youtube_adview_prediction.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "youtube_adview_prediction.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/sidharth178/Youtube-Adview-Prediction/blob/master/youtube_adview_prediction.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q6bqbVyb8ELZ"
      },
      "source": [
        "# <center>**===========YouTube AdView Prediction============**</center>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RH0yj_pW8yKF"
      },
      "source": [
        "# **1. Introduction**\n",
        "### **Objective:** \n",
        "To build a machine learning model which will predict youtube adview count based on other youtube metrics.\n",
        "### **Data Description:**\n",
        "- train.csv - the training set\n",
        "- test.csv - the test set\n",
        "- The file train.csv contains metrics and other details of about 15000 youtube videos. The metrics include number of views, likes, dislikes, comments and apart from that published date, duration and category are also included. The train.csv file also contains the metric number of adviews which is our target variable for prediction.\n",
        "\n",
        "### **Table of Content:**\n",
        "1. Introduction\n",
        "2. Install & Import Libraries\n",
        "3. Load Datasets\n",
        "4. Exploratory Data Analysis\n",
        "5. Feature Engineering\n",
        "6. Model Development\n",
        "7. Find Prediction\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iZwNON8xIs5Q"
      },
      "source": [
        "# **2. Install & Import Libraries**\n",
        "\n",
        "- Run the below cell, if you've not install these libraries before."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "collapsed": true,
        "id": "A-nNkYeYrgj6"
      },
      "source": [
        "# # use to visualize missing value\n",
        "# !pip install missingno\n",
        "\n",
        "# # use for hyper parameter tuning\n",
        "# !pip install optuna\n"
      ],
      "execution_count": 107,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "edVs2sxRI-58",
        "outputId": "9eee2dec-f1d6-4dac-ac23-9879de5f804f"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "import missingno as msno\n",
        "## Display all the columns of the dataframe\n",
        "pd.pandas.set_option('display.max_columns',None)\n",
        "\n",
        "from scipy import stats\n",
        "from scipy.stats import norm, skew # for some statistics\n",
        "import warnings # to ignore warning\n",
        "from sklearn.preprocessing import RobustScaler, PowerTransformer, LabelEncoder\n",
        "from sklearn.pipeline import make_pipeline\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "import optuna\n",
        "from sklearn.model_selection import KFold, cross_val_score\n",
        "from sklearn.linear_model import Ridge, Lasso\n",
        "from sklearn.ensemble import StackingRegressor, RandomForestRegressor, GradientBoostingRegressor\n",
        "from sklearn.svm import SVR\n",
        "import xgboost as xgb\n",
        "import lightgbm as lgb\n",
        "import joblib\n",
        "\n",
        "import warnings\n",
        "warnings.filterwarnings('ignore')\n",
        "print(\"Library Imported!!\")"
      ],
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Library Imported!!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YwLuqec9JD6y"
      },
      "source": [
        "# **3. Load Datasets**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TqC_8gRGDBzr"
      },
      "source": [
        "# load train and test dataset\n",
        "train_df = pd.read_csv(\"/content/train.csv\")\n",
        "test_df = pd.read_csv(\"/content/test.csv\")"
      ],
      "execution_count": 55,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6g2dd-GlIFIB"
      },
      "source": [
        "# **4. Exploratory Data Analysis**\n",
        "### 4.1. Train Data Exploration\n",
        "\n",
        "For both train and test dataset, We'll explore following things\n",
        "\n",
        "- First 5 rows\n",
        "- Data shape\n",
        "- Data information\n",
        "- Data types\n",
        "- Null value"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aNGZv-v0U12t"
      },
      "source": [
        "### 4.1.1. First 5 records"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 203
        },
        "id": "DNxoDXP5DB3p",
        "outputId": "9170fede-2445-422e-aeb1-b0f717c7bc74"
      },
      "source": [
        "train_df.head()"
      ],
      "execution_count": 56,
      "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>vidid</th>\n",
              "      <th>adview</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>published</th>\n",
              "      <th>duration</th>\n",
              "      <th>category</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>VID_18655</td>\n",
              "      <td>40</td>\n",
              "      <td>1031602</td>\n",
              "      <td>8523</td>\n",
              "      <td>363</td>\n",
              "      <td>1095</td>\n",
              "      <td>2016-09-14</td>\n",
              "      <td>PT7M37S</td>\n",
              "      <td>F</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>VID_14135</td>\n",
              "      <td>2</td>\n",
              "      <td>1707</td>\n",
              "      <td>56</td>\n",
              "      <td>2</td>\n",
              "      <td>6</td>\n",
              "      <td>2016-10-01</td>\n",
              "      <td>PT9M30S</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>VID_2187</td>\n",
              "      <td>1</td>\n",
              "      <td>2023</td>\n",
              "      <td>25</td>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>2016-07-02</td>\n",
              "      <td>PT2M16S</td>\n",
              "      <td>C</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>VID_23096</td>\n",
              "      <td>6</td>\n",
              "      <td>620860</td>\n",
              "      <td>777</td>\n",
              "      <td>161</td>\n",
              "      <td>153</td>\n",
              "      <td>2016-07-27</td>\n",
              "      <td>PT4M22S</td>\n",
              "      <td>H</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>VID_10175</td>\n",
              "      <td>1</td>\n",
              "      <td>666</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>2016-06-29</td>\n",
              "      <td>PT31S</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       vidid  adview    views likes dislikes comment   published duration  \\\n",
              "0  VID_18655      40  1031602  8523      363    1095  2016-09-14  PT7M37S   \n",
              "1  VID_14135       2     1707    56        2       6  2016-10-01  PT9M30S   \n",
              "2   VID_2187       1     2023    25        0       2  2016-07-02  PT2M16S   \n",
              "3  VID_23096       6   620860   777      161     153  2016-07-27  PT4M22S   \n",
              "4  VID_10175       1      666     1        0       0  2016-06-29    PT31S   \n",
              "\n",
              "  category  \n",
              "0        F  \n",
              "1        D  \n",
              "2        C  \n",
              "3        H  \n",
              "4        D  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 56
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Dwqhcr1SVHl_"
      },
      "source": [
        "### 4.1.2. Data Shape - Train Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2YgZcVkGDCT0",
        "outputId": "71ec2251-5e62-42c4-d942-cfbda5206e3d"
      },
      "source": [
        "train_df.shape"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(14999, 9)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0bpFSABeVdw6"
      },
      "source": [
        "### 4.1.3. Data Information - Train Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "collapsed": true,
        "id": "pF0s1eqYVROq",
        "outputId": "cf2247bc-11ef-442d-fa77-15f26ddeab6e"
      },
      "source": [
        "train_df.info()"
      ],
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 14999 entries, 0 to 14998\n",
            "Data columns (total 9 columns):\n",
            " #   Column     Non-Null Count  Dtype \n",
            "---  ------     --------------  ----- \n",
            " 0   vidid      14999 non-null  object\n",
            " 1   adview     14999 non-null  int64 \n",
            " 2   views      14999 non-null  object\n",
            " 3   likes      14999 non-null  object\n",
            " 4   dislikes   14999 non-null  object\n",
            " 5   comment    14999 non-null  object\n",
            " 6   published  14999 non-null  object\n",
            " 7   duration   14999 non-null  object\n",
            " 8   category   14999 non-null  object\n",
            "dtypes: int64(1), object(8)\n",
            "memory usage: 1.0+ MB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nsHU-MblVe0W"
      },
      "source": [
        "### 4.1.4. Statistical analysis - Train Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 388
        },
        "id": "B3x0Q839V07G",
        "outputId": "f1bac8c9-520a-4d6f-da03-6d9887b91ec4"
      },
      "source": [
        "train_df.describe(include='all')"
      ],
      "execution_count": 59,
      "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>vidid</th>\n",
              "      <th>adview</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>published</th>\n",
              "      <th>duration</th>\n",
              "      <th>category</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>14999</td>\n",
              "      <td>1.499900e+04</td>\n",
              "      <td>14999</td>\n",
              "      <td>14999</td>\n",
              "      <td>14999</td>\n",
              "      <td>14999</td>\n",
              "      <td>14999</td>\n",
              "      <td>14999</td>\n",
              "      <td>14999</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>unique</th>\n",
              "      <td>14999</td>\n",
              "      <td>NaN</td>\n",
              "      <td>14588</td>\n",
              "      <td>4789</td>\n",
              "      <td>1546</td>\n",
              "      <td>2007</td>\n",
              "      <td>2386</td>\n",
              "      <td>3146</td>\n",
              "      <td>8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>top</th>\n",
              "      <td>VID_12352</td>\n",
              "      <td>NaN</td>\n",
              "      <td>238</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>2016-08-26</td>\n",
              "      <td>PT31S</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>freq</th>\n",
              "      <td>1</td>\n",
              "      <td>NaN</td>\n",
              "      <td>4</td>\n",
              "      <td>174</td>\n",
              "      <td>1091</td>\n",
              "      <td>1290</td>\n",
              "      <td>42</td>\n",
              "      <td>147</td>\n",
              "      <td>7558</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>NaN</td>\n",
              "      <td>2.107791e+03</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>NaN</td>\n",
              "      <td>5.237711e+04</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>NaN</td>\n",
              "      <td>1.000000e+00</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>NaN</td>\n",
              "      <td>1.000000e+00</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>NaN</td>\n",
              "      <td>2.000000e+00</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>NaN</td>\n",
              "      <td>6.000000e+00</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>NaN</td>\n",
              "      <td>5.429665e+06</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "            vidid        adview  views  likes dislikes comment   published  \\\n",
              "count       14999  1.499900e+04  14999  14999    14999   14999       14999   \n",
              "unique      14999           NaN  14588   4789     1546    2007        2386   \n",
              "top     VID_12352           NaN    238      1        0       0  2016-08-26   \n",
              "freq            1           NaN      4    174     1091    1290          42   \n",
              "mean          NaN  2.107791e+03    NaN    NaN      NaN     NaN         NaN   \n",
              "std           NaN  5.237711e+04    NaN    NaN      NaN     NaN         NaN   \n",
              "min           NaN  1.000000e+00    NaN    NaN      NaN     NaN         NaN   \n",
              "25%           NaN  1.000000e+00    NaN    NaN      NaN     NaN         NaN   \n",
              "50%           NaN  2.000000e+00    NaN    NaN      NaN     NaN         NaN   \n",
              "75%           NaN  6.000000e+00    NaN    NaN      NaN     NaN         NaN   \n",
              "max           NaN  5.429665e+06    NaN    NaN      NaN     NaN         NaN   \n",
              "\n",
              "       duration category  \n",
              "count     14999    14999  \n",
              "unique     3146        8  \n",
              "top       PT31S        D  \n",
              "freq        147     7558  \n",
              "mean        NaN      NaN  \n",
              "std         NaN      NaN  \n",
              "min         NaN      NaN  \n",
              "25%         NaN      NaN  \n",
              "50%         NaN      NaN  \n",
              "75%         NaN      NaN  \n",
              "max         NaN      NaN  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 59
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yD_hyMVGWLZ0"
      },
      "source": [
        "### 4.1.5. Data Type - Train Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZWsab-RhVRQk",
        "outputId": "7a53b72d-5ad7-4973-aa9e-fb2eaab0b210"
      },
      "source": [
        "train_dtype = train_df.dtypes\n",
        "train_dtype.value_counts()"
      ],
      "execution_count": 60,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "object    8\n",
              "int64     1\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0mgEmXnXeGCg"
      },
      "source": [
        "### 4.1.6. Null Value - Train Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IulOWrP1WES3",
        "outputId": "5a45b281-31f7-4ca7-eedb-8cd9afce8867"
      },
      "source": [
        "train_df.isnull().sum().sort_values(ascending = False).head(10)"
      ],
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "category     0\n",
              "duration     0\n",
              "published    0\n",
              "comment      0\n",
              "dislikes     0\n",
              "likes        0\n",
              "views        0\n",
              "adview       0\n",
              "vidid        0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 61
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EP7aNnxuta5H"
      },
      "source": [
        "### 4.1.7. Visualize missing value using **Misingno** - Train Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "id": "WyQS0jHNrvGN",
        "outputId": "23f5aca0-05d3-4a59-d70e-16cdb9ac36c8"
      },
      "source": [
        "msno.matrix(train_df)"
      ],
      "execution_count": 62,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fb96dc11950>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 62
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1800x720 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "r77wbBEBjzjq"
      },
      "source": [
        "### 4.2. Test Data Exploration\n",
        "\n",
        "### 4.2.1. First 5 rows - Test Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 203
        },
        "id": "gruCW1xKVRU4",
        "outputId": "638aba79-21fb-41e0-d95b-b10a71dc1c43"
      },
      "source": [
        "test_df.head()"
      ],
      "execution_count": 63,
      "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>vidid</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>published</th>\n",
              "      <th>duration</th>\n",
              "      <th>category</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>VID_1054</td>\n",
              "      <td>440238</td>\n",
              "      <td>6153</td>\n",
              "      <td>218</td>\n",
              "      <td>1377</td>\n",
              "      <td>2017-02-18</td>\n",
              "      <td>PT7M29S</td>\n",
              "      <td>B</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>VID_18629</td>\n",
              "      <td>1040132</td>\n",
              "      <td>8171</td>\n",
              "      <td>340</td>\n",
              "      <td>1047</td>\n",
              "      <td>2016-06-28</td>\n",
              "      <td>PT6M29S</td>\n",
              "      <td>F</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>VID_13967</td>\n",
              "      <td>28534</td>\n",
              "      <td>31</td>\n",
              "      <td>11</td>\n",
              "      <td>1</td>\n",
              "      <td>2014-03-10</td>\n",
              "      <td>PT37M54S</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>VID_19442</td>\n",
              "      <td>1316715</td>\n",
              "      <td>2284</td>\n",
              "      <td>250</td>\n",
              "      <td>274</td>\n",
              "      <td>2010-06-05</td>\n",
              "      <td>PT9M55S</td>\n",
              "      <td>G</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>VID_770</td>\n",
              "      <td>1893173</td>\n",
              "      <td>2519</td>\n",
              "      <td>225</td>\n",
              "      <td>116</td>\n",
              "      <td>2016-09-03</td>\n",
              "      <td>PT3M8S</td>\n",
              "      <td>B</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       vidid    views likes dislikes comment   published  duration category\n",
              "0   VID_1054   440238  6153      218    1377  2017-02-18   PT7M29S        B\n",
              "1  VID_18629  1040132  8171      340    1047  2016-06-28   PT6M29S        F\n",
              "2  VID_13967    28534    31       11       1  2014-03-10  PT37M54S        D\n",
              "3  VID_19442  1316715  2284      250     274  2010-06-05   PT9M55S        G\n",
              "4    VID_770  1893173  2519      225     116  2016-09-03    PT3M8S        B"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 63
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y1D05eHDmbbi"
      },
      "source": [
        "### 4.2.2. Data Shape - Test Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YzEMBbHdmQee",
        "outputId": "af3e4977-3fbb-4421-b1d9-c33ed4565674"
      },
      "source": [
        "test_df.shape"
      ],
      "execution_count": 64,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(8764, 8)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 64
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OcHKdbMulbdc"
      },
      "source": [
        "### 4.2.3. Data Type - Test Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NZEjWQL0VRXB",
        "outputId": "a760cbfb-2ddf-43a6-8c6a-7175c9af057e"
      },
      "source": [
        "test_dtype = test_df.dtypes\n",
        "test_dtype.value_counts()"
      ],
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "object    8\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cNK5swKMltr0"
      },
      "source": [
        "### 4.2.4. Data Information - Test Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rYSQ1inMVRar",
        "outputId": "a4d0a8f7-36a8-4379-eae5-c2f6859220a9"
      },
      "source": [
        "test_df.info()"
      ],
      "execution_count": 66,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 8764 entries, 0 to 8763\n",
            "Data columns (total 8 columns):\n",
            " #   Column     Non-Null Count  Dtype \n",
            "---  ------     --------------  ----- \n",
            " 0   vidid      8764 non-null   object\n",
            " 1   views      8764 non-null   object\n",
            " 2   likes      8764 non-null   object\n",
            " 3   dislikes   8764 non-null   object\n",
            " 4   comment    8764 non-null   object\n",
            " 5   published  8764 non-null   object\n",
            " 6   duration   8764 non-null   object\n",
            " 7   category   8764 non-null   object\n",
            "dtypes: object(8)\n",
            "memory usage: 547.9+ KB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zAIzB-sCb0NK"
      },
      "source": [
        "### 4.2.5. Statistical analysis - Test Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 172
        },
        "id": "9_vZHDuVbzh_",
        "outputId": "38134cc5-a902-4206-b24f-e5fdd6ff5e0b"
      },
      "source": [
        "test_df.describe(include='all')"
      ],
      "execution_count": 67,
      "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>vidid</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>published</th>\n",
              "      <th>duration</th>\n",
              "      <th>category</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>8764</td>\n",
              "      <td>8764</td>\n",
              "      <td>8764</td>\n",
              "      <td>8764</td>\n",
              "      <td>8764</td>\n",
              "      <td>8764</td>\n",
              "      <td>8764</td>\n",
              "      <td>8764</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>unique</th>\n",
              "      <td>8764</td>\n",
              "      <td>8605</td>\n",
              "      <td>3434</td>\n",
              "      <td>1215</td>\n",
              "      <td>1556</td>\n",
              "      <td>2109</td>\n",
              "      <td>2330</td>\n",
              "      <td>8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>top</th>\n",
              "      <td>VID_14440</td>\n",
              "      <td>36679</td>\n",
              "      <td>2</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>2016-08-26</td>\n",
              "      <td>PT31S</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>freq</th>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>89</td>\n",
              "      <td>662</td>\n",
              "      <td>727</td>\n",
              "      <td>32</td>\n",
              "      <td>120</td>\n",
              "      <td>4419</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "            vidid  views likes dislikes comment   published duration category\n",
              "count        8764   8764  8764     8764    8764        8764     8764     8764\n",
              "unique       8764   8605  3434     1215    1556        2109     2330        8\n",
              "top     VID_14440  36679     2        0       0  2016-08-26    PT31S        D\n",
              "freq            1      3    89      662     727          32      120     4419"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 67
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iz9-fW3Bmxlq"
      },
      "source": [
        "### 4.2.6. Null Data - Test Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-kRfFhJQmQhQ",
        "outputId": "ff9e53bc-d406-44c2-83b8-688a22dcdc97"
      },
      "source": [
        "test_df.isnull().sum().sort_values(ascending = False).head(10)"
      ],
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "category     0\n",
              "duration     0\n",
              "published    0\n",
              "comment      0\n",
              "dislikes     0\n",
              "likes        0\n",
              "views        0\n",
              "vidid        0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 68
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8G5ezElLvPwy"
      },
      "source": [
        "### 4.2.7. Visualize missing value using **Misingno** - Test Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 517
        },
        "id": "JkJ3y5rWvU0M",
        "outputId": "9d2bfaec-cc92-47a4-d339-0e93b79805c5"
      },
      "source": [
        "msno.matrix(test_df)"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fb96db0c610>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 69
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1800x720 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hHSmyv0ef5oI"
      },
      "source": [
        "### 4.2.8. Report - Data Exploration\n",
        "- The shape of train and test datasets are (14999, 9) & (8764, 8)\n",
        "- There is no null value present in both dataset.\n",
        "- Some categorical columns should convert to numerical.\n",
        "- e.g 'views', 'likes', 'dislikes', 'comment'.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SPJbHi99qs9A"
      },
      "source": [
        "### 4.3. Train & Test Data Comparison\n",
        "\n",
        "Here we'll compare below things between train and test dataset.\n",
        "- Data Type\n",
        "- Null values\n",
        "- Data Distribution"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vaxYtEf3lkkr"
      },
      "source": [
        "### 4.3.1. Data Type Comparison"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 49
        },
        "id": "mHIS1zJkm65P",
        "outputId": "236bebfb-e6d3-4031-e890-0b05d63d03a9"
      },
      "source": [
        "# as 'SalePrice' Column is not available in test dataset. So we'll delete it.\n",
        "trn_dtype = train_dtype.drop('adview')\n",
        "trn_dtype.compare(test_dtype)"
      ],
      "execution_count": 70,
      "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>self</th>\n",
              "      <th>other</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "Empty DataFrame\n",
              "Columns: [self, other]\n",
              "Index: []"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 70
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "C6yMp9W8YOro"
      },
      "source": [
        "- The data type of each columns is same in both train and test dataframe"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-0OYo4q3mLbz"
      },
      "source": [
        "### 4.3.2. Null Value Comparison"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 49
        },
        "id": "i6LSUa3um69Q",
        "outputId": "7c9f579b-1f32-4991-a011-79be23c1b61e"
      },
      "source": [
        "null_train = train_df.isnull().sum()\n",
        "null_test = test_df.isnull().sum()\n",
        "null_train = null_train.drop('adview')\n",
        "null_comp_df = null_train.compare(null_test).sort_values(['self'],ascending = [False])\n",
        "null_comp_df  "
      ],
      "execution_count": 71,
      "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>self</th>\n",
              "      <th>other</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "Empty DataFrame\n",
              "Columns: [self, other]\n",
              "Index: []"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 71
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mcV8GtUIohxs"
      },
      "source": [
        "- Here we can see that there is no null value present in test and train dataset."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eTXQ96ch5mu6"
      },
      "source": [
        "### 4.3.3. Distribution Comparison\n",
        "Before going for distribution comparison,let's do some data preprocessing which will help in data analysis."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S19HPQ5vqGJv"
      },
      "source": [
        "### 4.3.3.1 Convert Categorical column to numerical"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o2-BcVSpqBQp"
      },
      "source": [
        "convert_col = ['views', 'likes', 'dislikes', 'comment']\n",
        "# these columns contain 'F' letter. So replace it by '0'. As we are converting columns to numerical.\n",
        "for col in convert_col:\n",
        "  train_df[col].replace({\"F\": 0}, inplace=True)\n",
        "  test_df[col].replace({\"F\": 0}, inplace=True)\n",
        "\n",
        "\n",
        "# Convert \"categorical\" feature to \"numerical\"\n",
        "for col in convert_col:\n",
        "  train_df[col] = train_df[col].astype('int')\n",
        "  test_df[col] = test_df[col].astype('int')"
      ],
      "execution_count": 72,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fsrOe5WQqlgO"
      },
      "source": [
        "### 4.3.3.2. Temporal variable analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 203
        },
        "id": "wYDiWyAotVQC",
        "outputId": "7877c235-4b8a-4780-f608-c9ee751e14f7"
      },
      "source": [
        "train_df.head()"
      ],
      "execution_count": 73,
      "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>vidid</th>\n",
              "      <th>adview</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>published</th>\n",
              "      <th>duration</th>\n",
              "      <th>category</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>VID_18655</td>\n",
              "      <td>40</td>\n",
              "      <td>1031602</td>\n",
              "      <td>8523</td>\n",
              "      <td>363</td>\n",
              "      <td>1095</td>\n",
              "      <td>2016-09-14</td>\n",
              "      <td>PT7M37S</td>\n",
              "      <td>F</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>VID_14135</td>\n",
              "      <td>2</td>\n",
              "      <td>1707</td>\n",
              "      <td>56</td>\n",
              "      <td>2</td>\n",
              "      <td>6</td>\n",
              "      <td>2016-10-01</td>\n",
              "      <td>PT9M30S</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>VID_2187</td>\n",
              "      <td>1</td>\n",
              "      <td>2023</td>\n",
              "      <td>25</td>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>2016-07-02</td>\n",
              "      <td>PT2M16S</td>\n",
              "      <td>C</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>VID_23096</td>\n",
              "      <td>6</td>\n",
              "      <td>620860</td>\n",
              "      <td>777</td>\n",
              "      <td>161</td>\n",
              "      <td>153</td>\n",
              "      <td>2016-07-27</td>\n",
              "      <td>PT4M22S</td>\n",
              "      <td>H</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>VID_10175</td>\n",
              "      <td>1</td>\n",
              "      <td>666</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>2016-06-29</td>\n",
              "      <td>PT31S</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       vidid  adview    views  likes  dislikes  comment   published duration  \\\n",
              "0  VID_18655      40  1031602   8523       363     1095  2016-09-14  PT7M37S   \n",
              "1  VID_14135       2     1707     56         2        6  2016-10-01  PT9M30S   \n",
              "2   VID_2187       1     2023     25         0        2  2016-07-02  PT2M16S   \n",
              "3  VID_23096       6   620860    777       161      153  2016-07-27  PT4M22S   \n",
              "4  VID_10175       1      666      1         0        0  2016-06-29    PT31S   \n",
              "\n",
              "  category  \n",
              "0        F  \n",
              "1        D  \n",
              "2        C  \n",
              "3        H  \n",
              "4        D  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 73
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ehdTdXj0qiQK"
      },
      "source": [
        "# convert \"duration\" column format into \"second\" format\n",
        "def checki(x):\n",
        "    y = x[2:]\n",
        "    h = ''\n",
        "    m = ''\n",
        "    s = ''\n",
        "    mm = ''\n",
        "    P = ['H','M','S']\n",
        "    for i in y:\n",
        "        if i not in P:\n",
        "            mm+=i\n",
        "        else:\n",
        "            if(i==\"H\"):\n",
        "                h = mm\n",
        "                mm = ''\n",
        "            elif(i == \"M\"):\n",
        "                m = mm\n",
        "                mm = ''\n",
        "            else:\n",
        "                s = mm\n",
        "                mm = ''\n",
        "    if(h==''):\n",
        "        h = '00'\n",
        "    if(m == ''):\n",
        "        m = '00'\n",
        "    if(s==''):\n",
        "        s='00'\n",
        "    bp = h+':'+m+':'+s\n",
        "    return bp\n",
        "\n",
        "train_mp = train_df[\"duration\"]\n",
        "test_mp = test_df[\"duration\"]\n",
        "\n",
        "train_time = train_mp.apply(checki)\n",
        "test_time = test_mp.apply(checki)\n",
        "\n",
        "def func_sec(time_string):\n",
        "    h, m, s = time_string.split(\":\")\n",
        "    return int(h) * 3600 + int(m) * 60 + int(s)\n",
        "\n",
        "train_time=train_time.apply(func_sec)\n",
        "test_time=test_time.apply(func_sec)\n",
        "\n",
        "train_df[\"duration\"]=train_time\n",
        "test_df[\"duration\"]=test_time\n",
        "# train_df.head()"
      ],
      "execution_count": 74,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B_bUANhOzmAk"
      },
      "source": [
        "### 4.3.3.3. Convert 'date' to 'year' format in 'published' column"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "L1w9QeI2vVpK"
      },
      "source": [
        "train_df['published'] = pd.DatetimeIndex(train_df['published']).year\n",
        "test_df['published'] = pd.DatetimeIndex(test_df['published']).year\n",
        "\n",
        "# convert to numerical feature\n",
        "train_df['published'] = train_df['published'].astype('int')\n",
        "test_df['published'] = test_df['published'].astype('int')"
      ],
      "execution_count": 75,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 203
        },
        "id": "xEK45v3LxCnr",
        "outputId": "4bb5bf94-2fe7-4479-9ad3-504beb13e585"
      },
      "source": [
        "train_df.head()"
      ],
      "execution_count": 76,
      "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>vidid</th>\n",
              "      <th>adview</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>published</th>\n",
              "      <th>duration</th>\n",
              "      <th>category</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>VID_18655</td>\n",
              "      <td>40</td>\n",
              "      <td>1031602</td>\n",
              "      <td>8523</td>\n",
              "      <td>363</td>\n",
              "      <td>1095</td>\n",
              "      <td>2016</td>\n",
              "      <td>457</td>\n",
              "      <td>F</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>VID_14135</td>\n",
              "      <td>2</td>\n",
              "      <td>1707</td>\n",
              "      <td>56</td>\n",
              "      <td>2</td>\n",
              "      <td>6</td>\n",
              "      <td>2016</td>\n",
              "      <td>570</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>VID_2187</td>\n",
              "      <td>1</td>\n",
              "      <td>2023</td>\n",
              "      <td>25</td>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>2016</td>\n",
              "      <td>136</td>\n",
              "      <td>C</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>VID_23096</td>\n",
              "      <td>6</td>\n",
              "      <td>620860</td>\n",
              "      <td>777</td>\n",
              "      <td>161</td>\n",
              "      <td>153</td>\n",
              "      <td>2016</td>\n",
              "      <td>262</td>\n",
              "      <td>H</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>VID_10175</td>\n",
              "      <td>1</td>\n",
              "      <td>666</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>2016</td>\n",
              "      <td>31</td>\n",
              "      <td>D</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       vidid  adview    views  likes  dislikes  comment  published  duration  \\\n",
              "0  VID_18655      40  1031602   8523       363     1095       2016       457   \n",
              "1  VID_14135       2     1707     56         2        6       2016       570   \n",
              "2   VID_2187       1     2023     25         0        2       2016       136   \n",
              "3  VID_23096       6   620860    777       161      153       2016       262   \n",
              "4  VID_10175       1      666      1         0        0       2016        31   \n",
              "\n",
              "  category  \n",
              "0        F  \n",
              "1        D  \n",
              "2        C  \n",
              "3        H  \n",
              "4        D  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 76
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IxaxNjKjgKKv",
        "outputId": "7a5c86a0-bc43-4de5-e85c-072bea117c8e"
      },
      "source": [
        "numerical_features = [col for col in train_df.columns if train_df[col].dtypes != 'O']\n",
        "discrete_features = [col for col in numerical_features if len(train_df[col].unique()) < 10 and col not in ['vidid']]\n",
        "continuous_features = [feature for feature in numerical_features if feature not in discrete_features+['vidid']]\n",
        "categorical_features = [col for col in train_df.columns if train_df[col].dtype == 'O']\n",
        "\n",
        "print(\"Total Number of Numerical Columns : \",len(numerical_features))\n",
        "print(\"Number of discrete features : \",len(discrete_features))\n",
        "print(\"No of continuous features are : \", len(continuous_features))\n",
        "print(\"Number of categorical features : \",len(categorical_features))"
      ],
      "execution_count": 77,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Total Number of Numerical Columns :  7\n",
            "Number of discrete features :  0\n",
            "No of continuous features are :  7\n",
            "Number of categorical features :  2\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3YdIZ1TxV1AO"
      },
      "source": [
        "### 4.3.3.4. Concat Train and Test datasets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iBGkVUWEmSuz"
      },
      "source": [
        "# combined train and test datasets\n",
        "combined_df = pd.concat([train_df,test_df],axis=0)\n",
        "\n",
        "combined_df[\"Label\"] = \"test\"\n",
        "combined_df[\"Label\"][:14999] = \"train\""
      ],
      "execution_count": 78,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uGp9-zq2nQIg"
      },
      "source": [
        "### 4.3.4. Distribution Comparison - Continuous"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 474
        },
        "id": "7c5BsoLGsl3A",
        "outputId": "3d295e11-46ab-49a9-ae2f-4a057c328e06"
      },
      "source": [
        "plt.figure(figsize=(20, 8))\n",
        "continuous_features = ['views', 'comment', 'likes', 'dislikes','published', 'duration']\n",
        "pos = 1\n",
        "for i, feature in enumerate(continuous_features):\n",
        "  plt.subplot(2 , 3 , pos)\n",
        "  sns.distplot(test_df[feature], hist = False, kde = True, kde_kws = {'linewidth': 3},color='r' )\n",
        "  sns.distplot(train_df[feature], hist = False, kde = True, kde_kws = {'linewidth': 3} ,color='b')\n",
        "  pos = pos + 1\n"
      ],
      "execution_count": 79,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x576 with 6 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QwCd_IwF_bPJ"
      },
      "source": [
        "Above distribution shows that:\n",
        "- The distribution of train and test data are similar for most continous features.\n",
        "- All distributions are not **normally distributed**."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CGBkcLd9_6br"
      },
      "source": [
        "### 4.3.5. Linearity Check\n",
        "Here we'll see the linearity between all features and the target variable."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 488
        },
        "id": "YTgiWIC3mS85",
        "outputId": "9b26a9c5-b5ec-4b27-bed7-4bf00c18f370"
      },
      "source": [
        "plt.figure(figsize=(20, 8))\n",
        "pos = 1\n",
        "for i, feature in enumerate(continuous_features):\n",
        "  plt.subplot(2 , 3 , pos)\n",
        "  sns.scatterplot(data=combined_df, x = feature, y= \"adview\")\n",
        "  pos = pos + 1\n"
      ],
      "execution_count": 80,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x576 with 6 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lKiydLMjSJNB"
      },
      "source": [
        "### 4.3.6. Distribution Comparison - Categorical \n",
        "- There are two categorical features. These are **\"category\", \"vidid\"**.\n",
        "- **\"vidid\"** is the id of video. So it has no impact to target variable.\n",
        "- So we'll only check the distribution of **\"category\"** column."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 341
        },
        "id": "Y41ZbAAamS-_",
        "outputId": "81901140-6b8a-4c3a-bfee-1f1a9a983d94"
      },
      "source": [
        "plt.figure(figsize=(30, 8))\n",
        "sns.countplot(data = combined_df, x = 'category', hue=\"Label\")"
      ],
      "execution_count": 81,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fb972109910>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 81
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 2160x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aeZq8T5nnHAN"
      },
      "source": [
        "Above distribution shows that:\n",
        "\n",
        "- The distribution of train and test data are similar for most categorical features.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fPP7bOYAmBoz"
      },
      "source": [
        "### 4.3.7. Distribution - Target Variable"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 307
        },
        "id": "j6ne0ZWFmIsT",
        "outputId": "5d4107c1-5e7f-4335-9c7a-ea15e664557a"
      },
      "source": [
        "sns.distplot(train_df[\"adview\"], hist = False, kde = True, kde_kws = {'linewidth': 3} ,color='b')"
      ],
      "execution_count": 82,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fb97219eed0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 82
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JG7-JA4Jd8T8"
      },
      "source": [
        "### 4.3.8. Data Correlation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 613
        },
        "id": "2rPQ_r3DOoik",
        "outputId": "c7a02de1-e7aa-4d82-d6c9-20b2230fd75f"
      },
      "source": [
        "training_corr = train_df.corr(method='spearman') \n",
        "plt.figure(figsize=(20,10))\n",
        "sns.heatmap(training_corr,cmap=\"YlGnBu\", linewidths=.5) "
      ],
      "execution_count": 83,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fb96e46ae50>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 83
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x720 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zdHtzXr64cF7"
      },
      "source": [
        "#**5. Feature Engineering**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Vcc7VuKBItVu"
      },
      "source": [
        "### 5.1. Drop Columns\n",
        "Here we'll drop unnecessary columns\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SH4ZnBsnWPgp",
        "outputId": "e8fc01ab-bb2b-47bf-d4bd-0b572127271e"
      },
      "source": [
        "drop_columns = [\"vidid\",'Label','published','duration']\n",
        "\n",
        "#  Drop columns\n",
        "print(\"Number of columns before dropping : \",len(combined_df.columns))\n",
        "print(\"Number of dropping columns : \",len(drop_columns))\n",
        "combined_df.drop(columns=drop_columns, inplace=True, errors='ignore')\n",
        "print(\"Number of columns after dropping : \",len(combined_df.columns))\n"
      ],
      "execution_count": 84,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of columns before dropping :  10\n",
            "Number of dropping columns :  4\n",
            "Number of columns after dropping :  6\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mQ0OC2Pl6dwU"
      },
      "source": [
        "### 5.2. Apply PowerTransformer to columns\n",
        "- We saw in distribution of continuous features that some features are not linear towards target feature. So we need to transform this. \n",
        "- Lets check the skewness of these distributions\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6DD6sC7y3pGW",
        "outputId": "c529ac29-c083-4c83-c43e-344e3a29eea3"
      },
      "source": [
        "# check the skew of all numerical features\n",
        "skew_check_col = ['views','likes','dislikes','comment']\n",
        "skewed_feats = combined_df[skew_check_col].apply(lambda x : skew(x.dropna())).sort_values(ascending = False)\n",
        "print('\\n Skew in numberical features: \\n')\n",
        "skewness_df = pd.DataFrame({'Skew' : skewed_feats})\n",
        "print(skewness_df.head(7))"
      ],
      "execution_count": 85,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            " Skew in numberical features: \n",
            "\n",
            "               Skew\n",
            "views     29.926939\n",
            "comment   18.761969\n",
            "dislikes  18.269315\n",
            "likes     11.914098\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YP4GZWuB-i0d",
        "outputId": "f64456ba-9ba1-4e3d-df41-d0e4eb86acee"
      },
      "source": [
        "\n",
        "for col in skew_check_col:\n",
        "  power = PowerTransformer(method='yeo-johnson', standardize=True)\n",
        "  combined_df[[col]] = power.fit_transform(combined_df[[col]]) # fit with combined_data to avoid overfitting with training data\n",
        "\n",
        "print('Number of skewed numerical features got transform : ', len(skew_check_col))"
      ],
      "execution_count": 86,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Number of skewed numerical features got transform :  4\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8u6RPmOS_xgg"
      },
      "source": [
        "### 5.7. Encoding Categorical Features\n",
        "\n",
        "### Get-Dummies\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1a4lHQtv-i2K"
      },
      "source": [
        "# Generate one-hot dummy columns\n",
        "combined_df = pd.get_dummies(combined_df).reset_index(drop=True)\n"
      ],
      "execution_count": 87,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 223
        },
        "id": "ZDNBu_erxces",
        "outputId": "58e1a1d0-196a-4d65-c658-96c3876fbb59"
      },
      "source": [
        "combined_df.head()"
      ],
      "execution_count": 88,
      "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>adview</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>category_A</th>\n",
              "      <th>category_B</th>\n",
              "      <th>category_C</th>\n",
              "      <th>category_D</th>\n",
              "      <th>category_E</th>\n",
              "      <th>category_F</th>\n",
              "      <th>category_G</th>\n",
              "      <th>category_H</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>40.0</td>\n",
              "      <td>1.000244</td>\n",
              "      <td>1.441680</td>\n",
              "      <td>1.103359</td>\n",
              "      <td>1.398827</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2.0</td>\n",
              "      <td>-1.693941</td>\n",
              "      <td>-0.819659</td>\n",
              "      <td>-1.200594</td>\n",
              "      <td>-0.781064</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1.0</td>\n",
              "      <td>-1.641413</td>\n",
              "      <td>-1.119185</td>\n",
              "      <td>-1.747128</td>\n",
              "      <td>-1.168045</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>6.0</td>\n",
              "      <td>0.722749</td>\n",
              "      <td>0.280058</td>\n",
              "      <td>0.723964</td>\n",
              "      <td>0.577704</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1.0</td>\n",
              "      <td>-1.970231</td>\n",
              "      <td>-2.009511</td>\n",
              "      <td>-1.747128</td>\n",
              "      <td>-1.679421</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   adview     views     likes  dislikes   comment  category_A  category_B  \\\n",
              "0    40.0  1.000244  1.441680  1.103359  1.398827           0           0   \n",
              "1     2.0 -1.693941 -0.819659 -1.200594 -0.781064           0           0   \n",
              "2     1.0 -1.641413 -1.119185 -1.747128 -1.168045           0           0   \n",
              "3     6.0  0.722749  0.280058  0.723964  0.577704           0           0   \n",
              "4     1.0 -1.970231 -2.009511 -1.747128 -1.679421           0           0   \n",
              "\n",
              "   category_C  category_D  category_E  category_F  category_G  category_H  \n",
              "0           0           0           0           1           0           0  \n",
              "1           0           1           0           0           0           0  \n",
              "2           1           0           0           0           0           0  \n",
              "3           0           0           0           0           0           1  \n",
              "4           0           1           0           0           0           0  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 88
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IIbQ0CZr-i72"
      },
      "source": [
        "new_train_data = combined_df.iloc[:len(train_df), :]\n",
        "new_test_data = combined_df.iloc[len(train_df):, :]\n",
        "X_train = new_train_data.drop('adview', axis=1)\n",
        "y_train = np.log1p(new_train_data['adview'].values.ravel())\n",
        "X_test = new_test_data.drop('adview', axis=1)"
      ],
      "execution_count": 89,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qFGC3j13-i9l",
        "outputId": "aa70203c-3fcf-4872-a2a5-39e82f928d38"
      },
      "source": [
        "# Make Pipeline\n",
        "pre_precessing_pipeline = make_pipeline(RobustScaler())\n",
        "\n",
        "X_train = pre_precessing_pipeline.fit_transform(X_train)\n",
        "X_test = pre_precessing_pipeline.transform(X_test)\n",
        "\n",
        "print(X_train.shape)\n",
        "print(X_test.shape)"
      ],
      "execution_count": 90,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(14999, 12)\n",
            "(8764, 12)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iH-Uzst4gHMQ"
      },
      "source": [
        "# **6. Model Development**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fl0B8vr6jC00"
      },
      "source": [
        "### 6.2. Hyperparameter Tuning using Optuna"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SJWqIAjTgFDO"
      },
      "source": [
        "RANDOM_SEED = 23\n",
        "\n",
        "# 10-fold CV\n",
        "kfolds = KFold(n_splits=10, shuffle=True, random_state=RANDOM_SEED)"
      ],
      "execution_count": 91,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EzF5b8S6gFGB"
      },
      "source": [
        "def tune(objective):\n",
        "    study = optuna.create_study(direction=\"maximize\")\n",
        "    study.optimize(objective, n_trials=100)\n",
        "\n",
        "    params = study.best_params\n",
        "    best_score = study.best_value\n",
        "    print(f\"Best score: {best_score} \\nOptimized parameters: {params}\")\n",
        "    return params"
      ],
      "execution_count": 92,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3oassg0djYQq"
      },
      "source": [
        "### 6.3. Ridge Regression"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "44_phjDHgFLD"
      },
      "source": [
        "def ridge_objective(trial):\n",
        "\n",
        "    _alpha = trial.suggest_float(\"alpha\", 0.1, 20)\n",
        "\n",
        "    ridge = Ridge(alpha=_alpha, random_state=RANDOM_SEED)\n",
        "\n",
        "    score = cross_val_score(\n",
        "        ridge,X_train,y_train, cv=kfolds, scoring=\"neg_root_mean_squared_error\"\n",
        "    ).mean()\n",
        "    return score\n",
        "\n",
        "# ridge_params = tune(ridge_objective)\n",
        "\n",
        "\n",
        "# Best score: -1.898690687982798 \n",
        "ridge_params = {'alpha': 19.99855836300504}"
      ],
      "execution_count": 93,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YV-MkANFgFNR",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6a17bf1f-8929-406c-b92d-67508bb4ba27"
      },
      "source": [
        "ridge = Ridge(**ridge_params, random_state=RANDOM_SEED)\n",
        "ridge.fit(X_train,y_train)"
      ],
      "execution_count": 94,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Ridge(alpha=19.99855836300504, copy_X=True, fit_intercept=True, max_iter=None,\n",
              "      normalize=False, random_state=23, solver='auto', tol=0.001)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 94
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pAlGJI0bpZAQ"
      },
      "source": [
        "### 6.4. Lasso Regression"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CzzkYSFNgFRh"
      },
      "source": [
        "def lasso_objective(trial):\n",
        "\n",
        "    _alpha = trial.suggest_float(\"alpha\", 0.0001, 1)\n",
        "\n",
        "    lasso = Lasso(alpha=_alpha, random_state=RANDOM_SEED)\n",
        "\n",
        "    score = cross_val_score(\n",
        "        lasso,X_train,y_train, cv=kfolds, scoring=\"neg_root_mean_squared_error\"\n",
        "    ).mean()\n",
        "    return score\n",
        "\n",
        "\n",
        "# lasso_params = tune(lasso_objective)\n",
        "\n",
        "# Best score: -1.8987548559962844 \n",
        "lasso_params =  {'alpha': 0.0009661425571276957}"
      ],
      "execution_count": 95,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hqU5XCv6gFT6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4cd794e0-1c5a-4086-ab19-90d0e97d659c"
      },
      "source": [
        "lasso = Lasso(**lasso_params, random_state=RANDOM_SEED)\n",
        "lasso.fit(X_train,y_train)"
      ],
      "execution_count": 96,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Lasso(alpha=0.0009661425571276957, copy_X=True, fit_intercept=True,\n",
              "      max_iter=1000, normalize=False, positive=False, precompute=False,\n",
              "      random_state=23, selection='cyclic', tol=0.0001, warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 96
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zLeveS_2xNeN"
      },
      "source": [
        "### 6.5. Gradient Boosting Regressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mwVuWK6xgFYY"
      },
      "source": [
        "def gbr_objective(trial):\n",
        "    _n_estimators = trial.suggest_int(\"n_estimators\", 50, 2000)\n",
        "    _learning_rate = trial.suggest_float(\"learning_rate\", 0.01, 1)\n",
        "    _max_depth = trial.suggest_int(\"max_depth\", 1, 20)\n",
        "    _min_samp_split = trial.suggest_int(\"min_samples_split\", 2, 20)\n",
        "    _min_samples_leaf = trial.suggest_int(\"min_samples_leaf\", 2, 20)\n",
        "    _max_features = trial.suggest_int(\"max_features\", 10, 50)\n",
        "\n",
        "    gbr = GradientBoostingRegressor(\n",
        "        n_estimators=_n_estimators,\n",
        "        learning_rate=_learning_rate,\n",
        "        max_depth=_max_depth, \n",
        "        max_features=_max_features,\n",
        "        min_samples_leaf=_min_samples_leaf,\n",
        "        min_samples_split=_min_samp_split,\n",
        "        \n",
        "        random_state=RANDOM_SEED,\n",
        "    )\n",
        "\n",
        "    score = cross_val_score(\n",
        "        gbr, X_train,y_train, cv=kfolds, scoring=\"neg_root_mean_squared_error\"\n",
        "    ).mean()\n",
        "    return score\n",
        "\n",
        "# gbr_params = tune(gbr_objective)\n",
        "\n",
        "# Best score: -1.8222372332051289 \n",
        "gbr_params = {'n_estimators': 1396, 'learning_rate': 0.014373145732630006, 'max_depth': 6, 'min_samples_split': 6, 'min_samples_leaf': 7, 'max_features': 10}"
      ],
      "execution_count": 97,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V8dMZcDqt4wV",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "33f5fece-176f-407d-9b11-0a5daa479ce0"
      },
      "source": [
        "gbr = GradientBoostingRegressor(random_state=RANDOM_SEED, **gbr_params)\n",
        "gbr.fit(X_train,y_train)"
      ],
      "execution_count": 98,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "GradientBoostingRegressor(alpha=0.9, ccp_alpha=0.0, criterion='friedman_mse',\n",
              "                          init=None, learning_rate=0.014373145732630006,\n",
              "                          loss='ls', max_depth=6, max_features=10,\n",
              "                          max_leaf_nodes=None, min_impurity_decrease=0.0,\n",
              "                          min_impurity_split=None, min_samples_leaf=7,\n",
              "                          min_samples_split=6, min_weight_fraction_leaf=0.0,\n",
              "                          n_estimators=1396, n_iter_no_change=None,\n",
              "                          presort='deprecated', random_state=23, subsample=1.0,\n",
              "                          tol=0.0001, validation_fraction=0.1, verbose=0,\n",
              "                          warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 98
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lh5-TjoFxxe7"
      },
      "source": [
        "### 6.6. XGBRegressor "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "93rj2E03t4zN"
      },
      "source": [
        "def xgb_objective(trial):\n",
        "    _n_estimators = trial.suggest_int(\"n_estimators\", 50, 2000)\n",
        "    _max_depth = trial.suggest_int(\"max_depth\", 1, 20)\n",
        "    _learning_rate = trial.suggest_float(\"learning_rate\", 0.01, 1)\n",
        "    _gamma = trial.suggest_float(\"gamma\", 0.01, 1)\n",
        "    _min_child_weight = trial.suggest_float(\"min_child_weight\", 0.1, 10)\n",
        "    _subsample = trial.suggest_float('subsample', 0.01, 1)\n",
        "    _reg_alpha = trial.suggest_float('reg_alpha', 0.01, 10)\n",
        "    _reg_lambda = trial.suggest_float('reg_lambda', 0.01, 10)\n",
        "\n",
        "    \n",
        "    xgbr = xgb.XGBRegressor(\n",
        "        n_estimators=_n_estimators,\n",
        "        max_depth=_max_depth, \n",
        "        learning_rate=_learning_rate,\n",
        "        gamma=_gamma,\n",
        "        min_child_weight=_min_child_weight,\n",
        "        subsample=_subsample,\n",
        "        reg_alpha=_reg_alpha,\n",
        "        reg_lambda=_reg_lambda,\n",
        "        random_state=RANDOM_SEED,\n",
        "    )\n",
        "    \n",
        "\n",
        "    score = cross_val_score(\n",
        "        xgbr, X_train,y_train, cv=kfolds, scoring=\"neg_root_mean_squared_error\"\n",
        "    ).mean()\n",
        "    return score\n",
        "\n",
        "\n",
        "# xgb_params = tune(xgb_objective)\n",
        "\n",
        "xgb_params =  {'n_estimators': 75, 'max_depth': 4, 'learning_rate': 0.27059503805300894, 'gamma': 0.6375378736305962, 'min_child_weight': 3.2347222003450633, 'subsample': 0.8792064649951686, 'reg_alpha': 8.764034303437914, 'reg_lambda': 7.475836220328881}\n",
        "# Best score : -1.8258592810003325."
      ],
      "execution_count": 99,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "waGAURNWt45H",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7de66240-3a3e-4137-b0e7-a0869887e255"
      },
      "source": [
        "xgbr = xgb.XGBRegressor(random_state=RANDOM_SEED, **xgb_params)\n",
        "xgbr.fit(X_train,y_train)"
      ],
      "execution_count": 100,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[16:30:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
              "             colsample_bynode=1, colsample_bytree=1, gamma=0.6375378736305962,\n",
              "             importance_type='gain', learning_rate=0.27059503805300894,\n",
              "             max_delta_step=0, max_depth=4, min_child_weight=3.2347222003450633,\n",
              "             missing=None, n_estimators=75, n_jobs=1, nthread=None,\n",
              "             objective='reg:linear', random_state=23,\n",
              "             reg_alpha=8.764034303437914, reg_lambda=7.475836220328881,\n",
              "             scale_pos_weight=1, seed=None, silent=None,\n",
              "             subsample=0.8792064649951686, verbosity=1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 100
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MD3xdWM4ySda"
      },
      "source": [
        "### 6.7. LGBMRegressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QETEMwbmt47s"
      },
      "source": [
        "import lightgbm as lgb\n",
        "\n",
        "def lgb_objective(trial):\n",
        "    _num_leaves = trial.suggest_int(\"num_leaves\", 50, 100)\n",
        "    _max_depth = trial.suggest_int(\"max_depth\", 1, 20)\n",
        "    _learning_rate = trial.suggest_float(\"learning_rate\", 0.01, 1)\n",
        "    _n_estimators = trial.suggest_int(\"n_estimators\", 50, 2000)\n",
        "    _min_child_weight = trial.suggest_float(\"min_child_weight\", 0.1, 10)\n",
        "    _reg_alpha = trial.suggest_float('reg_alpha', 0.01, 10)\n",
        "    _reg_lambda = trial.suggest_float('reg_lambda', 0.01, 10)\n",
        "    _subsample = trial.suggest_float('subsample', 0.01, 1)\n",
        "\n",
        "\n",
        "    \n",
        "    lgbr = lgb.LGBMRegressor(objective='regression',\n",
        "                             num_leaves=_num_leaves,\n",
        "                             max_depth=_max_depth,\n",
        "                             learning_rate=_learning_rate,\n",
        "                             n_estimators=_n_estimators,\n",
        "                             min_child_weight=_min_child_weight,\n",
        "                             subsample=_subsample,\n",
        "                             reg_alpha=_reg_alpha,\n",
        "                             reg_lambda=_reg_lambda,\n",
        "                             random_state=RANDOM_SEED,\n",
        "    )\n",
        "    \n",
        "\n",
        "    score = cross_val_score(\n",
        "        lgbr, X_train,y_train, cv=kfolds, scoring=\"neg_root_mean_squared_error\"\n",
        "    ).mean()\n",
        "    return score\n",
        "\n",
        "\n",
        "# lgb_params = tune(lgb_objective)\n",
        "\n",
        "# Best score: -1.824529794158143 \n",
        "lgb_params =  {'num_leaves': 84, 'max_depth': 10, 'learning_rate': 0.011076909667786489, 'n_estimators': 727, 'min_child_weight': 4.921109754366219, 'reg_alpha': 4.370797996109474, 'reg_lambda': 8.552921079737136, 'subsample': 0.4411906869457217}\n"
      ],
      "execution_count": 101,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6ZBxq_Qkt5Dy",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cb7f0dbe-160f-42e9-d162-c8399bae1573"
      },
      "source": [
        "lgbr = lgb.LGBMRegressor(objective='regression', random_state=RANDOM_SEED, **lgb_params)\n",
        "lgbr.fit(X_train,y_train)"
      ],
      "execution_count": 102,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LGBMRegressor(boosting_type='gbdt', class_weight=None, colsample_bytree=1.0,\n",
              "              importance_type='split', learning_rate=0.011076909667786489,\n",
              "              max_depth=10, min_child_samples=20,\n",
              "              min_child_weight=4.921109754366219, min_split_gain=0.0,\n",
              "              n_estimators=727, n_jobs=-1, num_leaves=84,\n",
              "              objective='regression', random_state=23,\n",
              "              reg_alpha=4.370797996109474, reg_lambda=8.552921079737136,\n",
              "              silent=True, subsample=0.4411906869457217,\n",
              "              subsample_for_bin=200000, subsample_freq=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 102
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "80iXjVmnyZUP"
      },
      "source": [
        "### 6.8. StackingRegressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vQcCR97Rt5Gq",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "94cbc17c-cb1a-4ebb-d440-bea4d5b1b1fc"
      },
      "source": [
        "# stack models\n",
        "stack = StackingRegressor(\n",
        "    estimators=[\n",
        "        ('ridge', ridge),\n",
        "        ('lasso', lasso),\n",
        "        ('gradientboostingregressor', gbr),\n",
        "        ('xgb', xgbr),\n",
        "        ('lgb', lgbr),\n",
        "        # ('svr', svr), # Not using this for now as its score is significantly worse than the others\n",
        "    ],\n",
        "    cv=kfolds)\n",
        "stack.fit(X_train,y_train)"
      ],
      "execution_count": 103,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[16:31:41] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:25] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:26] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:26] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:27] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:28] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:29] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:29] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:30] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:31] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:32] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "StackingRegressor(cv=KFold(n_splits=10, random_state=23, shuffle=True),\n",
              "                  estimators=[('ridge',\n",
              "                               Ridge(alpha=19.99855836300504, copy_X=True,\n",
              "                                     fit_intercept=True, max_iter=None,\n",
              "                                     normalize=False, random_state=23,\n",
              "                                     solver='auto', tol=0.001)),\n",
              "                              ('lasso',\n",
              "                               Lasso(alpha=0.0009661425571276957, copy_X=True,\n",
              "                                     fit_intercept=True, max_iter=1000,\n",
              "                                     normalize=False, positive=False,\n",
              "                                     precompu...\n",
              "                                             max_depth=10, min_child_samples=20,\n",
              "                                             min_child_weight=4.921109754366219,\n",
              "                                             min_split_gain=0.0,\n",
              "                                             n_estimators=727, n_jobs=-1,\n",
              "                                             num_leaves=84,\n",
              "                                             objective='regression',\n",
              "                                             random_state=23,\n",
              "                                             reg_alpha=4.370797996109474,\n",
              "                                             reg_lambda=8.552921079737136,\n",
              "                                             silent=True,\n",
              "                                             subsample=0.4411906869457217,\n",
              "                                             subsample_for_bin=200000,\n",
              "                                             subsample_freq=0))],\n",
              "                  final_estimator=None, n_jobs=None, passthrough=False,\n",
              "                  verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 103
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "koq8PaOASMx8"
      },
      "source": [
        "### 6.9. Save the Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u5rcU63bST5Y",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bb13367e-496a-4354-8d77-18ffcfd6111b"
      },
      "source": [
        "joblib.dump(stack, \"prediction_model.pkl\")\n",
        "model=joblib.load(\"prediction_model.pkl\")\n",
        "model"
      ],
      "execution_count": 104,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[16:38:53] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[16:38:53] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "StackingRegressor(cv=KFold(n_splits=10, random_state=23, shuffle=True),\n",
              "                  estimators=[('ridge',\n",
              "                               Ridge(alpha=19.99855836300504, copy_X=True,\n",
              "                                     fit_intercept=True, max_iter=None,\n",
              "                                     normalize=False, random_state=23,\n",
              "                                     solver='auto', tol=0.001)),\n",
              "                              ('lasso',\n",
              "                               Lasso(alpha=0.0009661425571276957, copy_X=True,\n",
              "                                     fit_intercept=True, max_iter=1000,\n",
              "                                     normalize=False, positive=False,\n",
              "                                     precompu...\n",
              "                                             max_depth=10, min_child_samples=20,\n",
              "                                             min_child_weight=4.921109754366219,\n",
              "                                             min_split_gain=0.0,\n",
              "                                             n_estimators=727, n_jobs=-1,\n",
              "                                             num_leaves=84,\n",
              "                                             objective='regression',\n",
              "                                             random_state=23,\n",
              "                                             reg_alpha=4.370797996109474,\n",
              "                                             reg_lambda=8.552921079737136,\n",
              "                                             silent=True,\n",
              "                                             subsample=0.4411906869457217,\n",
              "                                             subsample_for_bin=200000,\n",
              "                                             subsample_freq=0))],\n",
              "                  final_estimator=None, n_jobs=None, passthrough=False,\n",
              "                  verbose=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 104
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hUrY75J0whQS"
      },
      "source": [
        "# **7. Find Prediction**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VG2pNlIMuRtF",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "07ae3b0e-9e85-49da-d382-dc1dbeb2fddd"
      },
      "source": [
        "print('Predict submission')\n",
        "final_test_df = pd.read_csv(\"/content/test.csv\")\n",
        "\n",
        "final_test_df['AdView'] = np.round(np.expm1(model.predict(X_test))).astype(int)\n",
        "\n",
        "final_test_df.to_csv('submission_test.csv', index=False)"
      ],
      "execution_count": 105,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Predict submission\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UQg8JI_9uR17",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 203
        },
        "outputId": "617455b6-df29-4df3-d7c0-4cb1f239d325"
      },
      "source": [
        "final_test_df.head()"
      ],
      "execution_count": 106,
      "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>vidid</th>\n",
              "      <th>views</th>\n",
              "      <th>likes</th>\n",
              "      <th>dislikes</th>\n",
              "      <th>comment</th>\n",
              "      <th>published</th>\n",
              "      <th>duration</th>\n",
              "      <th>category</th>\n",
              "      <th>AdView</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>VID_1054</td>\n",
              "      <td>440238</td>\n",
              "      <td>6153</td>\n",
              "      <td>218</td>\n",
              "      <td>1377</td>\n",
              "      <td>2017-02-18</td>\n",
              "      <td>PT7M29S</td>\n",
              "      <td>B</td>\n",
              "      <td>6</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>VID_18629</td>\n",
              "      <td>1040132</td>\n",
              "      <td>8171</td>\n",
              "      <td>340</td>\n",
              "      <td>1047</td>\n",
              "      <td>2016-06-28</td>\n",
              "      <td>PT6M29S</td>\n",
              "      <td>F</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>VID_13967</td>\n",
              "      <td>28534</td>\n",
              "      <td>31</td>\n",
              "      <td>11</td>\n",
              "      <td>1</td>\n",
              "      <td>2014-03-10</td>\n",
              "      <td>PT37M54S</td>\n",
              "      <td>D</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>VID_19442</td>\n",
              "      <td>1316715</td>\n",
              "      <td>2284</td>\n",
              "      <td>250</td>\n",
              "      <td>274</td>\n",
              "      <td>2010-06-05</td>\n",
              "      <td>PT9M55S</td>\n",
              "      <td>G</td>\n",
              "      <td>6</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>VID_770</td>\n",
              "      <td>1893173</td>\n",
              "      <td>2519</td>\n",
              "      <td>225</td>\n",
              "      <td>116</td>\n",
              "      <td>2016-09-03</td>\n",
              "      <td>PT3M8S</td>\n",
              "      <td>B</td>\n",
              "      <td>6</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       vidid    views likes dislikes comment   published  duration category  \\\n",
              "0   VID_1054   440238  6153      218    1377  2017-02-18   PT7M29S        B   \n",
              "1  VID_18629  1040132  8171      340    1047  2016-06-28   PT6M29S        F   \n",
              "2  VID_13967    28534    31       11       1  2014-03-10  PT37M54S        D   \n",
              "3  VID_19442  1316715  2284      250     274  2010-06-05   PT9M55S        G   \n",
              "4    VID_770  1893173  2519      225     116  2016-09-03    PT3M8S        B   \n",
              "\n",
              "   AdView  \n",
              "0       6  \n",
              "1       4  \n",
              "2       3  \n",
              "3       6  \n",
              "4       6  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 106
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZFiX1ZYhKiiY"
      },
      "source": [
        ""
      ],
      "execution_count": 106,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bt2KgNJ0AULh"
      },
      "source": [
        ""
      ],
      "execution_count": 106,
      "outputs": []
    }
  ]
}