notebooks/0_data_preprocessing.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "5a0a3eab-a8db-493b-a344-73c9ddb139a8",
   "metadata": {},
   "source": [
    "# 0. Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4aeb45e5-b2f1-4018-8736-6a2135ed6dcc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<torch._C.Generator at 0x122e02836b0>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "import warnings\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from pandas.errors import SettingWithCopyWarning\n",
    "\n",
    "warnings.simplefilter(action=\"ignore\", category=SettingWithCopyWarning)\n",
    "warnings.simplefilter(action=\"ignore\", category=FutureWarning)\n",
    "\n",
    "import datetime\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import optuna\n",
    "import torch\n",
    "import torch.optim as optim\n",
    "from optuna.trial import TrialState\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from torch import nn\n",
    "from torch.utils.data import DataLoader, TensorDataset\n",
    "from torchsummary import summary\n",
    "from torchvision import datasets, transforms\n",
    "from torchvision.transforms import ToTensor\n",
    "from tqdm import tqdm\n",
    "\n",
    "import joblib\n",
    "\n",
    "torch.manual_seed(41)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e1b7e6af-ab41-4a52-ad48-1e7776da4220",
   "metadata": {},
   "source": [
    "# 1. Dataset Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "46e6de8b-fd65-4cf7-984a-5fd3d308dae0",
   "metadata": {},
   "source": [
    "## 1.1. Data Read In"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8ac24a1e-35f8-4573-9ffa-81ec500e98ba",
   "metadata": {},
   "source": [
    "Processed dataset from [Part 1](https://github.com/johnwslee/fine_dust_analysis) will be used as raw data for study."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "bf5d161d-67bd-4130-b5f4-f47e462ca958",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\n",
    "    \"../data/seoul_fine_dust_weather_2008_2021_for_ml.csv\",\n",
    "    index_col=0,\n",
    "    parse_dates=True,\n",
    ").reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "78c4d8f9-7c11-4330-9ff8-7a6d93c1a228",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date</th>\n",
       "      <th>temp(°C)</th>\n",
       "      <th>precipitation(mm)</th>\n",
       "      <th>wind_speed(m/s)</th>\n",
       "      <th>wind_direction</th>\n",
       "      <th>humidity(%)</th>\n",
       "      <th>local_P(hPa)</th>\n",
       "      <th>cloud_cover</th>\n",
       "      <th>lowest_ceiling(100m)</th>\n",
       "      <th>month</th>\n",
       "      <th>PM10_Counts</th>\n",
       "      <th>PM25_Counts</th>\n",
       "      <th>log_PM10_Counts</th>\n",
       "      <th>log_PM25_Counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2008-01-01 10:00:00</td>\n",
       "      <td>-7.6</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.6</td>\n",
       "      <td>340.0</td>\n",
       "      <td>47.0</td>\n",
       "      <td>1010.7</td>\n",
       "      <td>2.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>1</td>\n",
       "      <td>30.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>3.433987</td>\n",
       "      <td>2.484907</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2008-01-01 11:00:00</td>\n",
       "      <td>-6.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.1</td>\n",
       "      <td>290.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>1010.7</td>\n",
       "      <td>2.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>1</td>\n",
       "      <td>29.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>3.401197</td>\n",
       "      <td>2.639057</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2008-01-01 12:00:00</td>\n",
       "      <td>-5.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.1</td>\n",
       "      <td>290.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>1010.7</td>\n",
       "      <td>2.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>1</td>\n",
       "      <td>29.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>3.401197</td>\n",
       "      <td>2.564949</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2008-01-01 13:00:00</td>\n",
       "      <td>-5.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.1</td>\n",
       "      <td>290.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>1009.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>1</td>\n",
       "      <td>28.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>3.367296</td>\n",
       "      <td>2.564949</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2008-01-01 14:00:00</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.1</td>\n",
       "      <td>290.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>1009.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>1</td>\n",
       "      <td>27.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>3.332205</td>\n",
       "      <td>2.639057</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 date  temp(°C)  precipitation(mm)  wind_speed(m/s)  \\\n",
       "0 2008-01-01 10:00:00      -7.6                0.0              4.6   \n",
       "1 2008-01-01 11:00:00      -6.9                0.0              4.1   \n",
       "2 2008-01-01 12:00:00      -5.8                0.0              4.1   \n",
       "3 2008-01-01 13:00:00      -5.0                0.0              4.1   \n",
       "4 2008-01-01 14:00:00      -4.4                0.0              4.1   \n",
       "\n",
       "   wind_direction  humidity(%)  local_P(hPa)  cloud_cover  \\\n",
       "0           340.0         47.0        1010.7          2.0   \n",
       "1           290.0         42.0        1010.7          2.0   \n",
       "2           290.0         42.0        1010.7          2.0   \n",
       "3           290.0         36.0        1009.9          0.0   \n",
       "4           290.0         36.0        1009.9          0.0   \n",
       "\n",
       "   lowest_ceiling(100m)  month  PM10_Counts  PM25_Counts  log_PM10_Counts  \\\n",
       "0                  12.0      1         30.0         11.0         3.433987   \n",
       "1                  12.0      1         29.0         13.0         3.401197   \n",
       "2                  12.0      1         29.0         12.0         3.401197   \n",
       "3                  10.0      1         28.0         12.0         3.367296   \n",
       "4                  10.0      1         27.0         13.0         3.332205   \n",
       "\n",
       "   log_PM25_Counts  \n",
       "0         2.484907  \n",
       "1         2.639057  \n",
       "2         2.564949  \n",
       "3         2.564949  \n",
       "4         2.639057  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "64f61738-548c-4dfd-a0bf-0be9295029b0",
   "metadata": {},
   "source": [
    "The dataset above will be transformed so that it can be used for multi-scale timeseries classification. In order to do so, `wind_direction`, `humidity(%)`, `lowest_ceiling(100m)`, `temp(°C)`, `wind_speed(m/s)`, `local_P(hPa)`, `precipitation(mm)`, and `PM10_Counts` will be used as features, whereas the binary target, `Air_is_bad?` will be generated based on `PM10_Counts` of the next day."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "99454136-5685-477d-8682-b859929592b5",
   "metadata": {},
   "outputs": [],
   "source": [
    "new_columns = [\n",
    "    \"date\",\n",
    "    \"wind_direction\",\n",
    "    \"humidity(%)\",\n",
    "    \"lowest_ceiling(100m)\",\n",
    "    \"temp(°C)\",\n",
    "    \"wind_speed(m/s)\",\n",
    "    \"local_P(hPa)\",\n",
    "    \"precipitation(mm)\",\n",
    "    \"PM10_Counts\",\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "041f0731-4c90-4344-99e1-cf9b5c75581d",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df[new_columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "1b6ca8d9-2525-474a-9871-ae977dd73df7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date</th>\n",
       "      <th>wind_direction</th>\n",
       "      <th>humidity(%)</th>\n",
       "      <th>lowest_ceiling(100m)</th>\n",
       "      <th>temp(°C)</th>\n",
       "      <th>wind_speed(m/s)</th>\n",
       "      <th>local_P(hPa)</th>\n",
       "      <th>precipitation(mm)</th>\n",
       "      <th>PM10_Counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2008-01-01 10:00:00</td>\n",
       "      <td>340.0</td>\n",
       "      <td>47.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>-7.6</td>\n",
       "      <td>4.6</td>\n",
       "      <td>1010.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>30.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2008-01-01 11:00:00</td>\n",
       "      <td>290.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>-6.9</td>\n",
       "      <td>4.1</td>\n",
       "      <td>1010.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2008-01-01 12:00:00</td>\n",
       "      <td>290.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>-5.8</td>\n",
       "      <td>4.1</td>\n",
       "      <td>1010.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2008-01-01 13:00:00</td>\n",
       "      <td>290.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>-5.0</td>\n",
       "      <td>4.1</td>\n",
       "      <td>1009.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>28.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2008-01-01 14:00:00</td>\n",
       "      <td>290.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>-4.4</td>\n",
       "      <td>4.1</td>\n",
       "      <td>1009.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>27.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122728</th>\n",
       "      <td>2021-12-31 05:00:00</td>\n",
       "      <td>270.0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>-7.7</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1018.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122729</th>\n",
       "      <td>2021-12-31 06:00:00</td>\n",
       "      <td>290.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>-8.0</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1019.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>23.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122730</th>\n",
       "      <td>2021-12-31 07:00:00</td>\n",
       "      <td>270.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>-8.6</td>\n",
       "      <td>2.7</td>\n",
       "      <td>1019.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122731</th>\n",
       "      <td>2021-12-31 08:00:00</td>\n",
       "      <td>270.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>-8.8</td>\n",
       "      <td>2.9</td>\n",
       "      <td>1020.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122732</th>\n",
       "      <td>2021-12-31 09:00:00</td>\n",
       "      <td>270.0</td>\n",
       "      <td>40.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>-8.4</td>\n",
       "      <td>3.7</td>\n",
       "      <td>1021.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>122733 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                      date  wind_direction  humidity(%)  lowest_ceiling(100m)  \\\n",
       "0      2008-01-01 10:00:00           340.0         47.0                  12.0   \n",
       "1      2008-01-01 11:00:00           290.0         42.0                  12.0   \n",
       "2      2008-01-01 12:00:00           290.0         42.0                  12.0   \n",
       "3      2008-01-01 13:00:00           290.0         36.0                  10.0   \n",
       "4      2008-01-01 14:00:00           290.0         36.0                  10.0   \n",
       "...                    ...             ...          ...                   ...   \n",
       "122728 2021-12-31 05:00:00           270.0         39.0                  10.0   \n",
       "122729 2021-12-31 06:00:00           290.0         40.0                  10.0   \n",
       "122730 2021-12-31 07:00:00           270.0         42.0                  10.0   \n",
       "122731 2021-12-31 08:00:00           270.0         42.0                  10.0   \n",
       "122732 2021-12-31 09:00:00           270.0         40.0                  10.0   \n",
       "\n",
       "        temp(°C)  wind_speed(m/s)  local_P(hPa)  precipitation(mm)  \\\n",
       "0           -7.6              4.6        1010.7                0.0   \n",
       "1           -6.9              4.1        1010.7                0.0   \n",
       "2           -5.8              4.1        1010.7                0.0   \n",
       "3           -5.0              4.1        1009.9                0.0   \n",
       "4           -4.4              4.1        1009.9                0.0   \n",
       "...          ...              ...           ...                ...   \n",
       "122728      -7.7              3.5        1018.5                0.0   \n",
       "122729      -8.0              3.2        1019.2                0.0   \n",
       "122730      -8.6              2.7        1019.9                0.0   \n",
       "122731      -8.8              2.9        1020.8                0.0   \n",
       "122732      -8.4              3.7        1021.2                0.0   \n",
       "\n",
       "        PM10_Counts  \n",
       "0              30.0  \n",
       "1              29.0  \n",
       "2              29.0  \n",
       "3              28.0  \n",
       "4              27.0  \n",
       "...             ...  \n",
       "122728         25.0  \n",
       "122729         23.0  \n",
       "122730         22.0  \n",
       "122731         22.0  \n",
       "122732         22.0  \n",
       "\n",
       "[122733 rows x 9 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "84e5f1f6-1d96-414e-9432-642fc9f08343",
   "metadata": {},
   "source": [
    "Since the time starts from 10 for 2008-01-01, the dataset should start from 2008-01-02 so that all features have equal length of data per day. The same goes for 2021-12-31. So Feature will range from 2008-01-02 to 2021-12-29, while it will be from 2008-01-03 to 2021-12-30 for target."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "374404a2-0010-4f7d-b25a-ee021fcb2ff0",
   "metadata": {},
   "source": [
    "## 1.2. Target Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "f5ad6a13-a4f0-4f31-b765-e011e3c089dd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "46562"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc[df[\"PM10_Counts\"] > 45].shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "2835a72b-5cbf-4105-89ea-5502e0021215",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "122733"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "6307e271-11d6-4ba2-9887-a0699c54eb9b",
   "metadata": {},
   "outputs": [],
   "source": [
    "pm10_count = df[[\"date\", \"PM10_Counts\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "712ccbcd-7570-4ef2-bcb3-80ccff675ba6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date</th>\n",
       "      <th>PM10_Counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2008-01-01 10:00:00</td>\n",
       "      <td>30.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2008-01-01 11:00:00</td>\n",
       "      <td>29.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2008-01-01 12:00:00</td>\n",
       "      <td>29.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2008-01-01 13:00:00</td>\n",
       "      <td>28.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2008-01-01 14:00:00</td>\n",
       "      <td>27.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122728</th>\n",
       "      <td>2021-12-31 05:00:00</td>\n",
       "      <td>25.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122729</th>\n",
       "      <td>2021-12-31 06:00:00</td>\n",
       "      <td>23.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122730</th>\n",
       "      <td>2021-12-31 07:00:00</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122731</th>\n",
       "      <td>2021-12-31 08:00:00</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>122732</th>\n",
       "      <td>2021-12-31 09:00:00</td>\n",
       "      <td>22.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>122733 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                      date  PM10_Counts\n",
       "0      2008-01-01 10:00:00         30.0\n",
       "1      2008-01-01 11:00:00         29.0\n",
       "2      2008-01-01 12:00:00         29.0\n",
       "3      2008-01-01 13:00:00         28.0\n",
       "4      2008-01-01 14:00:00         27.0\n",
       "...                    ...          ...\n",
       "122728 2021-12-31 05:00:00         25.0\n",
       "122729 2021-12-31 06:00:00         23.0\n",
       "122730 2021-12-31 07:00:00         22.0\n",
       "122731 2021-12-31 08:00:00         22.0\n",
       "122732 2021-12-31 09:00:00         22.0\n",
       "\n",
       "[122733 rows x 2 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pm10_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "fd04b5bf-9175-422b-ba58-2f22f84ce533",
   "metadata": {},
   "outputs": [],
   "source": [
    "pm10_count[\"date_new\"] = pm10_count[\"date\"].dt.date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "adda68ed-62c4-4b5a-bbd4-7f5c1d2290cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "target = pm10_count.groupby(\"date_new\").mean().reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "fb422a12-30f7-47b4-826b-5a2f88c1198e",
   "metadata": {},
   "outputs": [],
   "source": [
    "target[\"date_new\"] = pd.to_datetime(target[\"date_new\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "d521ea63-70bd-4849-bedd-18e323f77364",
   "metadata": {},
   "outputs": [],
   "source": [
    "target[\"Air_is_bad?\"] = np.where(target[\"PM10_Counts\"] >= 45, 1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "46e733db-76ec-4f3a-bf23-4e63c2141009",
   "metadata": {},
   "outputs": [],
   "source": [
    "target = target[[\"date_new\", \"Air_is_bad?\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "8c7ae608-0b86-4a33-94df-57f5da8d4b10",
   "metadata": {},
   "outputs": [],
   "source": [
    "target = target.loc[\n",
    "    (target[\"date_new\"].dt.date >= datetime.date(2008, 1, 3))\n",
    "    & (target[\"date_new\"].dt.date <= datetime.date(2021, 12, 30))\n",
    "].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "28cdba5d-8143-4ea1-993b-b90b926a928a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date_new</th>\n",
       "      <th>Air_is_bad?</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2008-01-03</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2008-01-04</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2008-01-05</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2008-01-06</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2008-01-07</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5106</th>\n",
       "      <td>2021-12-26</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5107</th>\n",
       "      <td>2021-12-27</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5108</th>\n",
       "      <td>2021-12-28</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5109</th>\n",
       "      <td>2021-12-29</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5110</th>\n",
       "      <td>2021-12-30</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5111 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       date_new  Air_is_bad?\n",
       "0    2008-01-03            1\n",
       "1    2008-01-04            1\n",
       "2    2008-01-05            1\n",
       "3    2008-01-06            1\n",
       "4    2008-01-07            1\n",
       "...         ...          ...\n",
       "5106 2021-12-26            0\n",
       "5107 2021-12-27            0\n",
       "5108 2021-12-28            1\n",
       "5109 2021-12-29            1\n",
       "5110 2021-12-30            0\n",
       "\n",
       "[5111 rows x 2 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e2ab5bce-7e96-47b7-bff6-1e58aefc4851",
   "metadata": {},
   "source": [
    "## 1.3. Feature Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a391b3b0-698b-481a-95a8-5c19c023e1d1",
   "metadata": {},
   "source": [
    "The dataframe will be splitted into train and test set based on the year of 2016. The only reason why the data is splited now is to avoid fitting `StandardScaler()` on test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "7be59dc4-8eba-4e8e-9498-3b9a1f3f13c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df = df.loc[df[\"date\"].dt.date < datetime.date(2016, 1, 1)]\n",
    "test_df = df.loc[df[\"date\"].dt.date >= datetime.date(2016, 1, 1)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "1fb068bb-da45-44a5-ae6e-ed083a7ee260",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_names = list(df.columns[1:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "0a784da7-60d9-4433-8825-4b76a928bdd3",
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = StandardScaler()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "0d759dfd-d6f6-4519-9cc4-71fa45e96b4f",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df[feature_names] = scaler.fit_transform(train_df[feature_names])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "c600be03-155c-4b89-b5e9-2e2f07f504ed",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_df[feature_names] = scaler.transform(test_df[feature_names])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a282959c-b008-419e-afa2-bc64536468f4",
   "metadata": {},
   "source": [
    "The train/test df will be concatenated again just for convenience of dataset preparation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "bdacb460-eebe-4a5e-802b-d0c710443900",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['../models/scaler.pkl']"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "joblib.dump(scaler, \"../models/scaler.pkl\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "7f393522-5bf4-412a-a172-cd1fa07d8905",
   "metadata": {},
   "outputs": [],
   "source": [
    "features = pd.concat([train_df, test_df], axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "d9af4db7-d70d-40fc-8efa-e015ef4d745e",
   "metadata": {},
   "outputs": [],
   "source": [
    "features[\"date_new\"] = features[\"date\"].dt.date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "c9929c93-a31d-4680-b4aa-1725e64bb90e",
   "metadata": {},
   "outputs": [],
   "source": [
    "features = (\n",
    "    features.groupby(\"date_new\")\n",
    "    .agg(list)[\n",
    "        [\n",
    "            \"wind_direction\",\n",
    "            \"humidity(%)\",\n",
    "            \"lowest_ceiling(100m)\",\n",
    "            \"temp(°C)\",\n",
    "            \"wind_speed(m/s)\",\n",
    "            \"local_P(hPa)\",\n",
    "            \"precipitation(mm)\",\n",
    "            \"PM10_Counts\",\n",
    "        ]\n",
    "    ]\n",
    "    .reset_index()\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "88e070c0-3f25-4dd6-954d-a47453162799",
   "metadata": {},
   "outputs": [],
   "source": [
    "features[\"date_new\"] = pd.to_datetime(features[\"date_new\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "aa88c6b9-1058-486e-8cb7-8bbdb3525bfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "features = features.loc[\n",
    "    (features[\"date_new\"].dt.date >= datetime.date(2008, 1, 2))\n",
    "    & (features[\"date_new\"].dt.date <= datetime.date(2021, 12, 29))\n",
    "].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "4a255bdf-530f-4ff6-bb80-b11f4ec3d442",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date_new</th>\n",
       "      <th>wind_direction</th>\n",
       "      <th>humidity(%)</th>\n",
       "      <th>lowest_ceiling(100m)</th>\n",
       "      <th>temp(°C)</th>\n",
       "      <th>wind_speed(m/s)</th>\n",
       "      <th>local_P(hPa)</th>\n",
       "      <th>precipitation(mm)</th>\n",
       "      <th>PM10_Counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2008-01-02</td>\n",
       "      <td>[0.924452475006894, 1.1974764438963235, 1.1974...</td>\n",
       "      <td>[-0.16491600114460359, -0.665906724059201, -0....</td>\n",
       "      <td>[-0.5452500390488106, -0.5452500390488106, -0....</td>\n",
       "      <td>[-1.7379632433239907, -1.62958516404385, -1.53...</td>\n",
       "      <td>[-0.5164077188699413, 0.11713131692002768, 0.1...</td>\n",
       "      <td>[0.9043753409882764, 1.1817998051850536, 1.181...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.32022983414764833, -0.15608473051877228, -...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2008-01-03</td>\n",
       "      <td>[-1.7147792242575914, -1.2597392761085422, -1....</td>\n",
       "      <td>[-0.06471785656168408, -0.11481692885314383, -...</td>\n",
       "      <td>[-0.22368526895253332, -0.8668148091450879, -0...</td>\n",
       "      <td>[-1.3496084592368196, -1.1870413403166087, -1....</td>\n",
       "      <td>[-1.8538790166487649, -0.6571941712677122, -0....</td>\n",
       "      <td>[1.1187487905948783, 0.7782733118079257, 0.778...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.15608473051877228, 0.19956299401045927, 0....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2008-01-04</td>\n",
       "      <td>[0.924452475006894, -1.5327632449979716, 1.379...</td>\n",
       "      <td>[0.035480288021235415, -0.01461878427022433, -...</td>\n",
       "      <td>[-0.8668148091450879, -0.8668148091450879, -0....</td>\n",
       "      <td>[-1.2231673667433223, -1.2592933931700356, -1....</td>\n",
       "      <td>[0.046738090721142475, -0.5164077188699413, -1...</td>\n",
       "      <td>[1.1817998051850536, 1.3331222402014657, 1.333...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.500495683996732, 1.1570760985122364, 1.4032...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2008-01-05</td>\n",
       "      <td>[0.7424364957472743, -1.5327632449979716, -1.2...</td>\n",
       "      <td>[-0.6158076517677413, 0.035480288021235415, -0...</td>\n",
       "      <td>[-0.5452500390488106, -0.5452500390488106, -0....</td>\n",
       "      <td>[-1.1328523006765383, -1.12382079406986, -0.97...</td>\n",
       "      <td>[-0.9387670760632539, -0.6571941712677122, -1....</td>\n",
       "      <td>[0.9548161526604138, 0.9043753409882764, 0.904...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.500495683996732, 0.9929309948833603, 1.4032...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2008-01-06</td>\n",
       "      <td>[-1.5327632449979716, -0.8957073175893028, -0....</td>\n",
       "      <td>[0.8871645169760511, 0.5364710109358329, 0.436...</td>\n",
       "      <td>[-0.4380617823500515, -0.974003065843847, -0.9...</td>\n",
       "      <td>[-1.0515687412164327, -0.8619071024761865, -0....</td>\n",
       "      <td>[-1.0091603022621394, -1.0795535284610247, -1....</td>\n",
       "      <td>[0.9548161526604138, 0.9043753409882764, 0.904...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.36370809763933537, 1.512723823041468, 1.649...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5106</th>\n",
       "      <td>2021-12-25</td>\n",
       "      <td>[0.924452475006894, 0.924452475006894, 0.74243...</td>\n",
       "      <td>[-0.4655104348933621, -0.9164020855164998, -1....</td>\n",
       "      <td>[-0.11649701225377423, -0.11649701225377423, -...</td>\n",
       "      <td>[-1.8102152961774178, -2.252759119904659, -2.1...</td>\n",
       "      <td>[1.5953890670966224, 0.609883900312226, 0.6802...</td>\n",
       "      <td>[1.2700712256112905, 2.039293603611432, 2.0392...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.21079976506173095, -0.7853076277627973, -0...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5107</th>\n",
       "      <td>2021-12-26</td>\n",
       "      <td>[0.7424364957472743, 0.7424364957472743, 0.924...</td>\n",
       "      <td>[-0.7160057963506608, -0.8162039409335803, -0....</td>\n",
       "      <td>[-0.11649701225377423, -0.11649701225377423, -...</td>\n",
       "      <td>[-2.451452265251584, -2.388231719004835, -2.29...</td>\n",
       "      <td>[-0.3756212664721703, 0.6802771265111115, 0.18...</td>\n",
       "      <td>[2.3671588794803324, 2.518481314496759, 2.4680...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.8126651450342767, -0.9494527313916734, -0....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5108</th>\n",
       "      <td>2021-12-27</td>\n",
       "      <td>[1.1974764438963235, -1.2597392761085422, -0.8...</td>\n",
       "      <td>[-0.36531229031044254, -0.31521321801898283, -...</td>\n",
       "      <td>[-0.11649701225377423, -0.11649701225377423, -...</td>\n",
       "      <td>[-2.2437276132979806, -1.9185933754575586, -1....</td>\n",
       "      <td>[-0.23483481407439957, -0.5164077188699413, -0...</td>\n",
       "      <td>[2.1275650240376835, 1.963632386103219, 1.9258...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.8673801795772353, -0.48437493777652446, -0...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5109</th>\n",
       "      <td>2021-12-28</td>\n",
       "      <td>[-1.5327632449979716, -1.2597392761085422, -1....</td>\n",
       "      <td>[1.1376598784333498, 1.0374617338504304, 0.636...</td>\n",
       "      <td>[-0.3308735256512924, -0.3308735256512924, -0....</td>\n",
       "      <td>[-1.7199002301106339, -1.6657111904705635, -1....</td>\n",
       "      <td>[-0.30522804027328476, -0.3756212664721703, -0...</td>\n",
       "      <td>[1.7114283277425177, 1.7240385306605557, 1.686...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.18344224779025162, 0.06277540765306251, 0....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5110</th>\n",
       "      <td>2021-12-29</td>\n",
       "      <td>[-1.5327632449979716, -1.0777232968489225, -1....</td>\n",
       "      <td>[0.8871645169760511, 1.0374617338504304, 0.887...</td>\n",
       "      <td>[-0.11649701225377423, -0.4380617823500515, -0...</td>\n",
       "      <td>[-1.4399235253036036, -1.1328523006765383, -1....</td>\n",
       "      <td>[-0.7275873974665975, -1.0091603022621394, -0....</td>\n",
       "      <td>[0.9800365584964896, 0.43779783302097325, 0.39...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.14484795946750056, 0.36370809763933537, 0.5...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5111 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       date_new                                     wind_direction  \\\n",
       "0    2008-01-02  [0.924452475006894, 1.1974764438963235, 1.1974...   \n",
       "1    2008-01-03  [-1.7147792242575914, -1.2597392761085422, -1....   \n",
       "2    2008-01-04  [0.924452475006894, -1.5327632449979716, 1.379...   \n",
       "3    2008-01-05  [0.7424364957472743, -1.5327632449979716, -1.2...   \n",
       "4    2008-01-06  [-1.5327632449979716, -0.8957073175893028, -0....   \n",
       "...         ...                                                ...   \n",
       "5106 2021-12-25  [0.924452475006894, 0.924452475006894, 0.74243...   \n",
       "5107 2021-12-26  [0.7424364957472743, 0.7424364957472743, 0.924...   \n",
       "5108 2021-12-27  [1.1974764438963235, -1.2597392761085422, -0.8...   \n",
       "5109 2021-12-28  [-1.5327632449979716, -1.2597392761085422, -1....   \n",
       "5110 2021-12-29  [-1.5327632449979716, -1.0777232968489225, -1....   \n",
       "\n",
       "                                            humidity(%)  \\\n",
       "0     [-0.16491600114460359, -0.665906724059201, -0....   \n",
       "1     [-0.06471785656168408, -0.11481692885314383, -...   \n",
       "2     [0.035480288021235415, -0.01461878427022433, -...   \n",
       "3     [-0.6158076517677413, 0.035480288021235415, -0...   \n",
       "4     [0.8871645169760511, 0.5364710109358329, 0.436...   \n",
       "...                                                 ...   \n",
       "5106  [-0.4655104348933621, -0.9164020855164998, -1....   \n",
       "5107  [-0.7160057963506608, -0.8162039409335803, -0....   \n",
       "5108  [-0.36531229031044254, -0.31521321801898283, -...   \n",
       "5109  [1.1376598784333498, 1.0374617338504304, 0.636...   \n",
       "5110  [0.8871645169760511, 1.0374617338504304, 0.887...   \n",
       "\n",
       "                                   lowest_ceiling(100m)  \\\n",
       "0     [-0.5452500390488106, -0.5452500390488106, -0....   \n",
       "1     [-0.22368526895253332, -0.8668148091450879, -0...   \n",
       "2     [-0.8668148091450879, -0.8668148091450879, -0....   \n",
       "3     [-0.5452500390488106, -0.5452500390488106, -0....   \n",
       "4     [-0.4380617823500515, -0.974003065843847, -0.9...   \n",
       "...                                                 ...   \n",
       "5106  [-0.11649701225377423, -0.11649701225377423, -...   \n",
       "5107  [-0.11649701225377423, -0.11649701225377423, -...   \n",
       "5108  [-0.11649701225377423, -0.11649701225377423, -...   \n",
       "5109  [-0.3308735256512924, -0.3308735256512924, -0....   \n",
       "5110  [-0.11649701225377423, -0.4380617823500515, -0...   \n",
       "\n",
       "                                               temp(°C)  \\\n",
       "0     [-1.7379632433239907, -1.62958516404385, -1.53...   \n",
       "1     [-1.3496084592368196, -1.1870413403166087, -1....   \n",
       "2     [-1.2231673667433223, -1.2592933931700356, -1....   \n",
       "3     [-1.1328523006765383, -1.12382079406986, -0.97...   \n",
       "4     [-1.0515687412164327, -0.8619071024761865, -0....   \n",
       "...                                                 ...   \n",
       "5106  [-1.8102152961774178, -2.252759119904659, -2.1...   \n",
       "5107  [-2.451452265251584, -2.388231719004835, -2.29...   \n",
       "5108  [-2.2437276132979806, -1.9185933754575586, -1....   \n",
       "5109  [-1.7199002301106339, -1.6657111904705635, -1....   \n",
       "5110  [-1.4399235253036036, -1.1328523006765383, -1....   \n",
       "\n",
       "                                        wind_speed(m/s)  \\\n",
       "0     [-0.5164077188699413, 0.11713131692002768, 0.1...   \n",
       "1     [-1.8538790166487649, -0.6571941712677122, -0....   \n",
       "2     [0.046738090721142475, -0.5164077188699413, -1...   \n",
       "3     [-0.9387670760632539, -0.6571941712677122, -1....   \n",
       "4     [-1.0091603022621394, -1.0795535284610247, -1....   \n",
       "...                                                 ...   \n",
       "5106  [1.5953890670966224, 0.609883900312226, 0.6802...   \n",
       "5107  [-0.3756212664721703, 0.6802771265111115, 0.18...   \n",
       "5108  [-0.23483481407439957, -0.5164077188699413, -0...   \n",
       "5109  [-0.30522804027328476, -0.3756212664721703, -0...   \n",
       "5110  [-0.7275873974665975, -1.0091603022621394, -0....   \n",
       "\n",
       "                                           local_P(hPa)  \\\n",
       "0     [0.9043753409882764, 1.1817998051850536, 1.181...   \n",
       "1     [1.1187487905948783, 0.7782733118079257, 0.778...   \n",
       "2     [1.1817998051850536, 1.3331222402014657, 1.333...   \n",
       "3     [0.9548161526604138, 0.9043753409882764, 0.904...   \n",
       "4     [0.9548161526604138, 0.9043753409882764, 0.904...   \n",
       "...                                                 ...   \n",
       "5106  [1.2700712256112905, 2.039293603611432, 2.0392...   \n",
       "5107  [2.3671588794803324, 2.518481314496759, 2.4680...   \n",
       "5108  [2.1275650240376835, 1.963632386103219, 1.9258...   \n",
       "5109  [1.7114283277425177, 1.7240385306605557, 1.686...   \n",
       "5110  [0.9800365584964896, 0.43779783302097325, 0.39...   \n",
       "\n",
       "                                      precipitation(mm)  \\\n",
       "0     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "1     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "2     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "3     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "4     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "...                                                 ...   \n",
       "5106  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5107  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5108  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5109  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5110  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "\n",
       "                                            PM10_Counts  \n",
       "0     [-0.32022983414764833, -0.15608473051877228, -...  \n",
       "1     [-0.15608473051877228, 0.19956299401045927, 0....  \n",
       "2     [0.500495683996732, 1.1570760985122364, 1.4032...  \n",
       "3     [0.500495683996732, 0.9929309948833603, 1.4032...  \n",
       "4     [0.36370809763933537, 1.512723823041468, 1.649...  \n",
       "...                                                 ...  \n",
       "5106  [-0.21079976506173095, -0.7853076277627973, -0...  \n",
       "5107  [-0.8126651450342767, -0.9494527313916734, -0....  \n",
       "5108  [-0.8673801795772353, -0.48437493777652446, -0...  \n",
       "5109  [-0.18344224779025162, 0.06277540765306251, 0....  \n",
       "5110  [0.14484795946750056, 0.36370809763933537, 0.5...  \n",
       "\n",
       "[5111 rows x 9 columns]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63d1f821-e1d7-4fd4-b2be-216e6f4c3dae",
   "metadata": {},
   "source": [
    "## 1.4. Concatenation of Features and Target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "99e99638-f064-4daa-822e-79214a8ab21a",
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.concat([features, target[\"Air_is_bad?\"]], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "aab73ecc-119b-4fc9-b225-4c9d2f96b3c9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>date_new</th>\n",
       "      <th>wind_direction</th>\n",
       "      <th>humidity(%)</th>\n",
       "      <th>lowest_ceiling(100m)</th>\n",
       "      <th>temp(°C)</th>\n",
       "      <th>wind_speed(m/s)</th>\n",
       "      <th>local_P(hPa)</th>\n",
       "      <th>precipitation(mm)</th>\n",
       "      <th>PM10_Counts</th>\n",
       "      <th>Air_is_bad?</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2008-01-02</td>\n",
       "      <td>[0.924452475006894, 1.1974764438963235, 1.1974...</td>\n",
       "      <td>[-0.16491600114460359, -0.665906724059201, -0....</td>\n",
       "      <td>[-0.5452500390488106, -0.5452500390488106, -0....</td>\n",
       "      <td>[-1.7379632433239907, -1.62958516404385, -1.53...</td>\n",
       "      <td>[-0.5164077188699413, 0.11713131692002768, 0.1...</td>\n",
       "      <td>[0.9043753409882764, 1.1817998051850536, 1.181...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.32022983414764833, -0.15608473051877228, -...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2008-01-03</td>\n",
       "      <td>[-1.7147792242575914, -1.2597392761085422, -1....</td>\n",
       "      <td>[-0.06471785656168408, -0.11481692885314383, -...</td>\n",
       "      <td>[-0.22368526895253332, -0.8668148091450879, -0...</td>\n",
       "      <td>[-1.3496084592368196, -1.1870413403166087, -1....</td>\n",
       "      <td>[-1.8538790166487649, -0.6571941712677122, -0....</td>\n",
       "      <td>[1.1187487905948783, 0.7782733118079257, 0.778...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.15608473051877228, 0.19956299401045927, 0....</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2008-01-04</td>\n",
       "      <td>[0.924452475006894, -1.5327632449979716, 1.379...</td>\n",
       "      <td>[0.035480288021235415, -0.01461878427022433, -...</td>\n",
       "      <td>[-0.8668148091450879, -0.8668148091450879, -0....</td>\n",
       "      <td>[-1.2231673667433223, -1.2592933931700356, -1....</td>\n",
       "      <td>[0.046738090721142475, -0.5164077188699413, -1...</td>\n",
       "      <td>[1.1817998051850536, 1.3331222402014657, 1.333...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.500495683996732, 1.1570760985122364, 1.4032...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2008-01-05</td>\n",
       "      <td>[0.7424364957472743, -1.5327632449979716, -1.2...</td>\n",
       "      <td>[-0.6158076517677413, 0.035480288021235415, -0...</td>\n",
       "      <td>[-0.5452500390488106, -0.5452500390488106, -0....</td>\n",
       "      <td>[-1.1328523006765383, -1.12382079406986, -0.97...</td>\n",
       "      <td>[-0.9387670760632539, -0.6571941712677122, -1....</td>\n",
       "      <td>[0.9548161526604138, 0.9043753409882764, 0.904...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.500495683996732, 0.9929309948833603, 1.4032...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2008-01-06</td>\n",
       "      <td>[-1.5327632449979716, -0.8957073175893028, -0....</td>\n",
       "      <td>[0.8871645169760511, 0.5364710109358329, 0.436...</td>\n",
       "      <td>[-0.4380617823500515, -0.974003065843847, -0.9...</td>\n",
       "      <td>[-1.0515687412164327, -0.8619071024761865, -0....</td>\n",
       "      <td>[-1.0091603022621394, -1.0795535284610247, -1....</td>\n",
       "      <td>[0.9548161526604138, 0.9043753409882764, 0.904...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.36370809763933537, 1.512723823041468, 1.649...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5106</th>\n",
       "      <td>2021-12-25</td>\n",
       "      <td>[0.924452475006894, 0.924452475006894, 0.74243...</td>\n",
       "      <td>[-0.4655104348933621, -0.9164020855164998, -1....</td>\n",
       "      <td>[-0.11649701225377423, -0.11649701225377423, -...</td>\n",
       "      <td>[-1.8102152961774178, -2.252759119904659, -2.1...</td>\n",
       "      <td>[1.5953890670966224, 0.609883900312226, 0.6802...</td>\n",
       "      <td>[1.2700712256112905, 2.039293603611432, 2.0392...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.21079976506173095, -0.7853076277627973, -0...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5107</th>\n",
       "      <td>2021-12-26</td>\n",
       "      <td>[0.7424364957472743, 0.7424364957472743, 0.924...</td>\n",
       "      <td>[-0.7160057963506608, -0.8162039409335803, -0....</td>\n",
       "      <td>[-0.11649701225377423, -0.11649701225377423, -...</td>\n",
       "      <td>[-2.451452265251584, -2.388231719004835, -2.29...</td>\n",
       "      <td>[-0.3756212664721703, 0.6802771265111115, 0.18...</td>\n",
       "      <td>[2.3671588794803324, 2.518481314496759, 2.4680...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.8126651450342767, -0.9494527313916734, -0....</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5108</th>\n",
       "      <td>2021-12-27</td>\n",
       "      <td>[1.1974764438963235, -1.2597392761085422, -0.8...</td>\n",
       "      <td>[-0.36531229031044254, -0.31521321801898283, -...</td>\n",
       "      <td>[-0.11649701225377423, -0.11649701225377423, -...</td>\n",
       "      <td>[-2.2437276132979806, -1.9185933754575586, -1....</td>\n",
       "      <td>[-0.23483481407439957, -0.5164077188699413, -0...</td>\n",
       "      <td>[2.1275650240376835, 1.963632386103219, 1.9258...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.8673801795772353, -0.48437493777652446, -0...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5109</th>\n",
       "      <td>2021-12-28</td>\n",
       "      <td>[-1.5327632449979716, -1.2597392761085422, -1....</td>\n",
       "      <td>[1.1376598784333498, 1.0374617338504304, 0.636...</td>\n",
       "      <td>[-0.3308735256512924, -0.3308735256512924, -0....</td>\n",
       "      <td>[-1.7199002301106339, -1.6657111904705635, -1....</td>\n",
       "      <td>[-0.30522804027328476, -0.3756212664721703, -0...</td>\n",
       "      <td>[1.7114283277425177, 1.7240385306605557, 1.686...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[-0.18344224779025162, 0.06277540765306251, 0....</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5110</th>\n",
       "      <td>2021-12-29</td>\n",
       "      <td>[-1.5327632449979716, -1.0777232968489225, -1....</td>\n",
       "      <td>[0.8871645169760511, 1.0374617338504304, 0.887...</td>\n",
       "      <td>[-0.11649701225377423, -0.4380617823500515, -0...</td>\n",
       "      <td>[-1.4399235253036036, -1.1328523006765383, -1....</td>\n",
       "      <td>[-0.7275873974665975, -1.0091603022621394, -0....</td>\n",
       "      <td>[0.9800365584964896, 0.43779783302097325, 0.39...</td>\n",
       "      <td>[-0.11906781300263639, -0.11906781300263639, -...</td>\n",
       "      <td>[0.14484795946750056, 0.36370809763933537, 0.5...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5111 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       date_new                                     wind_direction  \\\n",
       "0    2008-01-02  [0.924452475006894, 1.1974764438963235, 1.1974...   \n",
       "1    2008-01-03  [-1.7147792242575914, -1.2597392761085422, -1....   \n",
       "2    2008-01-04  [0.924452475006894, -1.5327632449979716, 1.379...   \n",
       "3    2008-01-05  [0.7424364957472743, -1.5327632449979716, -1.2...   \n",
       "4    2008-01-06  [-1.5327632449979716, -0.8957073175893028, -0....   \n",
       "...         ...                                                ...   \n",
       "5106 2021-12-25  [0.924452475006894, 0.924452475006894, 0.74243...   \n",
       "5107 2021-12-26  [0.7424364957472743, 0.7424364957472743, 0.924...   \n",
       "5108 2021-12-27  [1.1974764438963235, -1.2597392761085422, -0.8...   \n",
       "5109 2021-12-28  [-1.5327632449979716, -1.2597392761085422, -1....   \n",
       "5110 2021-12-29  [-1.5327632449979716, -1.0777232968489225, -1....   \n",
       "\n",
       "                                            humidity(%)  \\\n",
       "0     [-0.16491600114460359, -0.665906724059201, -0....   \n",
       "1     [-0.06471785656168408, -0.11481692885314383, -...   \n",
       "2     [0.035480288021235415, -0.01461878427022433, -...   \n",
       "3     [-0.6158076517677413, 0.035480288021235415, -0...   \n",
       "4     [0.8871645169760511, 0.5364710109358329, 0.436...   \n",
       "...                                                 ...   \n",
       "5106  [-0.4655104348933621, -0.9164020855164998, -1....   \n",
       "5107  [-0.7160057963506608, -0.8162039409335803, -0....   \n",
       "5108  [-0.36531229031044254, -0.31521321801898283, -...   \n",
       "5109  [1.1376598784333498, 1.0374617338504304, 0.636...   \n",
       "5110  [0.8871645169760511, 1.0374617338504304, 0.887...   \n",
       "\n",
       "                                   lowest_ceiling(100m)  \\\n",
       "0     [-0.5452500390488106, -0.5452500390488106, -0....   \n",
       "1     [-0.22368526895253332, -0.8668148091450879, -0...   \n",
       "2     [-0.8668148091450879, -0.8668148091450879, -0....   \n",
       "3     [-0.5452500390488106, -0.5452500390488106, -0....   \n",
       "4     [-0.4380617823500515, -0.974003065843847, -0.9...   \n",
       "...                                                 ...   \n",
       "5106  [-0.11649701225377423, -0.11649701225377423, -...   \n",
       "5107  [-0.11649701225377423, -0.11649701225377423, -...   \n",
       "5108  [-0.11649701225377423, -0.11649701225377423, -...   \n",
       "5109  [-0.3308735256512924, -0.3308735256512924, -0....   \n",
       "5110  [-0.11649701225377423, -0.4380617823500515, -0...   \n",
       "\n",
       "                                               temp(°C)  \\\n",
       "0     [-1.7379632433239907, -1.62958516404385, -1.53...   \n",
       "1     [-1.3496084592368196, -1.1870413403166087, -1....   \n",
       "2     [-1.2231673667433223, -1.2592933931700356, -1....   \n",
       "3     [-1.1328523006765383, -1.12382079406986, -0.97...   \n",
       "4     [-1.0515687412164327, -0.8619071024761865, -0....   \n",
       "...                                                 ...   \n",
       "5106  [-1.8102152961774178, -2.252759119904659, -2.1...   \n",
       "5107  [-2.451452265251584, -2.388231719004835, -2.29...   \n",
       "5108  [-2.2437276132979806, -1.9185933754575586, -1....   \n",
       "5109  [-1.7199002301106339, -1.6657111904705635, -1....   \n",
       "5110  [-1.4399235253036036, -1.1328523006765383, -1....   \n",
       "\n",
       "                                        wind_speed(m/s)  \\\n",
       "0     [-0.5164077188699413, 0.11713131692002768, 0.1...   \n",
       "1     [-1.8538790166487649, -0.6571941712677122, -0....   \n",
       "2     [0.046738090721142475, -0.5164077188699413, -1...   \n",
       "3     [-0.9387670760632539, -0.6571941712677122, -1....   \n",
       "4     [-1.0091603022621394, -1.0795535284610247, -1....   \n",
       "...                                                 ...   \n",
       "5106  [1.5953890670966224, 0.609883900312226, 0.6802...   \n",
       "5107  [-0.3756212664721703, 0.6802771265111115, 0.18...   \n",
       "5108  [-0.23483481407439957, -0.5164077188699413, -0...   \n",
       "5109  [-0.30522804027328476, -0.3756212664721703, -0...   \n",
       "5110  [-0.7275873974665975, -1.0091603022621394, -0....   \n",
       "\n",
       "                                           local_P(hPa)  \\\n",
       "0     [0.9043753409882764, 1.1817998051850536, 1.181...   \n",
       "1     [1.1187487905948783, 0.7782733118079257, 0.778...   \n",
       "2     [1.1817998051850536, 1.3331222402014657, 1.333...   \n",
       "3     [0.9548161526604138, 0.9043753409882764, 0.904...   \n",
       "4     [0.9548161526604138, 0.9043753409882764, 0.904...   \n",
       "...                                                 ...   \n",
       "5106  [1.2700712256112905, 2.039293603611432, 2.0392...   \n",
       "5107  [2.3671588794803324, 2.518481314496759, 2.4680...   \n",
       "5108  [2.1275650240376835, 1.963632386103219, 1.9258...   \n",
       "5109  [1.7114283277425177, 1.7240385306605557, 1.686...   \n",
       "5110  [0.9800365584964896, 0.43779783302097325, 0.39...   \n",
       "\n",
       "                                      precipitation(mm)  \\\n",
       "0     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "1     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "2     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "3     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "4     [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "...                                                 ...   \n",
       "5106  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5107  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5108  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5109  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "5110  [-0.11906781300263639, -0.11906781300263639, -...   \n",
       "\n",
       "                                            PM10_Counts  Air_is_bad?  \n",
       "0     [-0.32022983414764833, -0.15608473051877228, -...            1  \n",
       "1     [-0.15608473051877228, 0.19956299401045927, 0....            1  \n",
       "2     [0.500495683996732, 1.1570760985122364, 1.4032...            1  \n",
       "3     [0.500495683996732, 0.9929309948833603, 1.4032...            1  \n",
       "4     [0.36370809763933537, 1.512723823041468, 1.649...            1  \n",
       "...                                                 ...          ...  \n",
       "5106  [-0.21079976506173095, -0.7853076277627973, -0...            0  \n",
       "5107  [-0.8126651450342767, -0.9494527313916734, -0....            0  \n",
       "5108  [-0.8673801795772353, -0.48437493777652446, -0...            1  \n",
       "5109  [-0.18344224779025162, 0.06277540765306251, 0....            1  \n",
       "5110  [0.14484795946750056, 0.36370809763933537, 0.5...            0  \n",
       "\n",
       "[5111 rows x 10 columns]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "0ba34e05-8d64-43a0-9a7b-86dea14e925c",
   "metadata": {},
   "outputs": [],
   "source": [
    "data.to_csv(\"../data/fine_dust_dataset.csv\", index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "193919cb-7ded-4968-8b45-26e2a7f5965e",
   "metadata": {},
   "source": [
    "# 2. DataLoader Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e3ca19a2-3c80-4d1c-8796-b4f641c0e6ad",
   "metadata": {},
   "source": [
    "## 2.1. Preparation of X in `numpy.array` format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "24bae604-4d51-48f8-a7b3-8136336f0efe",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 68%|████████████████████████████████████████████████████                        | 3500/5111 [00:00<00:00, 5793.36it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2791 1 23\n",
      "2791 2 23\n",
      "2791 3 23\n",
      "2791 4 23\n",
      "2791 5 23\n",
      "2791 6 23\n",
      "2801 1 23\n",
      "2801 2 23\n",
      "2801 3 23\n",
      "2801 4 23\n",
      "2801 5 23\n",
      "2801 6 23\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████| 5111/5111 [00:00<00:00, 5755.63it/s]\n"
     ]
    }
   ],
   "source": [
    "# To filter out data with inconsistent length\n",
    "\n",
    "for i in tqdm(range(len(data))):\n",
    "    for j in range(1, 7):\n",
    "        if len(data.iloc[i, j]) != 24:\n",
    "            print(i, j, len(data.iloc[i, j]))\n",
    "        else:\n",
    "            continue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "a69b2cff-6d0c-4f1a-842a-332d6d4be4a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data.drop([2791, 2801], axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "8b2a5ea0-544e-4387-b1c9-a96ddd932317",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████| 5109/5109 [00:01<00:00, 3577.53it/s]\n"
     ]
    }
   ],
   "source": [
    "feature_array = []\n",
    "for i in tqdm(range(len(data))):\n",
    "    row_array = []\n",
    "    for j in range(1, 9):\n",
    "        row_array.append(np.array(data.iloc[i, j]))\n",
    "    stacked_row_array = np.stack(row_array, axis=0)\n",
    "    feature_array.append(stacked_row_array)\n",
    "\n",
    "X = np.stack(feature_array, axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "232a79aa-8824-43c2-b9fb-937b02dada9e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5109, 8, 24)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bf3a9165-24e4-4c51-bb7f-5ae32fe4b461",
   "metadata": {},
   "source": [
    "## 2.2. Preparation of y in `numpy.array` format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "626cbae9-078c-4f09-9663-82e331283a65",
   "metadata": {},
   "outputs": [],
   "source": [
    "y = np.array(data[\"Air_is_bad?\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "a0fc6944-5040-4fe5-82f0-15391b34ddf9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5109,)"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e54ad713-103a-4ee6-89c6-7f52fffdc664",
   "metadata": {},
   "source": [
    "## 2.3. Train/Valid/Test Split for Deep Learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "f1423643-9a9d-4180-9cec-36f07a77ad92",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2919"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# To find the index to split the data into train and test set\n",
    "len(data.loc[data[\"date_new\"].dt.date < datetime.date(2016, 1, 1)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "7deea8fa-d60e-4ce9-b754-480151a85363",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3650"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# To find the index to split the test data into valid and test set\n",
    "len(data.loc[data[\"date_new\"].dt.date < datetime.date(2018, 1, 1)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "f9ac3ec1-e6fa-4c47-9cc0-ff506d9113ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = X[:2919, :, :]\n",
    "X_valid = X[2919:3650, :, :]\n",
    "X_test = X[3650:, :, :]\n",
    "\n",
    "y_train = y[:2919]\n",
    "y_valid = y[2919:3650]\n",
    "y_test = y[3650:]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bdf15e47-5983-461d-9713-243d634a3414",
   "metadata": {},
   "source": [
    "## 2.4. DataLoaders"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "5c839db3-0129-40d5-84bb-85c2766bbac6",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_tensor = torch.Tensor(X_train)\n",
    "X_valid_tensor = torch.Tensor(X_valid)\n",
    "X_test_tensor = torch.Tensor(X_test)\n",
    "\n",
    "y_train_tensor = torch.Tensor(y_train)\n",
    "y_valid_tensor = torch.Tensor(y_valid)\n",
    "y_test_tensor = torch.Tensor(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "570f1bf9-5a7b-4528-b9d7-6d704f4123c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset = TensorDataset(X_train_tensor, y_train_tensor)\n",
    "trainloader = DataLoader(train_dataset, batch_size=28, shuffle=True)\n",
    "\n",
    "valid_dataset = TensorDataset(X_valid_tensor, y_valid_tensor)\n",
    "validloader = DataLoader(valid_dataset, batch_size=10, shuffle=True)\n",
    "\n",
    "test_dataset = TensorDataset(X_test_tensor, y_test_tensor)\n",
    "testloader = DataLoader(test_dataset, batch_size=10, shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74ffb75c-ce6f-48bb-8b7c-d6b3cf4e784f",
   "metadata": {},
   "source": [
    "## 2.5. Sample Check"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "3b8afae1-086b-414c-9619-5f4e8cb100a7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 900x400 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "weather, label = next(iter(trainloader))\n",
    "i = np.random.randint(0, 28)\n",
    "weather = weather[i]\n",
    "weather = scaler.inverse_transform(weather.permute(1, 0)).transpose()\n",
    "fig, axs = plt.subplots(nrows=2, ncols=4, figsize=(9, 4))\n",
    "r = 0\n",
    "c = 0\n",
    "for j in range(8):\n",
    "    axs[r, c].plot(weather[j, :])\n",
    "    axs[r, c].set_title(f\"{feature_names[j]}\")\n",
    "    if c < 3:\n",
    "        c += 1\n",
    "    else:\n",
    "        r += 1\n",
    "        c = 0\n",
    "fig.suptitle(f\"Air Quality Bad? {bool(label[i])}\")\n",
    "plt.tight_layout()\n",
    "plt.show();\n",
    "fig.savefig(\"../img/features_for_DL.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c771c587-8c40-43b6-b19c-86f697534bbb",
   "metadata": {},
   "source": [
    "As mentioned earlier, the 24-hour data of 8 features shown above will used to predict the air quality of the next day using multi-scale 1-d CNN network."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c50261fe-4e6a-45d3-949c-543d1970142e",
   "metadata": {},
   "source": [
    "## 2.6. DataLoader Save"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "a5d8e952-0328-4735-b254-eeedee77dd73",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(trainloader, \"../models/trainloader.pth\")\n",
    "torch.save(validloader, \"../models/validloader.pth\")\n",
    "torch.save(testloader, \"../models/testloader.pth\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:finedust_env]",
   "language": "python",
   "name": "conda-env-finedust_env-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}