVLMZz9Z/j+23fc/na\nZRb1WETTMk1vOyaV/fbt20e1atVcHYZKp5TeNxGJMsYUTG0bEWkMDDfGtLGnhwIYYz5KZX034IIx\nxjvTAnegLY1KKaWUusmvv/7K7t27ee2113JkwgjWYzbuKXgPXWt0dXUoVLi7AuP/M55jrxzj/Rbv\nsy18Gy1/aEm9ifWYtmsa1+Ov31b5Pj4+vPzyyxw6dIipU6eSN29eevfuTYUKFRg9ejSXL18m/kY8\nM3fPpM7XdXh42sMcv3ycCe0mEPpyKMMDh7Pu/9Zxj9c9tPqhFUsPLc2kPVdKZZHNQGURKS8i7kB3\nYL7jCiJSwmGyA5A5w7amQFsalVJKKXWT1q1bs3v3bkJDQ29qycoJDp4/yH0T7mN48+EMCxzm6nBu\nEhMXw487f2T0htHsP7ef0neVpl/9fjxY3hp91d3t9o6pMYZff/2VUaNGsXLlSjwKelCgcQEu1r5I\n1fJVGdJ0CD1q9SCfW74k2525eoY2U9uw58wefuz8I4/VeOy24lDZS1sac6eMtDTa6/wH+BTrkRvf\nG2M+EJF3gS3GmPki8hFWshgHXACeM8bsz4p90OG4lFJKKZXEjh07+O233/joo49yZMIIMOGPCeTL\nk4//1v+vq0NJUYG8BXim7jP0rtObJX8t4ePfP+bNlW8mLmt4b0OalWnG/WXup3HpxtyV/650lS8i\nNGvRjP3e+9k9azdnfj1D9Ipo8q7JS5NeTWjUstFNCSNAsYLFWPXUKtpPb0/3n7tz6dol+tTtkyn7\nrJTKXMaYxcDiZPPecfh7KDA0O2LRlkallFJKJfHUU0/x888/c/z4ce6++25Xh3OTSzGXKDW2FEFV\ng5gSNMXV4TjtVOQp1h9bz7pj61h3fB3bwrcRb+LJI3nwu8ePZqWbcX/Z+2lWphklC5VMtZyImAg+\n/+NzPt30KeeiztG8bHPeuP8NKpgKjB07lu+//55r167RsWNHXn/9dRo3vnmQoKjrUXSZ1YWlh5by\ncauPGdjkn8GOrl27xunTpzl16lTi68yZM+TNmxcfH58kL29v78S/CxQokGO7Mt8p9u3bR9WqVfU4\n5yLGGPbv35+hlsacRJNGpZRSSiUKCwujfPny9O/fP3Ho+Zzm042fMmDZALb03UK9kvVcHU6GXbl2\nhU0nNllJ5LF1bAjbQNR161ED5X3K06xMs8TWyKpFqnLm6hnGbhzLF5u/4ErsFR6p8ghDmw2lSekm\nSco9e/YsEyZMYMKECVy4cIFmzZrx+uuv06BBgyTJYNjJMCavm8zBowcpm7csntc8OXXqFBcvXszQ\n/ri7u6eYTCa8unbtSv369W/7uP2bHTlyhEKFCuHr66uJYy5gjOH8+fNcuXKF8uXLJ1mmSWMaNGlU\nOUlsbCzXr1+nYMFcc80qpVSWGjx4MKNHj+bQoUM3fcnJCeJvxFNlQhVKeJVg3f+tc3U4mep6/HV2\nnN5ByNEQ1h23EskzV88A4Ovhy9XrV4mNj6Vrja4MaTqE2sVv/ezMq1ev8v333zNmzBiOHj2a4jqe\nnp7kvSsvl/NdpkLpCrT1b0uJEiUoXrx4klexYsWIj48nIiKCS5cuERERkeIrtWWXLl3iq6++4skn\nn8z04/Zvcv36dcLCwoiJiXF1KMpJBQoUoFSpUuTLl7S7uCaNadCkUeUUly5dIjAwkLNnz7JmzRoq\nVqzo6pCUUsqlrly5QunSpWndujWzZs1ydTgpWnBgAR1mdGDmozNzxKipWckYw6ELh1h3bB0hx0Lw\nzOfJyw1fprJv5XSVExcXx9y5czl9+jT33HNPkmTQy8sLYwxDlg9h1O+j6FGrB5M6TkrxfsjM2B9t\nHVPKokljGjRpVDlBbGws7dq1Y+3atXh5eeHt7c3atWspU6aMq0NTSimX+fTTTxkwYAAbN26kYcOG\nmVp23I04nl3wLBExEfjd40etYrXwu8ePCndXwC2Pm9PltJzSkgPnD3D4pcNZktj8m41YN4KhK4by\nSJVHmPXoLDzyebg6JKXuWJo0plRJsDwLPAvg7uZe79pb17K8TqVSc+PGDZ544gmmT5/OlClTqFGj\nBg8++CBFixZlzZo1lCyZ+uADSil1p4qLi6NSpUqULl2akJCQTC//g7Uf8NaqtyjnU46jEUcxWN8/\nPPJ6ULNYzcQkstY9tahVrBZFCxa9qYzdZ3ZT68tafPTQRwxpNiTTY1Tw1Zav6L+oPw+UfYD5j89P\n96iuSinnaNKYBm1pVK42aNAgRo8ezUcffcSQIdaXjo0bN9KqVStKlSrFmjVrKFasmIujVEqp7DVz\n5ky6d+/O3Llz6dixY6aWveXkFhp/15gu1bowvct0ouOi2XNmD7vO7GLX6V3sPLOTnad3ci7qXOI2\nxb2K/5NI2v+O/2M803dPJ2xAGL6evpkao/rH9F3TeXLuk9S+pzZLei5JMYFXSt0eTRrToEmjcqWE\nrlfPP/8848ePT3Jvxdq1a2nbti2VK1dm1apVFC5c2IWRKqVU9jHG0LBhQyIiIti/fz958uTJtLKj\nrkdR9+u6RMZGsuu5XdztkfIjPIwxnL562koiT+9k1xnr371n93It/p8eSn3q9OGbDt9kWnwqZYsO\nLuLRnx6lnE85fuv1G6XuKuXqkJS6o2jSmAZNGpWrzJo1i+7duxMUFMSsWbNwc7v5Hprly5fzyCOP\nULNmTVasWIG3t7cLIlVKqey1du1amjdvzpdffkm/fv0ytewXFr/A55s/Z3mv5TxU4aF0bx93I45D\nFw6x8/RODl04RG//3pQoVCJTY1QpW3t0LY9Me4TCHoX5rddv6R6ARymVOk0a06BJo3KF1atX06ZN\nGxo0aMCvv/6Kh0fqN/cvWrSIoKAgAgICWLZsGV5eXtkYqVJKZb8OHTqwYcMGjh49iqenZ6aVu/TQ\nUtr92I4BjQbwSZtPMq1clX22ntxK2x/b4iZurHpqFdWKVkt7I6VUmlyVNIqIpzEmKr3bZV7/E6Vy\nqF27dtGpUycqVqzIvHnzbpkwAjz88MPMmDGDTZs20b59e6Ki0n1dKaVUrnHgwAEWLFhA//79MzVh\nPBd1jt7zelOzWE0+fOjDTCtXZa96JesR0juEq9ev8unGT10djlIqg0SkiYjsBfbb07VF5Atnt9ek\nUd3Rjh8/Trt27ShYsCBLly51+j7Fzp07M2XKFNasWUNQUJA+RFcpdcf65JNPyJ8/P88//3ymlWmM\n4b8L/8uF6AtMDZpKgbwFMq1slf2qFqlKo1KN2Hxys6tDUUpl3FigDXAewBizA3jA2Y3TTBolWESC\n5QkJlnfs6TISLA0yGKxS2ebixYu0a9eOK1eusGTJknQ/g7FHjx589913/Prrr3Tt2pXY2NgsilQp\npVzjzJkzTJkyhSeffDJTR42evGMyc/bN4f0W71O7eO1MK1e5TkDJAHad2UVMnP6IqlRuZYw5nmxW\nvLPbOtPS+AXQGHjcnr4CfO5sBUq5QkxMDJ06deLgwYPMnTsXPz+/DJXTu3dvvvjiCxYsWEDPnj2J\ni4vL5EiVUsp1vvjiC2JiYnj11VczrcwjF4/w0pKXaF62Oa82zrxylWsFlAwg7kYc209td3UoSqmM\nOS4iTQAjIvlEZCCwz9mNnUkaG5ph5nkgBsAMMxcB9wyFqlQ2iI+Pp1evXqxdu5YpU6bQokWL2yrv\nueee45NPPmH27Nk8/fTTxMc7/aOMUkrlWNHR0Xz++ee0b9+eqlWrZkqZ8Tfi6fVLL0SEyZ0m45bn\n5lGqVe5Uv2R9wHrmplIqV+oHPA/cC5wA/IH+zm6c14l1rkuwuAEGQIKlKHAj/XEqlfWMMQwYMIDZ\ns2fzySef0L1790wpd8CAAURHR/Pmm29SoEABJk6cmKnPMVNKqew2ZcoUzp07x2uvvZZpZY5cP5L1\nx9fzQ9APlPUpm2nlKtcrdVcp7il4j97XqFTudZ8xpqfjDBFpCqx3ZmNnvvWOA34BikmwfACsAz5K\nb5RKZYfRo0czfvx4Xn31VQYMGJCpZb/xxhu89dZbfPfdd7z00ktk9+NqlFIqs9y4cYNPPvmE+vXr\n88ADTo+DcEtbT25l2OphdK3RlZ61eqa9gcpVRISAewPYfEKTRqWyi4i0FZEDInJIRIbcYr0uImJE\npP4tihvv5LwUpdnSaIaZHyVYtgIPAQJ0MsOM0/1flcouP/74I6+//jrdu3fn448/zpI63n33XWJi\nYhg9ejQeHh6MGjUKEXFq2+joaI4dO8bRo0cTX2FhYXh5eVGuXDnKli2b+G/RokWdLlcppdJr4cKF\nHDx4kOnTp2fKZ03U9Sie+OUJ7il4D18+/KV+ft2hAkoGsOjgIq5cu0Kh/IVcHY5SdzQRccMaR6YV\nEAZsFpH5xpi9ydYrBLwMbEqlnMZAE6CoiDjeaH4X4PQ9BGkmjRIsP5hhphf2Mz2SzVMqR1i+fDm9\ne/emRYsWTJo0Kcu6jooIo0aNSpI4vvvuuxhjuHjxYmIymDw5PHr0KGfPnk1SlpubGyVKlODKlStc\nunQpyTIPDw/Kli2bJJFM+Lds2bKUKFFCu8cqpTJs9OjRlClThkcffTRTyhv822D2n9vPb71+o7CH\nc482UrlP/ZL1MRj+DP+T5uWauzocpe50DYBDxpjDACIyA+gI7E223nvASGBQKuW4A15YeZ/jrz2X\nAaf/E3DmnsYajhP2/Y31nK1A5SzR0dEcPHiQffv2sX//fvbt28fhw4cpXrw4NWrUSHxVrVo1Ux/y\nnJW2bdtGUFAQVatW5ZdffiF//vxZWp+I8NlnnxETE8N7773HjBkzCA8PJzIyMsl6Hh4elClThrJl\ny1KnTp3EhC9h3r333kvevNYlGBERkZhchoaGJvl769atnDt3LknZ7u7ulC5dmvLly/PQQw/RuXNn\nqlSpkqX7rZS6M/zxxx+EhIQwduzYxM+g27Hs0DImbJ7Ayw1fpmWFlpkQocqpAkoGALD55GZNGpW6\nfXlFxHFkqYnGmIkO0/cCjo/ICAMaOhYgInWB0saYRSKSYtJojFkDrBGRScaYoxkONrUFEixDgTcA\nDwmWy1hdUwFigYmpbadyhvPnz7Nv374kyeH+/fsJDQ1NvBdPRChfvjwVK1YkNDSUZcuWcf369cRl\nFSpUSEwiq1evnphMenh4ZGnsxhiuXbtGdHQ0MTExKf6b8PfVq1cZOnQod999N0uWLMHb2ztLY0uQ\nJ08evvrqK4oVK8aePXto165dYlKY8CpSpIjTXbR8fHzw8fGhdu2Un2cWGRnJsWPHbkooDxw4wNCh\nQxk6dCg1atSgc+fOdOnSBT8/P+0eppRK0ZgxY/D29uaZZ5657bLOR52n97ze1Chag48e0uEO7nRF\nCxalrHdZHQxHqcwRZ4y51T2ItyQieYBPgKed3CRKRD7GahAskDDTGPOgU/WlNZiHBMtHZpgZ6mQw\naSr4YUFz9Y2rmVWcSxljiI2NJTo6GoCCBQuSL1++LK3zxo0bXL58mYiICC5dukRERAQREREcPnw4\nSYLo2DJVoEAB7rvvPqpWrUq1atUS/61cuXKSBPD69escOnSIPXv2JHkdPHgw8fmEefLkSZJM1qhR\ngypVqhAfH09kZCRXr1516uW4blRU1E2JYXr4+vqydu1aqlevnjkHOZc5fvw4v/zyC3PmzCEkJIQb\nN25QoUIFOnfuTOfOnWnYsKF2ZVVKAXDkyBEqVarEwIEDGTly5G2VZYzhsZ8eY/6B+Wzuu5naxVP+\n0UvdWR6d9SjbTm3j75f+dnUoSuVqIhJljCl4i+WNgeHGmDb29FAAY8xH9rQ38DeQ0NWtOHAB6GCM\nuenZOCLyKzATGIj1+I2ngLPGmMFOxevMCJASLHcDlXHMSoeZtc5UkFxOTBqXLl3KkiVLbkpcUmvl\nSlgWExNz0wia7u7ueHl5UbBgQby8vJL8ndK8hL/j4+OTJIKOCaHjvMuXL6e6H76+vkmSwoR/y5Qp\ng5tbxp+VFRsby19//ZWYRO7duzcxmXTmmYVubm4ULFjwli8PDw8KFCiQoX+LFy+Ol5dXhvfvTnLm\nzBnmz5/PnDlzWL58OdevX6dkyZIEBQXRuXNnHnjggUzpjqaUyn327t1L//79Wb9+PUeOHKFUqVK3\nVd7k7ZN5et7TjGw5ktebvp5JUaqcbuS6kQxZMYRzg87h6+nr6nCUyrWcSBrzAgexBiM9AWwGehhj\n9qSy/mpgYEoJo718qzGmnojsNMb42fM2G2MCnInXmYFw+mCNyFMK2A40AjYATjVl5gZ//vknU6ZM\noUCBAikmJj4+PkmmU1oHSNKCFhkZmeTvkydP3jQvpYQrT548+Pj44O3tnfhvxYoVk0wn/9fb25ty\n5cpRpEiRLDk+7u7uia2KjmJjYzl48CB///037u7uqSaE7u7u2lUymxQrVow+ffrQp08fIiIiWLRo\nEXPmzOH777/n888/x9fXlw4dOtClSxdatmyZ5fd/KqVc7/jx4wwfPpxJkybh5eXFhAkTbjthPHLx\nCC8ueZEHyj7Aa40z7zmPKucLuNf6frnl5BbaVGrj4miUunMZY+JE5AVgGdYop98bY/aIyLvAFmPM\n/HQWed3+N1xEHgZOAk6PXOZM99RdQACw0Qwz/hIsVYEPzTDTOZ2BAjmzpdEVEu7bS0gg3dzc8PHx\noWDBgppgqUwXFRXF0qVLmTNnDgsWLODy5csUKlSIRx55hNdee4169XRsK6XuNBcuXGDEiBGMHz+e\nGzdu8Pzzz/PGG2/c9g+M8TfiCZwcyM7TO9nZbydlfcpmUsQqN7gUcwmfkT683+J93nzgTVeHo1Su\nlVZLYxbU9wgQApTGej7jXUCws8mnMzc6xZhhJgZAgiW/GWb2A/dlMF5lExEKFCiAr68vZcuWpVSp\nUnh5eWnCqLKEp6cnnTt3ZurUqZw9e5YlS5bQvXt3lixZQv369enYsSPbtm1zdZhKqUwQHR3NqFGj\nqFixIqNHj6Zr164cPHiQTz75JFN6pHz8+8esO7aOCe0maML4L+RdwJsqvlV0MBylchljzEJjzCVj\nzG5jTAtjTD1ghbPbO5M0hkmw+ABzgd8kWOYBGR6uVSnlWu7u7rRt25aJEydy9OhR3nvvPdauXUvd\nunUJCgpix44drg5RKZUBcXFxfPfdd1SuXJnBgwfTtGlTtm/fzuTJkylbNnOSuz/D/+SdVe/wWPXH\neMLviUwpU+U+ASUDNGlUKhcRkXtFpL6IuNvTxUTkQ+AvZ8tIM2k0w0yQGWYizDAzHHgb+A7olMGY\nlVI5yF133cVbb71FaGgow4cPZ9WqVfj7+9OlSxd27drl6vCUUk4wxjB37lz8/Pzo06cPpUqVYvXq\n1SxcuBA/P79MrWvAsgEULViUrx75SnvG/IsFlAzg5JWTnLxy0tWhKKXSICKvYI1LMx7YKCJ9gH2A\nB+D0/Um3TBolWNwkWPYnTJthZo0ZZuabYSY2Y2ErpXIib29vhg0bRmhoKO+88w7Lly/Hz8+Pxx57\njN27d7s6PKVUKkJCQmjatClBQUHcuHGDOXPmsGHDBpo3z/wHr1+Lu8bGsI30qNmDwh5Oj52g7kD1\nS1qPlttyMsVBGpVSOcuzwH3GmMZYDX8TgNbGmAHGmHBnC7ll0miGmXjggARLmdsKVSmVK/j4+BAc\nHMyRI0d46623WLZsGX5+fnTr1o29e/e6OjyllG337t20b9+eBx54gKNHjzJx4kR2795NUFBQlrUA\nbj+1ndj4WBqWapgl5avco06JOriJG5tPaBdVpXKBGGPMBQBjzDHggDFma3oLceaexruBPRIsKyRY\n5ie80luRUir3KFy4MO+99x5Hjhxh6NChLF68mJo1a/L444+zb98+V4en1L+SMYYdO3bQu3dv/Pz8\nCAkJ4aOPPuKvv/6ib9++Wf4M1j9O/AFAw3s1afy388znSY1iNfS+RqVyh1IiMi7hBZRINu0UZ/6H\neTvjMSqlcjNfX18++OADBgwYwJgxYxg/fjwzZ86kR48evPPOO1SpUsXVISp1R7tx4wabNm1izpw5\nzJkzh8OHD5M/f35ee+01hgwZgq9v9j1cfdOJTZTwKkGpu27vGY/qzlC/RH3mHZiHMUbvb1UqZxuU\nbDrdrYzgxHMaM5s+p1Gp3Ovs2bOMHj2aCRMmEBMTQ/fu3enTpw/NmzcnTx5nOi4opdISFxfHmjVr\nmDNnDr/88gvh4eHky5ePli1b0rlzZzp27EjRokWzPa7K4ytTs1hNfun2S7bXrXKer7Z8xXOLnuPw\nS4cpf3d5V4ejVK6T3c9pvF1Z25fFJsHyLNZNmLi7uWdHlUqpLFC0aFFGjhzJa6+9xscff8zXX3/N\ntGnTKFWqFD169KBXr17UrFnT1WEqlevExMSwfPly5syZw7x587hw4QKenp60a9eOzp078/DDD+Pt\n7e2y+M5HnefQhUM8U+cZl8WgcpaAkgEAbD65WZNGpf4FtKVRKZVhUVFRLFiwgB9++IGlS5cSHx9P\n7dq1eeKJJ+jRowclS5Z0dYhK5VhXrlxhyZIlzJkzh0WLFhEZGYm3tzft27enc+fOtGnTBk9PT1eH\nCcCSv5bwn2n/YeWTK2lRvoWrw1E5QGx8LIU+KsTLDV9mVKtRrg5HqVwnt7U0an8ypVSGeXp60q1b\nNxYuXEh4eDjjx48nf/78DBo0iFKlStGqVSumTJnClStXXB2qUjnC6dOnmTRpEh06dKBo0aJ069aN\nVatW0aNHD5YuXcqZM2f44YcfCAoKyjEJI1j3MwqS+KgFpdzd3Kl9T20dDEepf4k0WxolWHYByVe6\nBGwB3jfDzPn0VKgtjUrd+Q4ePMjUqVOZOnUqR44cwcPDg06dOtGrVy9atWqV5aM8KpVTxMbGsn79\nepYtW8ayZcvYvn07/TwwBwAAIABJREFUAGXKlKFz58507tyZJk2a4Obm5uJIb63dj+0IuxzGrud2\nuToUlYM8v+h5ftj5AxFDIsgj2g6hVHpkd0ujiBQF+gLlcLhF0Rjzf05t70TSOAqIB6bZs7oDnsAp\noJkZZtqnJ2BNGpX69zDGsGHDBn744QdmzpzJxYsXKVasGN27d6dXr17Uq1dPR91Td5xDhw6xdOlS\nli1bxqpVq7h69Sp58+alSZMmtGnThrZt21KnTp1cc+4bYyjycRGCqgbxbYdvXR2OykEmbZ9E73m9\n2dt/L9WKVnN1OErlKi5IGn8HQrBGT41PmG+M+dmZ7Z35ub+lGWbqOkzvkmD50wwzdSVYnkhXtEqp\nfxURoUmTJjRp0oTPPvuMxYsXM3XqVL766ivGjRtHuXLlaNq0KY0bN6Zx48bUqlWLfPnyuTpspdLl\nypUrrFy5MrE18fDhwwBUqFCBJ598kjZt2tCiRQvuuusuF0eaMYcuHOJC9AV9PqO6SUJ35S0nt2jS\nqFTO52mMGZzRjZ1JGt0kWBqYYeYPAAmWACChH01cRitWSv27uLu706lTJzp16sTFixeZPXs2ixcv\nZsWK/2/vzqPrqsv9j7+fnIxN0zadm7TQWlpaylBoUoYKKAotWIsDMitcBFQEB+51XVQ0N/pDEcV7\n1atXK+DIIFccsBb7U1F/2pbTlk60lLZASpt0pFNCmunkPL8/zkk8mU9Ckn2SfF5r7bX3/u599n52\n1rdpnvMd9p959NFHgdgYyaKiouYk8rzzzmPChAm9Gkd9fT0VFRXs3r2brKwspk6dyoQJEwZMq0+y\n6urqOH78eJvl2LFjHD9+HDNj/vz5zJs3j6ysrKDDHVCi0SgbNmxoThJXrVpFJBIhNzeXSy65hLvv\nvpuFCxdyyimnBB1qrwhXhAE4d7KSRmlp9tjZ5GbksnbvWj541geDDkdEOrfMzK5w9+U9+XAy3VOL\ngUeA4YABlcCtwFbgXV7iT3bnhuqeKiKJ3J3XXnuN5557jtWrV7N69Wo2bNhAJBL7TmratGnNCeT5\n55/PWWed1Wlr5PHjx3nttdfYvXt3u+t9+/bR+vdednY2U6dO7XAZP358IEllTU0Nhw8fbnc5cuRI\ncwLYXmJYV1eX1D2ysrIoKipiwYIFLFiwgAsuuICxY8f28ZOlvoaGBvbs2UNZWVmbZceOHRw5cgSA\nuXPnsmjRIhYuXMgFF1xAZubge63UXcvv4pGNj3D8nuOkp2k8srR00Y8uoiHawOoPrw46FJEBJYDu\nqVVALlAPNMSL3d2T6gaT9Cs3rNRGAniJH+9BnM2UNIpIV2pqali/fj2rV69uTib37t0LxBK8ptbI\nMWPGtEkKKysrW1wrMzOTk046qXk5+eSTm7dra2vZtWtXm+Xw4Zbze+Xk5HDyySe3SCRPPvlkcnJy\ncPcul2g02m7Z0aNHO0wKDx8+TE1NTYc/o5ycHEaNGsWoUaMYOXJkiyWZsrq6OlavXs3KlStZuXIl\nzz//PA0Nsf9DTj311OYkcsGCBcycOXPQtcRGo1H279/fblJYVlbGnj17iEajzeeHQiGmTJnCtGnT\nmD59OhdddBGXXXZZr7eEp6L5P5xPTkYOf7v5b0GHIinoX1f8K99b9z0q76kkI6ThBSLJSiZpNLNF\nwLeI9fJ8yN3vb3X8o8DHiY1RfAO43d1f7JN4k2hpzALeT+uZdkr8Sz25oZJGEekud6e8vLy5JfK5\n555j/fr11NfXk5+f35wItrceP348aWndm9WvqqqK1157rTmJLCsra5FUNrUy9Ya0tDRGjx7NmDFj\n2l06Opadnd1rMUAsUV+3bl1zErlq1arm5xw3bhwXXHBBcxI5kLq0VlZWsn37drZt28ZLL73UvC4r\nK2vTGjtp0iSmTZvWZpk6dSpTpkwZkrP+1kZqGfHVEXzqvE/pXXzSrsdfeJzrf3U9Gz6ygbkT5wYd\njsiA0VXSaGYhYAdwKVAOrAWuS0wKzWyEu1fGt5cAd7j7ok6uuQS4KL77V3dflmy8yfwP+Ftir9h4\nHkiuv5OISC8yM6ZMmcKUKVO4+uqrgdiYvfr6evLy8nr9fnl5eZx++umcfvrp7R6vrKxk9+7d1NXV\nYWYtlrS0tDZl7R1LS0trbv3rblLbF3Jycrjwwgu58MILgVhL3Pbt25uTyJUrV/Lb3/4WiHVpPfXU\nUxk5ciQjRoxoXvLy8pLazsvL69VXTLg7+/bta5MYbtu2rbmFGiA9PZ0ZM2Ywa9YsFi9e3CIxbGo5\nlpY27d9EQ7RBk+BIh4oLiwFYW7FWSaNI75oPvOzurwKY2RPAlUBz0tiUMMbl0vY1ic3M7H6gGHg0\nXvRJM1vg7p9NJphkksbJXtJxxioiEoSsrKzAWrtGjBjRYUI5WKSlpTF79mxmz57NrbfeCsDBgwdZ\ntWoVK1euZPv27VRVVbFv3z527NhBZWUllZWVnXapTZSTk8OwYcN6tOTk5LB///4WCWJVVVXztUeM\nGMGsWbO49NJLmTVrFrNmzWL27Nm85S1v0ey83aRJcKQr0/OnMyp7FOv2ruO2ebcFHY7IQJJuZusS\n9pe6+9KE/UJgT8J+OdDml7GZfRy4G8gELunkflcAc909Gv/cT4ANQK8ljaus1M7wEtcbfUVEhrDx\n48c3z4DbkUgkQlVVVXMSWVlZ2Wa/srKS6upqTpw40WapqqriwIEDbcrr6+vb3KuwsJDZs2dz0003\nMXv27OYEcdKkSYNuDGZQwhVhCvIKmDxictChSIoyM4oKili7d23QoYgMNBF3L3qzF3H37wLfNbPr\ngXuBmzo5fRTQNMZmZHfuk0zS+FbgZiu1MmLdUw1wL/Ezu3MjEREZ/NLT08nPzyc/P79Xr9vY2EhN\nTQ0nTpygurqasWPH9knXZGkpXB5W11TpUnFBMV9f9XVqI7Vkp/fueGuRIawCmJKwPzle1pEngP/p\n5PhXgQ1m9hdi+dxFwD3JBpNM0nh5shcTERHpC6FQiOHDhzN8+PCgQxkyXj/xOq8cfYXbzlGXQ+lc\ncUExkWiEjfs3ct7k84IOR2SwWAvMMLNpxJLFa4HrE08wsxnuvjO++y5gJx1w98fN7K/ExjUC/Lu7\n7082mA5nX7BSa3pnR1UHi4iIiAxSayrWABrPKF1rmgxn3d51XZwpIsly9whwJ7AC2AY86e5bzexL\n8VlQAe40s61mtpHYuMY2XVPNbFZ8fQ4widjYyHKgIF6WlM5aGh8DFhObNdWJNWM2PwfwlmRvIiIi\nIgNLuDxMmqVRVPCmh9zIIFeYV8iE3Aka1yjSy9x9ObC8VdkXE7Y/mcRl7gZuBx5s7xZ0PnlOsw6T\nRi/xxfH1tGQuJCIiIoNHuCLMnHFzGJ6pLsHSOTOjuLCYtRVKGkVSjbvfHt+83N1rE4+ZWdKDkDtM\nGq208+ZKL/H1yd5EREREBg53Z03FGt4/+/1BhyIDRHFBMb/f8Xuq6qrIy9IkVSIpaBXQOr9rr6xd\nnXVPba8Js0nSTZkiIiIysOw8spOjtUc1nlGSVlRQhOOs37eei6deHHQ4IhJnZhOJvfMxx8zO5p9D\nDkcAw5K9TmfdU9/+piIUERGRASlcHgZgfuH8gCORgaK4IDYZztq9a5U0iqSWhcDNxF7Z8SD/TBor\ngc8le5EuX7lhpZYN3EHsfY0O/B34vpe07BMrIiIig0O4IkxuRi5zxs0JOhQZIMbljuPkkSdrMhyR\nFOPuPwF+Ymbvd/enenqdDl+5keCnwBzgO8B/x7d/1tMbioiISGoLV4QpKigilBYKOhQZQIoLi/Xa\nDZHUNc/MRjXtmFm+mf2fZD+cTNJ4upf4h73E/xJfbiOWOIqIiMggUxupZdP+TZxbqPGM0j1Fk4p4\n9eirHD5xOOhQRKSty939WNOOux8Frkj2w8kkjeut1M5r2rFSOxfQ10giIiKD0Mb9G2mINmgSHOm2\n4sLYuEa1NoqkpJCZZTXtmFkOkNXJ+S10OaYRmAesslLbHd8/CdhupfYC4F7iZ3YnWhEREUldTZPg\nqKVRumvepHlALGlceMrCgKMRkVYeBf5sZj+K7/8L8JNkP5xM0rioJ1GJiIjIwBOuCFOYV0jhiMKg\nQ5EBZmT2SGaOmanJcERSkLt/zcw2Ae+MF33Z3Vck+/kuk0Yv8des1PKBKYnne4mv726wIiIiktrC\nFWF1TZUeKy4o5i+7/hJ0GCLSvm1AxN3/ZGbDzCzP3auS+WCXYxqt1L4MbAa+TezdHg8C33gz0YqI\niEjqOVR9iFePvqquqdJjxQXF7K3ay96qvUGHIiIJzOw24JfAD+JFhcBvkv18Mt1Trwame4nXdz+8\nGCu124HbATJDmT29jIiIiPShNRVrAI1nlJ5LnAxnyalLAo5GRBJ8HJgPhAHcfaeZjU/2w8nMnroF\nGNXlWZ3wEl/qJV7kJV6UnpZMnioiIiL9LVwRJs3SmFcwL+hQZICaO3EuIQuxtkLjGkVSTJ37PxsB\nzSwd8GQ/nEwG91Vgg5XaFqCuqdBLXF8fiYiIDCLhijCnjz+d4ZnDgw5FBqhhGcOYM36OJsMRST1/\nM7PPATlmdilwB/C7ZD+cTNL4E+BrwAtAtEchioiISEqLepQ1FWv4wGkfCDoUGeCKC4r5zUu/wd0x\ns6DDEZGYe4APE8vpPgIsd/cfJvvhZJLGE17i3+5hcCIiIjIA7Dy8k2O1x5hfOD/oUGSAKyoo4uEN\nD7Pr2C6m5U8LOhyRAcvMFgHfAkLAQ+5+f6vjdwO3AhHgEHCLu7/WweXucvdvAT9M+Pwn42VdSiZp\n/LuV2leBp2nZPVWv3BARERkkwhVhQJPgyJtXXBCbDGft3rVKGkV6yMxCwHeBS4FyYK2ZPe3uLyac\ntgEocvcTZvYx4AHgmg4ueROxBDTRze2UtSuZpPHs+Pq8hDIHLknmBiIiIpL61lSsYXjmcE4bd1rQ\nocgAd8aEM8gMZbJu7zqunnN10OGIDFTzgZfd/VUAM3sCuBJoThrdPfGlqM8BN7a+iJldB1wPTDOz\npxMO5QFHkg2my6TRS/ztyV5MREREBqZwRZiigiJCaaGgQ5EBLjOUydyJczUZjkjn0s1sXcL+Undf\nmrBfCOxJ2C8HOusK8mHgmXbKVwH7gLHAgwnlVcDmpINN5iQrtXcBc4DspjIv8S8lexMRERFJXbWR\nWjbt38Td598ddCgySBRNKuJnm39G1KOkWTJveBMZciLuXtQbFzKzG4Ei4OLWx+JjHF8Dzjezk4EZ\n7v4nM8sBcoglj13q8l+xldr3ifWNvQsw4APAyck+hIiIiKS2Dfs20BBt0HhG6TXFhcVU1Vex/fXt\nQYciMlBVAFMS9ifHy1ows3cCnweWuHtd6+MJ590G/BL4QcL1fpNsMMl89XOBl/iHgKNe4qXA+cDM\nZG8gIiIiqa15EpzJShqldzRNhrNu77ouzhSRDqwFZpjZNDPLBK4lNjFpMzM7m1gSuMTdD3ZxvY8D\nC4BKAHffCYxPNphkksaa+PqElVoB0ABMSvYGIiIiktrCFWEmj5hMQV5B0KHIIDFr7CxyM3I1rlGk\nh9w9AtwJrAC2AU+6+1Yz+5KZLYmf9nVgOPC/Zrax1UQ3rdW5e33TjpmlE5vcNCnJjGlcZqU2Kh7U\n+vjFk34RpIiIiKS2cHlYXVOlV4XSQpwz6RwljSJvgrsvB5a3KvtiwvY7u3G5v5nZ54AcM7sUuAP4\nXbIf7rKl0Uv8y17ix7zEnyI2lnGWl/wzWBERERm4DlUfouxYmZJG6XXFBcVs3L+RhsaGoEMREbgH\nOAS8AHyEWDJ6b7If7jBptFIrtlKbmLD/IeBJ4MtWaqN7HK6IiIikDI1nlL5SVFBEbaSWrYe2Bh2K\nyJDn7lFiE9/c4e5XufsP3T3p7qmdtTT+AKgHsFK7CLgf+ClwHFjayedERERkgAiXhwlZiHmT5gUd\nigwyxYWxyXDWVqiLqkhQLOY/zOx1YDuw3cwOmVm3eo52ljSGvMSPxLevAZZ6iT/lJf4F4JSehS0i\nIiKpJFwR5vTxp5ObmRt0KDLITM+fTn52vmZQFQnWp4nNmlrs7qPdfTRwLrDAzD6d7EU6TRqt1Jom\nynkH8GzCsWQm0BEREZEUFvUoayrWML9wftChyCBkZhQVFGkyHJFgfRC4zt3Lmgrc/VXgRuBDyV6k\ns6TxceBvVmq/Jfbajb8DWKmdQqyLqoiIiAxgOw7v4HjdcU2CI32mqKCIFw6+QG2kNuhQRIaqDHd/\nvXWhux8CMpK9SIdJo5f4fcC/Aj8G3uolzQMl04C7uhWqiIiIpJw1FWsATYIjfae4oJhINMLG/RuD\nDkVkqKrv4bEWOu1m6iX+XDtlO5K9uIiIiKSucHmY4ZnDmT12dtChyCDVNBnOur3rOG/yeQFHIzIk\nnWVmle2UG5Cd7EU0NlFERGSICleEKS4oJpQWCjoUGaQK8wqZkDtB4xpFAuLuvfILvrMxjSIiIjJI\n1TTUsOnAJo1nlD5lZhQXFuu1GyIDnJJGERGRIWjD/g1EohGNZ5Q+V1xQzEuvv0RVXVXQoYhIDylp\nFBERGYLC5WEAtTRKnysuKMZx1u9bH3QoItJDShpFRESGoHBFmCkjpjApb1LQocggd+7kc8kMZfLY\nC48FHYqI9JCSRhERkSEoXBFW11TpF6NzRnPr2bfyo40/Yvfx3UGHIyI9oKRRRERkiDlYfZBdx3ap\na6r0m39/678DcP8/7g84EhHpCSWNIiIiQ4zGM0p/O2nkSdxy9i08vOFhyivLgw5HRLqpX5JGK7Xb\nrdTWWamti0Qj/XFLERER6UC4IkzIQpwz6ZygQ5Eh5J633kPUozyw8oGgQxEZEMxskZltN7OXzeye\ndo5fZGbrzSxiZlf1ZSz9kjR6iS/1Ei/yEi9KT0vvj1uKiIhIB8IVYU4ffzq5mblBhyJDyNRRU7np\nrJtY+vxS9lXtCzockZRmZiHgu8DlwGnAdWZ2WqvTdgM3A30+y5S6p4qIiAwhUY+ypmKNuqZKID77\n1s8SiUb4+qqvBx2KSKqbD7zs7q+6ez3wBHBl4gnuvsvdNwPRvg5GSaOIiMgQsuPwDirrKjVzqgRi\n+ujp3HjmjXx/3fc58MaBoMMRCVK6ma1LWG5vdbwQ2JOwXx4vC4SSRhERkSFEk+BI0D5/4eepa6zj\nG6u+EXQoIkGKuHtRwrI06IA6o6RRRERkCAlXhMnLzGPW2FlBhyJD1IwxM7j+jOv53rrvcaj6UNDh\niKSqCmBKwv7keFkglDSKiIgMIeGKMMWFxYTSQkGHIkPY5y/8PDUNNXxz9TeDDkUkVa0FZpjZNDPL\nBK4Fng4qGCWNIiIiQ0RNQw2bD2xW11QJ3Kyxs7jm9Gv477X/zeETh4MORyTluHsEuBNYAWwDnnT3\nrWb2JTNbAmBmxWZWDnwA+IGZbe2reJQ0ioiIDBHr960nEo0oaZSUcO+F91JdX81/PvefQYcikpLc\nfbm7z3T36e5+X7zsi+7+dHx7rbtPdvdcdx/j7nP6KhYljSIiIkPE6vLVAJo5VVLCnPFzuOq0q/h2\n+NscrTkadDgi0gkljSIiIoNU1KM8v/d5Sv9aStHSIj7zx88wa+wsJg6fGHRoIgDce9G9VNVX8a3w\nt4IORUQ6Ye7erzfM/UquV3+uul/vKSIiMlRU11fz57I/s2zHMpbtWMa+N/ZhGOdPOZ/FMxZz45k3\nMmXklK4vJNJP3veL9/Fs2bO89qnXGJk9MuhwRPqFmZ1w99yg40hWetABiATF3Xlg5QMcrjnMfZfc\nR0YoI+iQRER6ZM/xPbEkcecyni17ltpILXmZeSw6ZRGLZy7m8lMuZ1zuuKDDFGnXFy76Ar9+6dd8\nZ813uPeie4MOR0TaoaRRhqT6xnpu+91t/HTTTwHYemgrT171JLmZA+YLHxEZwqIeZW3FWn6343cs\n27GMTQc2ATA9fzofnfdRFs9czIUnX0hmKDPgSEW6dvaks1ly6hK+ufqbfOLcTzAia0TQIYlIK+qe\nKkPO8drjXPW/V/GnV/9E6dtKmZA7gTuW30FRQRG/v/73jB02NugQRURaqK6vZuuhrWw+sJmVe1ay\nfOdyDlYfJGQhFpy0gHfPfDeLZy7m1DGnYmZBhyvSbev2rqP4h8V85ZKv8NkLPxt0OCJ9Tt1TRVJY\nRWUFVzx2BS8eepEfXfkjbp57MwAThk/guqeuY8EjC1hx4wqmjpoaaJwiMjRFPUrZ0TI2H9gcWw7G\n1q8ceQUn9iXvqOxRXH7K5SyeuZhFpyxidM7ogKMWefOKCoq4YsYVPLj6Qe469y6GZw4POiQRSaCW\nRhkyXjjwAlc8dgXHa4/zy6t/yWXTL2tx/B+7/8G7H3832enZ/OGGP3DWxLMCilREhoIjNUd44cAL\nvHDwheYkccvBLVQ3xP6PNIwZY2Zw5oQzOWP8GZw54UzOnHAmU0dNJc00+bkMPs+VP8f5D5/PA+98\ngM8s+EzQ4Yj0qYHW0qikUYaEZ8ue5b2/eC+5Gbksv2E5cyfObfe8rQe3sujRRVTWVfKba37D26e9\nvZ8jFZHBJupRXjnyChv3b2TD/g1sOrCJzQc2U15Z3nzOmJwxzUlhU5I4Z/wchmUMCzBykf638OcL\n2bBvA7s+tUv1XwY1JY1dUNIo/e3nm3/OLb+9hZljZrL8huWcNPKkTs8vryxn0c8XsfPITn723p9x\n9Zyr+ylSERnoaiO1bDm4hY37N/4zSdy/qbn1MD0tndPGnRZLDsf/M0mcOHyixiKKAKv2rGLBIwt4\n8LIHufv8u4MOR6TPKGnsgpJG6S/uzv3/uJ/PPfs53jb1bfz6ml8zKntUUp89WnOUK5+4kn/s/gf/\ntei/+MS5n+jjaEWkL7k7NZEa3qh/gxMNJwhZiKz0LLJCWc3r7iZth08cbk4ONx6Irbcd2kajNwKQ\nl5nH3IlzmTtxLmdPPJu5E+dy2rjTyErP6otHFBk03vnTd7Ll4BbKPllGTkZO0OGI9ImBljRqIhwZ\nlCLRCHcuv5MfPP8Drj/jeh5Z8ki3/lDLz8lnxY0ruOFXN/DJP3ySvVV7+eo7vqqWAJF+FIlGOFZ7\njCM1RzhaczS2rj3K0ZqjVNZVUt1QTXV9NdUN1bxR/0aL/er6lmUnGk40TyTTkcxQZosksmmdnZ7d\nogxg26Ft7Knc0/zZwrxC5k6cy3tOfU9zojgtf5rGHor0wBcv/iIX//hifrj+h/rSViRFqKVRBp3q\n+mqufepalu1Yxj0L7uG+d9zX4z/cGqON3PXMXfzPuv/hQ2d9iIfe/RAZoYxejlhkcGmMNlLdEEvU\nmpK4pu0TDSeobqimsq6ybTJYe7RFWVV9Vaf3MYzhmcPJzcwlNyOX3Mzc2H58Ozej/f1hGcNo9Ebq\nInXUNdY1r2sjte2XJezXReqIRCPMHDOzufVw7sS5jMsd108/XZGh4W0/fhs7j+zklU+8QnZ6dtDh\niPQ6tTSKBOjAGwdY/Phi1u9bz/eu+B4fK/7Ym7peKC3Ed6/4LoV5hdz7l3s58MYBfnn1LzUVuPSL\nxmhjh0lNbaSWmoYaaiI1nGg4QU1DfN3RfkJ5TaSGukgdZoZhpFlam+00S8OwFtuJ5zV6Y4dJYV1j\nXdLPmBXKIj8nn9E5o8nPzmfyiMmcOeFM8rPzyc+Ol+e03R6RNYLs9Gy1/osMUiUXl3DJTy/hkQ2P\ncEfxHUGHIzLkqaVRBo3tr2/n8kcvZ/8b+3niqidYcuqSXr3+w+sf5iPLPsLZk87m99f/nvG543v1\n+pL6ItFIc7JWG6nlRMOJFktT8tSirD6hLNJyP7EVq73tSDTypuLNSc8hJyOHYRnDyEmPr+P7maFM\nIDazp7vH1niL7aZjidtNx9IsrUUL3rCMYc37XW0PyxjGiKwRjM4ZrfFKItIud+eiH1/ErmO7ePmu\nlzUWWAadgdbSqKRRBoWVu1ey5IklhCzEsuuXMb9wfp/cZ9mOZVz9v1dTOKKQFTeu4C35b+mT+6Sa\nqEepaahpM2asJlJD1KPNS2O0scV+i2Pe8bHEhCQxYenqWGJ5Mp9tfby+sZ6aSCwBTEwGm8pa7/ck\niTOsOWFKTJpyMnLISc9pMVYuO9Ry7Fx2enanY+uartlecqhWOBEZ6P74yh+57OeX8f13fZ+PFH0k\n6HBEepWSxi4oaZTe9tSLT3HDr27gpJEn8cwNzzB99PQ+vd9z5c/xrsfeRXpaOs/c8AznTDqnw3Oj\nHqWqropjtcc4VnuMo7VHm7ebFncnI5RBelo6GWkZZIQy2qw7Opaelk4kGqEh2kB9Yz31jfU0NCZs\nJ5S3d6wuUvfProWtEsLE9YmGE336M+0NaZbWohtl66Wpm2XieZmhTHIycshOzyYnPbbOTs9uLkss\nb+9460SweT8hSezJrJwiIhJrbbzgkQvYW7WXnXftbO4hITIYJJM0mtki4FtACHjI3e9vdTwL+Ckw\nDzgMXOPuu/okXiWNUHa0jF3HdnXaCpLs0tS9q711Ylev1mug3T902/zh28EfxECb6yXeo6vjHbUI\nJbYcdVXeVNbeec3ndHJuZ61VHbVSNUYb2bh/I+dNPo+nr3uascPG9kudeen1l1j484UcqTnCLXNv\n4Y36NzhWd4yjNS2TwuN1x4l6tF9i6o6MtAwyQ5lkhjKbk5zErobN6/bKEtY56TmE0kJt6mPI2pal\nWVqbcw0jlBZqUa/bS+7aO9Y03i5kISVlIiKD1DM7n+GKx67goXc/xIfP+XDQ4Yj0mq6SRjMLATuA\nS4FyYC1wnbu/mHDOHcCZ7v5RM7sWeK+7X9Mn8SpphC/+5Yt8+f99OegwBqTWSUJTUpCYNLQu6+yc\nZBKP1uefkn8KX3nHV/p9bNTeqr287xfvY8vBLYzKHsWo7FHk5+Q3b4/KaqcsYWmazCPN0ppbCxsa\nG1qsI9FIm7LWx9LT0psTwMxQJhmhjJb7CQliUwulkiwRERkI3J1zHzqX10+8TunbSoMORwag7PRs\nPjDnA0GH0UZilEpOAAAG5UlEQVQSSeP5wH+4+8L4/mcB3P2rCeesiJ+z2szSgf3AOO+DBK9fkkYr\ntduB2wEyQ5nz6u5Nfma9/lB2tIzdx3cn1dLXYQtgq9kGW687O2bE/oBvPRYrmbFfTa1wTddIvC7Q\n5h4dHe8siessuRMRERHpS02tjSI9MSF3Avv/bX/QYbRhZvXACwlFS919acLxq4BF7n5rfP+DwLnu\nfmfCOVvi55TH91+Jn/N6b8fbL6/c8BJfCiyFWEtjf9yzO6blT2Na/rSgwxARERGRVi6fcTnlny6n\nNlIbdCgyAIXSQkGH0JGIuxcFHUSy9J5GEREREUlphSMKgw5BpL9VAFMS9ifHy9o7pzzePXUksQlx\nep36F4qIiIiIiKSWtcAMM5tmZpnAtcDTrc55Grgpvn0V8GxfjGcEtTSKiIiIiIikFHePmNmdwApi\nr9x4xN23mtmXgHXu/jTwMPAzM3sZOEIssewTmj1VRERERESkHyXznsZUou6pIiIiIiIi0iEljSIi\nIiIiItKhfu+eapMtSgU1/XrTmHQgEsB9JbWpXkgyVE8kGaon0l2qM5IM1ZPBKcfdB0wDXr8njUEx\ns3UD6V0o0j9ULyQZqieSDNUT6S7VGUmG6omkggGT3YqIiIiIiEj/U9IoIiIiIiIiHRpKSePSoAOQ\nlKR6IclQPZFkqJ5Id6nOSDJUTyRwQ2ZMo4iIiIiIiHTfUGppFBERERERkW5K2aTRzKaY2V/M7EUz\n22pmn4yXjzazP5rZzvg6P14+y8xWm1mdmf1bq2t9On6NLWb2uJlld3DPm+LX3WlmNyWU32dme8zs\njb58ZulaitWLP5jZpvg1vm9mob58dkleitWTv5rZdjPbGF/G9+WzS/JSpZ6YWV5C/dhoZq+b2X/1\n9fNL96VKnYmXX2Nmm+PX+FpfPrd0T0D15A9mdszMlrUqv9PMXjYzN7OxffXMMgS4e0ouwCTgnPh2\nHrADOA14ALgnXn4P8LX49nigGLgP+LeE6xQCZcTehQLwJHBzO/cbDbwaX+fHt/Pjx86Lx/NG0D+X\nob6kWL0YEV8b8BRwbdA/Hy0pWU/+ChQF/TPRktr1pNV5zwMXBf3z0ZK6dQYYA+wGxsXP+wnwjqB/\nPlqCqSfxY+8A3g0sa1V+NjAV2AWMDfpno2XgLinb0uju+9x9fXy7CthG7B/PlcR+ORJfvyd+zkF3\nXws0tHO5dCDHzNKBYcDeds5ZCPzR3Y+4+1Hgj8Ci+LWfc/d9vfZw0mMpVi8qE66TCWiAcIpIpXoi\nqSsV64mZzST2B+Tf3+TjSR9IoTrzFmCnux+Kn/cn4P298IjSCwKoJ7j7n4Gqdso3uPuuN/VAIqRw\n99REZjaV2DclYWBCQgK3H5jQ2WfdvQL4BrFv5PYBx939/7ZzaiGwJ2G/PF4mKSoV6oWZrQAOEvtF\n/cuePIf0rVSoJ8CP4t0Ov2Bm1pPnkL6VIvUE4FrgF+6uL6FSXMB15mXgVDObGk8m3gNM6fHDSJ/p\np3oi0udSPmk0s+HEuv59KqFlB4D4f6qd/sca7y9+JTANKAByzezGPgpX+kmq1At3X0isG0oWcEl3\nPy99K0XqyQ3ufgZwYXz5YDc/L30sRepJk2uBx3v4WeknQdeZeKvjx4BfEGuV3gU0duMRpB8EXU9E\nelNKJ41mlkHsH9uj7v6rePEBM5sUPz6JWCtPZ94JlLn7IXdvAH4FXGBm5yZMOrAEqKDlt3ST42WS\nYlKtXrh7LfBbYr/YJUWkSj2Jf1Pc1EXpMWB+7zyh9IZUqSfxe50FpLv7873ycNInUqXOuPvv3P1c\ndz8f2E5s3JykiH6uJyJ9LmWTxngXroeBbe7+zYRDTwNNs4fdROyP9c7sBs4zs2Hxa74jfs2wu8+N\nL08DK4DLzCw//s3OZfEySSGpUi/MbHjCL/504F3AS731nPLmpFA9SW+arS7+B8RiYEtvPae8OalS\nTxKucx1qZUxpqVRnLD4Tc7z8DuCh3nlKebMCqCcifc9TYDae9hbgrcSa7TcDG+PLFcRmDPszsJPY\nwO/R8fMnEuvrXwkci283zW5ZSuwP+i3Az4CsDu55C7FxAi8D/5JQ/kD8etH4+j+C/vkM1SVV6gWx\ncQhr43FsAb5DrIUg8J+RlpSqJ7nEZsLcDGwFvgWEgv75aEmtepJw7FVgVtA/Fy0Do84Q+4Lhxfii\n2btTaAmonvwdOATUxD+/MF7+ifh+hNgkOg8F/fPRMjAXc9dYexEREREREWlfynZPFRERERERkeAp\naRQREREREZEOKWkUERERERGRDilpFBERERERkQ4paRQREREREZEOKWkUERERERGRDilpFBERERER\nkQ4paRQREREREZEO/X8myhuyBDk7aAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 1080x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA2gAAADSCAYAAAA/r+OzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 1080x216 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bgeG2rL39PUg",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "Points to note :\n",
        "*   In the plots above, green lines representing data/sampling density use the left y-axis.\n",
        "*   The black line is the detection rate for training data uses the right y-axis. \n",
        "* It can be seen that a majority of train data are observations between August and September 2018 and a majority of test data are observations between October and November 2018. \n",
        "* It is evident that there exists a relation between detection rate and sampling rate as the detection rate roughly increases and decreases with sampling rate for the training data.  \n",
        "<br>\n",
        "The last point indicates that it would be useful to engineer a feature that can capture the sampling rate at the time the observation was sampled, as a machine is more likely to have malware if it is sampled from a period with a high sampling rate.\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Zr1GyuLKJXS4",
        "colab_type": "text"
      },
      "source": [
        "# Data Cleaning"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WYPthVPUJlOo",
        "colab_type": "text"
      },
      "source": [
        "Dropping Machine Identifier in test and train data as it is not required for training or prediction.\n",
        "Dropping HasDetections in training as it is not required for training."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0__wK9pQ2nDf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train.drop(['HasDetections'], inplace=True, axis=1)\n",
        "test.drop(['MachineIdentifier'], inplace=True, axis=1)\n",
        "if not(pickled):\n",
        "  train.drop(['MachineIdentifier'], inplace=True, axis=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e57jXifbJoBM",
        "colab_type": "text"
      },
      "source": [
        "Dropping columns with more than 70% missing values"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s9AtP8x06d1Q",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  train.drop(['PuaMode','Census_ProcessorClass','DefaultBrowsersIdentifier','Census_IsFlightingInternal','Census_InternalBatteryType'], axis = 1,inplace=True) \n",
        "  test.drop(['PuaMode','Census_ProcessorClass','DefaultBrowsersIdentifier','Census_IsFlightingInternal','Census_InternalBatteryType'], axis = 1,inplace=True) "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yhe3CCYYzNrD",
        "colab_type": "text"
      },
      "source": [
        "Dropping Date attribute in train and test as it is hard to encode. A new feature indicative of time will be added in feature engineering.\n",
        "Dropping 'dummy' as it was only required for visualization."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Tpdel9fHxk4l",
        "colab": {}
      },
      "source": [
        "train.drop(['Date'], axis = 1,inplace=True) \n",
        "test.drop(['Date'], axis = 1,inplace=True) \n",
        "train.drop(['dummy'], axis = 1,inplace=True) \n",
        "test.drop(['dummy'], axis = 1,inplace=True) "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fvStgX-P7W4L",
        "colab_type": "text"
      },
      "source": [
        "Dealing with SmartScreen : SmartScreen had some inconsistent values. There were values like 'promt', 'promprt', etc for the value 'prompt','of' in place of 'off', 'enabled' for 'on','00000' for '0' and so on. Those inconsistent values too had to be replaced by the proper values.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7Lu8gFgK8ApV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  for category in train['SmartScreen'].unique():\n",
        "      print(category)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kDw15ql17esr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  train['SmartScreen'] = train['SmartScreen'].str.lower()\n",
        "  train['SmartScreen'].replace(\n",
        "      {'promt': 'prompt', \n",
        "      'promprt': 'prompt',\n",
        "      'enabled': 'on',\n",
        "      'of': 'off',\n",
        "      'deny': '0',\n",
        "      '00000000': '0',\n",
        "      'requiredadmin': 'requireadmin'},\n",
        "      inplace=True)\n",
        "  train['SmartScreen'].replace({'0': np.nan}, inplace=True)\n",
        "  train['SmartScreen'] = train['SmartScreen'].astype('category')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iwRrgcJSwUXM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  test['SmartScreen'] = test['SmartScreen'].str.lower()\n",
        "  test['SmartScreen'].replace(\n",
        "      {'promt': 'prompt', \n",
        "      'promprt': 'prompt',\n",
        "      'enabled': 'on',\n",
        "      'of': 'off',\n",
        "      'deny': '0',\n",
        "      '00000000': '0',\n",
        "      'requiredadmin': 'requireadmin'},\n",
        "      inplace=True)\n",
        "  test['SmartScreen'].replace({'0': np.nan}, inplace=True)\n",
        "  test['SmartScreen'] = test['SmartScreen'].astype('category')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d6-e3BLtJ5Tt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  for category in train['SmartScreen'].unique():\n",
        "    print(category)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PPeoKYafBkib",
        "colab_type": "text"
      },
      "source": [
        "##Imputing Median values. \n",
        "Several attributes have missing values, these are imputed with medians. Medians are preferred as they are more resistant to outliers."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rvLkoebB8nk0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "median_imputer = SimpleImputer(strategy='median')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0A7gJsQC8EM6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  train['AVProductsEnabled'] = median_imputer.fit_transform(train['AVProductsEnabled'].values.reshape(-1,1))\n",
        "  train['AVProductsInstalled'] = median_imputer.fit_transform(train['AVProductsInstalled'].values.reshape(-1,1))\n",
        "  train['Census_InternalBatteryNumberOfCharges'] = median_imputer.fit_transform(train['Census_InternalBatteryNumberOfCharges'].values.reshape(-1,1))\n",
        "  train['Census_InternalPrimaryDiagonalDisplaySizeInInches'] = median_imputer.fit_transform(train['Census_InternalPrimaryDiagonalDisplaySizeInInches'].values.reshape(-1,1))\n",
        "  train['Census_InternalPrimaryDisplayResolutionHorizontal'] = median_imputer.fit_transform(train['Census_InternalPrimaryDisplayResolutionHorizontal'].values.reshape(-1,1))\n",
        "  train['Census_InternalPrimaryDisplayResolutionVertical'] = median_imputer.fit_transform(train['Census_InternalPrimaryDisplayResolutionVertical'].values.reshape(-1,1))\n",
        "  train['Census_PrimaryDiskTotalCapacity'] = median_imputer.fit_transform(train['Census_PrimaryDiskTotalCapacity'].values.reshape(-1,1))\n",
        "  train['Census_ProcessorCoreCount'] = median_imputer.fit_transform(train['Census_ProcessorCoreCount'].values.reshape(-1,1))\n",
        "  train['Census_SystemVolumeTotalCapacity'] = median_imputer.fit_transform(train['Census_SystemVolumeTotalCapacity'].values.reshape(-1,1))\n",
        "  train['Census_TotalPhysicalRAM'] = median_imputer.fit_transform(train['Census_TotalPhysicalRAM'].values.reshape(-1,1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wz6KtT9H8kIX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  test['AVProductsEnabled'] = median_imputer.fit_transform(test['AVProductsEnabled'].values.reshape(-1,1))\n",
        "  test['AVProductsInstalled'] = median_imputer.fit_transform(test['AVProductsInstalled'].values.reshape(-1,1))\n",
        "  test['Census_InternalBatteryNumberOfCharges'] = median_imputer.fit_transform(test['Census_InternalBatteryNumberOfCharges'].values.reshape(-1,1))\n",
        "  test['Census_InternalPrimaryDiagonalDisplaySizeInInches'] = median_imputer.fit_transform(test['Census_InternalPrimaryDiagonalDisplaySizeInInches'].values.reshape(-1,1))\n",
        "  test['Census_InternalPrimaryDisplayResolutionHorizontal'] = median_imputer.fit_transform(test['Census_InternalPrimaryDisplayResolutionHorizontal'].values.reshape(-1,1))\n",
        "  test['Census_InternalPrimaryDisplayResolutionVertical'] = median_imputer.fit_transform(test['Census_InternalPrimaryDisplayResolutionVertical'].values.reshape(-1,1))\n",
        "  test['Census_PrimaryDiskTotalCapacity'] = median_imputer.fit_transform(test['Census_PrimaryDiskTotalCapacity'].values.reshape(-1,1))\n",
        "  test['Census_ProcessorCoreCount'] = median_imputer.fit_transform(test['Census_ProcessorCoreCount'].values.reshape(-1,1))\n",
        "  test['Census_SystemVolumeTotalCapacity'] = median_imputer.fit_transform(test['Census_SystemVolumeTotalCapacity'].values.reshape(-1,1))\n",
        "  test['Census_TotalPhysicalRAM'] = median_imputer.fit_transform(test['Census_TotalPhysicalRAM'].values.reshape(-1,1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2swGBGSdCNvf",
        "colab_type": "text"
      },
      "source": [
        "# Feature Engineering"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tFesdPfk8HRu",
        "colab_type": "text"
      },
      "source": [
        "To capture the sampling rate, we engineer a new feature 'WeekNumber' which is the number of weeks that had passed since 1st Jan 2018 when the observation was sampled. If an observation was sampled before 1st Jan 2018, this attribute will have a negative value.  \n",
        "    \n",
        "The newly engineered feature is now representative the sampling rate at the time at which the observation was sampled. This new feature is important for predicting whether the observation is of a system with a malware detection. An observation with 'WeekNumber' value equal to the most frequent value of 'WeekNumber' will have been sampled from the time period when sampling rate was very high. When the sampling rate is high detection rate is high as seen in Fig. 5. Thus an observation with a common 'WeekNumber' value is more likely to have malware detection.\n",
        "\n",
        "We later apply sorted frequency encoding on the attribute as the sorted frequency of 'WeekNumber' is essentially sampling rate, and sampling rate directly correlates with the target which we are trying to predict (Malware detection rate).\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "42zxd0czCRu1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  beginning = datetime(2018,1,1) \n",
        "  week_dict = {}\n",
        "  for i in Av_datedict: \n",
        "      week_dict[i] = (Av_datedict[i]-beginning).days//7\n",
        "  train['WeekNumber'] = train['AvSigVersion'].map(week_dict)\n",
        "  test['WeekNumber'] = test['AvSigVersion'].map(week_dict)\n",
        "  train['WeekNumber'].head()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "blsoIRsq8WFX",
        "colab_type": "text"
      },
      "source": [
        "## Feature Encoding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xX7JsavdKxq7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "numerical_attributes = [\n",
        "'AVProductStatesIdentifier',\n",
        "'Census_FirmwareManufacturerIdentifier', \n",
        "'Census_FirmwareVersionIdentifier', \n",
        "'Census_OEMModelIdentifier', \n",
        "'Census_OEMNameIdentifier', \n",
        "'Census_OSBuildNumber', \n",
        "'Census_OSBuildRevision',\n",
        "'Census_OSInstallLanguageIdentifier', \n",
        "'Census_OSUILocaleIdentifier', \n",
        "'Census_ProcessorManufacturerIdentifier', \n",
        "'Census_ProcessorModelIdentifier', \n",
        "'CountryIdentifier', \n",
        "'CityIdentifier',\n",
        "'GeoNameIdentifier', \n",
        "'IeVerIdentifier', \n",
        "'LocaleEnglishNameIdentifier', \n",
        "'OrganizationIdentifier',\n",
        "'OsBuild', \n",
        "'OsSuite', \n",
        "'RtpStateBitfield', \n",
        "'UacLuaenable',\n",
        "'Wdft_RegionIdentifier',\n",
        "'WeekNumber' \n",
        "]\n",
        "categorical_attributes = [\n",
        "'AppVersion', \n",
        "'AvSigVersion',  \n",
        "'Census_ActivationChannel', \n",
        "'Census_DeviceFamily', \n",
        "'Census_PrimaryDiskTypeName', \n",
        "'Census_ChassisTypeName',\n",
        "'Census_GenuineStateName',\n",
        "'Census_MDC2FormFactor', \n",
        "'Census_OSArchitecture', \n",
        "'Census_OSBranch', \n",
        "'Census_OSEdition', \n",
        "'Census_OSInstallTypeName', \n",
        "'Census_OSSkuName', \n",
        "'Census_OSVersion',\n",
        "'Census_OSWUAutoUpdateOptionsName', \n",
        "'Census_PowerPlatformRoleName', \n",
        "'EngineVersion',\n",
        "'Platform',\n",
        "'Processor',\n",
        "'ProductName',\n",
        "'OsBuildLab', \n",
        "'OsPlatformSubRelease', \n",
        "'OsVer', \n",
        "'SkuEdition', \n",
        "'SmartScreen', \n",
        "]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GzwvWlRRKwkp",
        "colab_type": "text"
      },
      "source": [
        "### Remove categories in train missing from test. The dataset contains many categories that are present in training data for an attribute, but are missing from testing data. Such categories are useless, and replaced with -1 prior value."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dzU4311lMMK1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(pickled):\n",
        "  for col in numerical_attributes:\n",
        "      train.loc[~train[col].isin(test[col].unique()), col] = -1\n",
        "\n",
        "  for col in categorical_attributes:\n",
        "      train[col] = train[col].cat.add_categories([-1])\n",
        "      train.loc[~train[col].isin(test[col].unique()), col] = -1"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FKcXrP-4OhBR",
        "colab_type": "text"
      },
      "source": [
        "### Encoding\n",
        "Frequency encoding is applied on the attributes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s3QSIyHiOyqU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def frequency_encoding(variable):\n",
        "    t = pd.concat([train[variable], test[variable]]).value_counts().reset_index()\n",
        "    t = t.reset_index()\n",
        "    t.loc[t[variable] == 1, 'level_0'] = np.nan\n",
        "    t.set_index('index', inplace=True)\n",
        "    max_label = t['level_0'].max() + 1\n",
        "    t.fillna(max_label, inplace=True)\n",
        "    return t.to_dict()['level_0']\n",
        "    \n",
        "if not(pickled):\n",
        "  for col in tqdm(numerical_attributes):\n",
        "      freq_enc_dict = frequency_encoding(col)\n",
        "      train[col] = train[col].map(lambda x: freq_enc_dict.get(x, np.nan)).astype('float32')\n",
        "      test[col] = test[col].map(lambda x: freq_enc_dict.get(x, np.nan)).astype('float32')\n",
        "      \n",
        "  for col in tqdm(categorical_attributes):\n",
        "      freq_enc_dict = frequency_encoding(col)\n",
        "      train[col] = train[col].map(lambda x: freq_enc_dict.get(x, np.nan)).astype('float32')\n",
        "      test[col] = test[col].map(lambda x: freq_enc_dict.get(x, np.nan)).astype('float32')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fm0jrXXS4SfO",
        "colab_type": "code",
        "outputId": "cdb4c7e7-873d-4b06-e766-2bd45342b011",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 217
        }
      },
      "source": [
        "train.head()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>ProductName</th>\n",
              "      <th>EngineVersion</th>\n",
              "      <th>AppVersion</th>\n",
              "      <th>AvSigVersion</th>\n",
              "      <th>IsBeta</th>\n",
              "      <th>RtpStateBitfield</th>\n",
              "      <th>IsSxsPassiveMode</th>\n",
              "      <th>AVProductStatesIdentifier</th>\n",
              "      <th>AVProductsInstalled</th>\n",
              "      <th>AVProductsEnabled</th>\n",
              "      <th>HasTpm</th>\n",
              "      <th>CountryIdentifier</th>\n",
              "      <th>CityIdentifier</th>\n",
              "      <th>OrganizationIdentifier</th>\n",
              "      <th>GeoNameIdentifier</th>\n",
              "      <th>LocaleEnglishNameIdentifier</th>\n",
              "      <th>Platform</th>\n",
              "      <th>Processor</th>\n",
              "      <th>OsVer</th>\n",
              "      <th>OsBuild</th>\n",
              "      <th>OsSuite</th>\n",
              "      <th>OsPlatformSubRelease</th>\n",
              "      <th>OsBuildLab</th>\n",
              "      <th>SkuEdition</th>\n",
              "      <th>IsProtected</th>\n",
              "      <th>AutoSampleOptIn</th>\n",
              "      <th>SMode</th>\n",
              "      <th>IeVerIdentifier</th>\n",
              "      <th>SmartScreen</th>\n",
              "      <th>Firewall</th>\n",
              "      <th>UacLuaenable</th>\n",
              "      <th>Census_MDC2FormFactor</th>\n",
              "      <th>Census_DeviceFamily</th>\n",
              "      <th>Census_OEMNameIdentifier</th>\n",
              "      <th>Census_OEMModelIdentifier</th>\n",
              "      <th>Census_ProcessorCoreCount</th>\n",
              "      <th>Census_ProcessorManufacturerIdentifier</th>\n",
              "      <th>Census_ProcessorModelIdentifier</th>\n",
              "      <th>Census_PrimaryDiskTotalCapacity</th>\n",
              "      <th>Census_PrimaryDiskTypeName</th>\n",
              "      <th>Census_SystemVolumeTotalCapacity</th>\n",
              "      <th>Census_HasOpticalDiskDrive</th>\n",
              "      <th>Census_TotalPhysicalRAM</th>\n",
              "      <th>Census_ChassisTypeName</th>\n",
              "      <th>Census_InternalPrimaryDiagonalDisplaySizeInInches</th>\n",
              "      <th>Census_InternalPrimaryDisplayResolutionHorizontal</th>\n",
              "      <th>Census_InternalPrimaryDisplayResolutionVertical</th>\n",
              "      <th>Census_PowerPlatformRoleName</th>\n",
              "      <th>Census_InternalBatteryNumberOfCharges</th>\n",
              "      <th>Census_OSVersion</th>\n",
              "      <th>Census_OSArchitecture</th>\n",
              "      <th>Census_OSBranch</th>\n",
              "      <th>Census_OSBuildNumber</th>\n",
              "      <th>Census_OSBuildRevision</th>\n",
              "      <th>Census_OSEdition</th>\n",
              "      <th>Census_OSSkuName</th>\n",
              "      <th>Census_OSInstallTypeName</th>\n",
              "      <th>Census_OSInstallLanguageIdentifier</th>\n",
              "      <th>Census_OSUILocaleIdentifier</th>\n",
              "      <th>Census_OSWUAutoUpdateOptionsName</th>\n",
              "      <th>Census_IsPortableOperatingSystem</th>\n",
              "      <th>Census_GenuineStateName</th>\n",
              "      <th>Census_ActivationChannel</th>\n",
              "      <th>Census_IsFlightsDisabled</th>\n",
              "      <th>Census_FlightRing</th>\n",
              "      <th>Census_ThresholdOptIn</th>\n",
              "      <th>Census_FirmwareManufacturerIdentifier</th>\n",
              "      <th>Census_FirmwareVersionIdentifier</th>\n",
              "      <th>Census_IsSecureBootEnabled</th>\n",
              "      <th>Census_IsWIMBootEnabled</th>\n",
              "      <th>Census_IsVirtualDevice</th>\n",
              "      <th>Census_IsTouchEnabled</th>\n",
              "      <th>Census_IsPenCapable</th>\n",
              "      <th>Census_IsAlwaysOnAlwaysConnectedCapable</th>\n",
              "      <th>Wdft_IsGamer</th>\n",
              "      <th>Wdft_RegionIdentifier</th>\n",
              "      <th>WeekNumber</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>134.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>1.0</td>\n",
              "      <td>206.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>21705.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>136.0</td>\n",
              "      <td>476940.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>299451.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>18.906250</td>\n",
              "      <td>1440.0</td>\n",
              "      <td>900.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.294967e+09</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2573.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>6.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>4.0</td>\n",
              "      <td>172.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>108360.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>135.0</td>\n",
              "      <td>476940.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>102385.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>13.898438</td>\n",
              "      <td>1366.0</td>\n",
              "      <td>768.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.000000e+00</td>\n",
              "      <td>19.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>16.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>NOT_SET</td>\n",
              "      <td>NaN</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1416.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>6.0</td>\n",
              "      <td>18.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>336.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>55.0</td>\n",
              "      <td>674.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>46.0</td>\n",
              "      <td>48.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>192.0</td>\n",
              "      <td>114473.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>113907.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>21.500000</td>\n",
              "      <td>1920.0</td>\n",
              "      <td>1080.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.294967e+09</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>185.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>25.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>31.0</td>\n",
              "      <td>21.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>26.0</td>\n",
              "      <td>22.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>159.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>63.0</td>\n",
              "      <td>238475.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>227116.0</td>\n",
              "      <td>0</td>\n",
              "      <td>4096.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>18.500000</td>\n",
              "      <td>1366.0</td>\n",
              "      <td>768.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.294967e+09</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>17.0</td>\n",
              "      <td>17.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>NaN</td>\n",
              "      <td>3.0</td>\n",
              "      <td>31.0</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>18.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1</td>\n",
              "      <td>40.0</td>\n",
              "      <td>14.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>2269.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>52.0</td>\n",
              "      <td>476940.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>101900.0</td>\n",
              "      <td>0</td>\n",
              "      <td>6144.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>14.000000</td>\n",
              "      <td>1366.0</td>\n",
              "      <td>768.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.000000e+00</td>\n",
              "      <td>15.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>17.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>2.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>Retail</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>130.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   ProductName  EngineVersion  ...  Wdft_RegionIdentifier  WeekNumber\n",
              "0          0.0            1.0  ...                    0.0         7.0\n",
              "1          0.0            6.0  ...                    6.0        18.0\n",
              "2          0.0            1.0  ...                    2.0         1.0\n",
              "3          0.0            1.0  ...                    2.0         1.0\n",
              "4          0.0            1.0  ...                    3.0         1.0\n",
              "\n",
              "[5 rows x 77 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "T-em8DxFo0r8",
        "colab_type": "text"
      },
      "source": [
        "Set need_pickle to 1 if you want to pickle the dataset. Pickling takes about 1 hour."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YjUoL-cZoupz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "need_pickle = 0"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "92wEyRkHo8E_",
        "colab_type": "text"
      },
      "source": [
        "Pickling the dataset."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x17IbdzxI5P7",
        "colab_type": "code",
        "outputId": "874a3dcd-fa2c-4875-a810-1c3f2dd769cb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "source": [
        "%%time\n",
        "if not(pickled) and need_pickle:\n",
        "  pickle_train = train\n",
        "  pickle_test = test\n",
        "\n",
        "  pickle_train['Date'] = train_date\n",
        "  pickle_test['Date'] = test_date\n",
        "\n",
        "  pickle_train['HasDetections'] = target_df\n",
        "  pickle_test['MachineIdentifier'] = test_machine_id \n",
        "\n",
        "  dumpPickle('train_df.pkl', pickle_train)\n",
        "  dumpPickle('test_df.pkl', pickle_test)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 4 µs, sys: 0 ns, total: 4 µs\n",
            "Wall time: 6.2 µs\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7FNuKUxyP-oi",
        "colab_type": "text"
      },
      "source": [
        "# Model "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jEMXerg7P-U_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train, x_test, y_train, y_test = train_test_split(train, target_df, test_size = 0.2, random_state = 0)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WxeK0v1WRpBn",
        "colab_type": "text"
      },
      "source": [
        "# LGBM"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YCiC15xfBpbS",
        "colab_type": "text"
      },
      "source": [
        "Boosting involves converting an ensemble of weak learners, usually decision trees into a strong learner.\n",
        "LightGBM is a gradient boosting framework that uses a tree based learning algorithm.\n",
        "\n",
        "The advantages of using LGBM for the given problem and dataset is as follows: \n",
        "\n",
        "\n",
        "1. As stated before, the given dataset consists of several categorical variables with high cardinalities. LGBM internally handles categorical variables by applying integer encoding. The disadvantage of having to develop an embedding system is forgone.\n",
        "2.   The given dataset has missing values for several attributes. LGBM is better equipped than neural networks to handle these missing values. This is done internally by ignoring missing value while splitting and later allocating it to the side reduces loss the most\n",
        "3.   LGBM is better at memory optimization for large data sets and takes lower memory to run. It also supports parallel and GPU learning speeding up the learning process \n",
        "\n",
        "To summarize, the best model for handling missing values, handling categorical features is the Light Gradient Boosting Machine. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_evD458pBexT",
        "colab_type": "text"
      },
      "source": [
        "Loading model pickle if it exists."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0A7-dPtITZJm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "LGBMModelPickleName = 'LGBMModel.pkl'\n",
        "model_pickled = pickleExists(LGBMModelPickleName) \n",
        "if model_pickled:\n",
        "  LGBM_model = loadPickle(LGBMModelPickleName)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d65_3fkwQIfU",
        "colab_type": "text"
      },
      "source": [
        "## Training"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xMyey7kaCYxy",
        "colab_type": "text"
      },
      "source": [
        "Training an LGBM model with the following parameters:\n",
        "\n",
        "*   number of leaves = 250\n",
        "*   learning rate = 0.02\n",
        "*   number of boosted trees to fit = 1800\n",
        "*   boosting type =  Traditional Gradient Boosting Decision Tree"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cjME18s5QIHq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "@jit(parallel=True)\n",
        "def mean_score(array):\n",
        "    return np.mean(array)\n",
        "\n",
        "@jit(parallel=True)\n",
        "def jit_div(array, divisor):\n",
        "    return np.divide(array, divisor)\n",
        "    \n",
        "def training(x_train, y_train):\n",
        "    train_scores = []\n",
        "    train_predictions = np.zeros(train.shape[0], dtype='float32')\n",
        "    test_predictions = np.zeros(test.shape[0], dtype='float32')\n",
        "\n",
        "    LGBM_model = lgb.LGBMClassifier(boosting_type='gbdt', \n",
        "                              num_leaves=250, \n",
        "                              n_estimators=1800, \n",
        "                              learning_rate=0.02)\n",
        "\n",
        "    LGBM_model.fit(x_train, y_train, eval_metric='auc',  \n",
        "            eval_set=[(x_test, y_test)], \n",
        "            verbose=50,\n",
        "            early_stopping_rounds=100)\n",
        "    return LGBM_model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-Uk08qI9Qxqt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not(model_pickled):\n",
        "  LGBM_model = training(x_train, y_train)\n",
        "  dumpPickle('LGBMModel.pkl', LGBM_model)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4BPm574aQp3W",
        "colab_type": "text"
      },
      "source": [
        "## Predicting"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s6nvlcZYREbp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "pred = LGBM_model.predict(x_test)\n",
        "test_score = roc_auc_score(y_test, pred)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r0CRxdHaRMO9",
        "colab_type": "code",
        "outputId": "e31a1047-4502-4d01-ebce-e1e37c27c38a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        }
      },
      "source": [
        "print(f'sklearn auc: {test_score:.4f}')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "sklearn auc: 0.6712\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NjvzF6VIRWld",
        "colab_type": "text"
      },
      "source": [
        "## Feature Importance"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cvfR40ZCRaDP",
        "colab_type": "code",
        "outputId": "4c0336be-c2f6-421c-af90-6a8d6e1161fd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 337
        }
      },
      "source": [
        "feat_imp = pd.Series(LGBM_model.feature_importances_, index = train.columns)\n",
        "feat_imp.nlargest(20).plot(kind='barh', figsize=(5,5))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f75eeadb550>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 53
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjIAAAEvCAYAAABfbCuBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nM7O25YaMmZmZtS03ZMzMzKxt/X/iIgacHU/YsQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 360x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n7glZKTfCsgm",
        "colab_type": "text"
      },
      "source": [
        "The above graph depicts the feature importance of 20 of the most important features of the train data for the model.  Note that newly engineered feature 'WeekNo' is one of the most important attributes for predicting malware. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7b1OV9IUR4Sz",
        "colab_type": "text"
      },
      "source": [
        "## Submitting to kaggle"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m4G1CjZsR665",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "submit = 0"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aOvXJOKvR8kR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if submit:\n",
        "  test_predictions = np.zeros(test.shape[0], dtype='float32')\n",
        "  pred_prob = LGBM_model.predict_proba(test)[:,1]\n",
        "  test_predictions = np.add(test_predictions, pred_prob).astype('float32')\n",
        "  test['MachineIdentifier'] = test_machine_id\n",
        "  test['HasDetections'] = test_predictions\n",
        "  test[['MachineIdentifier','HasDetections']].to_csv('submission.csv', index=False)\n",
        "  !kaggle competitions submit -c microsoft-malware-prediction -f submission.csv -m \"LGBM\""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vNGQ7ehEDOC9",
        "colab_type": "text"
      },
      "source": [
        "## Kaggle score for the LGBM model : \n",
        "###The score is calculated based on AUC. \n",
        "![](https://drive.google.com/uc?id=1Qw3iGCky7T7H-QMB-St3Q8LeQPnEJ5LY)\n",
        "<hr>\n",
        "<br>\n",
        "\n",
        "### Although, the score may seem low, it is consistent with scores in the private leaderboard when the competition was active:\n",
        "<br>\n",
        "\n",
        "## Private leaderboard : \n",
        "![](https://drive.google.com/uc?id=1RX7Pq7XqtfBC6ct1ylenjHTPEp1wQT9J)\n",
        "\n"
      ]
    }
  ]
}