preprocessing.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ff5d234c",
   "metadata": {},
   "source": [
    "# Urban Sounds Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "ebf9bb93",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Dataset: https://urbansounddataset.weebly.com/urbansound8k.html"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "36e28f0b",
   "metadata": {},
   "source": [
    "### 1. Spectrograms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6877c2b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Spectrograms: https://drive.google.com/drive/folders/1xey3vAVNDjWxnSfhuUsf_5dGANZWmCog?usp=sharing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70f060e1",
   "metadata": {},
   "source": [
    "### 2. Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "07b1e5fb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Packages\n",
    "\n",
    "import librosa\n",
    "import cv2\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import os \n",
    "import librosa.display\n",
    "from PIL import Image\n",
    "import PIL\n",
    "import glob\n",
    "import random\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "19b98d05",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read the dataset file\n",
    "\n",
    "metadata_path = r\"C:\\Users\\pc\\Desktop\\UrbanSound8K\\metadata\\UrbanSound8K.csv\"\n",
    "metadata = pd.read_csv(metadata_path)\n",
    "df = metadata.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "7b7195e6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The shape of the dataset is:  (8732, 8)\n"
     ]
    }
   ],
   "source": [
    "#The shape of the data\n",
    "\n",
    "print(\"The shape of the dataset is: \", df.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "f77bb2d8",
   "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>slice_file_name</th>\n",
       "      <th>fsID</th>\n",
       "      <th>start</th>\n",
       "      <th>end</th>\n",
       "      <th>salience</th>\n",
       "      <th>fold</th>\n",
       "      <th>classID</th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>100032-3-0-0.wav</td>\n",
       "      <td>100032</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.317551</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>dog_bark</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100263-2-0-117.wav</td>\n",
       "      <td>100263</td>\n",
       "      <td>58.5</td>\n",
       "      <td>62.500000</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>children_playing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>100263-2-0-121.wav</td>\n",
       "      <td>100263</td>\n",
       "      <td>60.5</td>\n",
       "      <td>64.500000</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>children_playing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>100263-2-0-126.wav</td>\n",
       "      <td>100263</td>\n",
       "      <td>63.0</td>\n",
       "      <td>67.000000</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>children_playing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>100263-2-0-137.wav</td>\n",
       "      <td>100263</td>\n",
       "      <td>68.5</td>\n",
       "      <td>72.500000</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>children_playing</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      slice_file_name    fsID  start        end  salience  fold  classID  \\\n",
       "0    100032-3-0-0.wav  100032    0.0   0.317551         1     5        3   \n",
       "1  100263-2-0-117.wav  100263   58.5  62.500000         1     5        2   \n",
       "2  100263-2-0-121.wav  100263   60.5  64.500000         1     5        2   \n",
       "3  100263-2-0-126.wav  100263   63.0  67.000000         1     5        2   \n",
       "4  100263-2-0-137.wav  100263   68.5  72.500000         1     5        2   \n",
       "\n",
       "              class  \n",
       "0          dog_bark  \n",
       "1  children_playing  \n",
       "2  children_playing  \n",
       "3  children_playing  \n",
       "4  children_playing  "
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Displaying the first 5 rows of the data\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "d61244b1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "slice_file_name    8732\n",
       "fsID               1297\n",
       "start              4878\n",
       "end                5020\n",
       "salience              2\n",
       "fold                 10\n",
       "classID              10\n",
       "class                10\n",
       "dtype: int64"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Number of unique values in each column\n",
    "\n",
    "df.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "710cd9ae",
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>fsID</th>\n",
       "      <td>8732.0</td>\n",
       "      <td>116033.493816</td>\n",
       "      <td>57991.017218</td>\n",
       "      <td>344.000000</td>\n",
       "      <td>69942.250000</td>\n",
       "      <td>118279.000000</td>\n",
       "      <td>166942.000000</td>\n",
       "      <td>209992.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>start</th>\n",
       "      <td>8732.0</td>\n",
       "      <td>38.645409</td>\n",
       "      <td>74.292126</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>10.376492</td>\n",
       "      <td>35.131372</td>\n",
       "      <td>600.125356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>end</th>\n",
       "      <td>8732.0</td>\n",
       "      <td>42.253312</td>\n",
       "      <td>74.369669</td>\n",
       "      <td>0.105962</td>\n",
       "      <td>6.839398</td>\n",
       "      <td>14.000000</td>\n",
       "      <td>38.866979</td>\n",
       "      <td>604.125356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>salience</th>\n",
       "      <td>8732.0</td>\n",
       "      <td>1.347000</td>\n",
       "      <td>0.476043</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>2.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fold</th>\n",
       "      <td>8732.0</td>\n",
       "      <td>5.385937</td>\n",
       "      <td>2.846820</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>10.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>classID</th>\n",
       "      <td>8732.0</td>\n",
       "      <td>4.592877</td>\n",
       "      <td>2.894544</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>9.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           count           mean           std         min           25%  \\\n",
       "fsID      8732.0  116033.493816  57991.017218  344.000000  69942.250000   \n",
       "start     8732.0      38.645409     74.292126    0.000000      3.000000   \n",
       "end       8732.0      42.253312     74.369669    0.105962      6.839398   \n",
       "salience  8732.0       1.347000      0.476043    1.000000      1.000000   \n",
       "fold      8732.0       5.385937      2.846820    1.000000      3.000000   \n",
       "classID   8732.0       4.592877      2.894544    0.000000      2.000000   \n",
       "\n",
       "                    50%            75%            max  \n",
       "fsID      118279.000000  166942.000000  209992.000000  \n",
       "start         10.376492      35.131372     600.125356  \n",
       "end           14.000000      38.866979     604.125356  \n",
       "salience       1.000000       2.000000       2.000000  \n",
       "fold           5.000000       8.000000      10.000000  \n",
       "classID        4.000000       7.000000       9.000000  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Summary\n",
    "\n",
    "df.describe().transpose()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "36cd232b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dog_bark            1000\n",
       "children_playing    1000\n",
       "air_conditioner     1000\n",
       "street_music        1000\n",
       "engine_idling       1000\n",
       "jackhammer          1000\n",
       "drilling            1000\n",
       "siren                929\n",
       "car_horn             429\n",
       "gun_shot             374\n",
       "Name: class, dtype: int64"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[\"class\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "c00c6cee",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize the target column\n",
    "\n",
    "plt.figure(figsize = (8,4))\n",
    "sns.barplot(x = df[\"class\"].value_counts().keys(),\n",
    "            y = df[\"class\"].value_counts(),\n",
    "            palette=\"magma\")\n",
    "plt.grid(color='#95a5a6', linestyle='-.', linewidth=1, axis='y', alpha=0.7)\n",
    "plt.xlabel(\"Class Names\", weight=\"bold\", color=\"#355764\", fontsize=14, labelpad = 20)\n",
    "plt.ylabel(\"Number of the samples\", weight=\"bold\", color=\"#355764\", fontsize=14, labelpad=20)\n",
    "plt.xticks(rotation=90, fontsize=12)\n",
    "plt.yticks(rotation=90, fontsize=12)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "508475a1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating spectrogram file path (fold + file name)\n",
    "\n",
    "df[\"specs_path\"] = '\\\\' + df['classID'].astype(str) + '\\\\' + df['slice_file_name'].astype(str).str.strip('.wav') + '.png'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "c252a957",
   "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>specs_path</th>\n",
       "      <th>classID</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>\\3\\100032-3-0-0.png</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>\\2\\100263-2-0-117.png</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>\\2\\100263-2-0-121.png</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>\\2\\100263-2-0-126.png</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>\\2\\100263-2-0-137.png</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              specs_path  classID\n",
       "0    \\3\\100032-3-0-0.png        3\n",
       "1  \\2\\100263-2-0-117.png        2\n",
       "2  \\2\\100263-2-0-121.png        2\n",
       "3  \\2\\100263-2-0-126.png        2\n",
       "4  \\2\\100263-2-0-137.png        2"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Update the dataset \n",
    "\n",
    "df = df[[\"specs_path\",\"classID\"]]\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "613286d9",
   "metadata": {},
   "source": [
    "##### 2.1 Grayscale and Resize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "ba46fea7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# The funtion rgb to grayscale\n",
    "\n",
    "def rgb_to_grayscale(loc, path):\n",
    "    \n",
    "    # grayscale\n",
    "    gray_images = cv2.imread(loc+path, 0) \n",
    "    \n",
    "    # resize\n",
    "    images = cv2.resize(gray_images, (64,64)) \n",
    "        \n",
    "    return images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "0c298360",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Loop to save all specs grayscale \n",
    "\n",
    "for spec in df[\"specs_path\"]:\n",
    "    loc = r'C:\\Users\\pc\\Desktop\\spectrograms' # location\n",
    "    path = spec # image path\n",
    "    gray_pic = rgb_to_grayscale(loc, path)\n",
    "    cv2.imwrite(loc+path, gray_pic)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a1d2f9d8",
   "metadata": {},
   "source": [
    "##### 2.2 Normalization "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "ff447231",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating a list [image,label]\n",
    "\n",
    "img_arr = []\n",
    "\n",
    "for i in range(len(df[\"specs_path\"])):\n",
    "    loc = r'C:\\Users\\pc\\Desktop\\spectrograms' # location\n",
    "    path = df[\"specs_path\"][i] # image path\n",
    "    gray_imgs = cv2.imread(loc+path, 0)\n",
    "    img_arr.append([gray_imgs, df[\"classID\"][i]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "737d047c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Shuffle the data\n",
    "\n",
    "random.shuffle(img_arr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "0eeefc35",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating a dataframe \n",
    "\n",
    "img_df = pd.DataFrame(img_arr, columns=[\"images\",\"classID\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "03a8ea12",
   "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>images</th>\n",
       "      <th>classID</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>[[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              images  classID\n",
       "0  [[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...        3\n",
       "1  [[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...        0\n",
       "2  [[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...        4\n",
       "3  [[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...        5\n",
       "4  [[249, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...        7"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "img_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "71641d48",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64, 64)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Shape of the sample image\n",
    "\n",
    "img_df[\"images\"][0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "6ef63366",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating the X, y\n",
    "\n",
    "X = np.array(img_df[\"images\"].tolist())\n",
    "y = np.array(img_df[\"classID\"].tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "d0246f20",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8732, 64, 64)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Shape of the X (Features)\n",
    "\n",
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "815bdb75",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8732,)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Shape of the y (Target)\n",
    "\n",
    "y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "e68ead51",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Normalization\n",
    "# we can divide X by 255\n",
    "\n",
    "X = X/255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "53bdfcd3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.97647059, 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.        ],\n",
       "       [0.97647059, 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.        ],\n",
       "       [0.97647059, 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.        ],\n",
       "       ...,\n",
       "       [0.97647059, 0.        , 0.        , ..., 0.        , 0.        ,\n",
       "        0.        ],\n",
       "       [0.97647059, 0.        , 0.        , ..., 0.01568627, 0.        ,\n",
       "        0.        ],\n",
       "       [0.99607843, 0.97647059, 0.97647059, ..., 0.97647059, 0.97647059,\n",
       "        0.97647059]])"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X[0] # Now each values scaled between 0 and 1."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d0198cc",
   "metadata": {},
   "source": [
    "##### 2.3 Dataset Splitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "7d42950a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Length of X:  8732\n",
      "Length of y:  8732\n"
     ]
    }
   ],
   "source": [
    "print(\"Length of X: \",len(X))\n",
    "print(\"Length of y: \",len(y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "5a2c0973",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Roundly %80 train sets, %10 validation sets, %10 test sets\n",
    "\n",
    "# Train datasets\n",
    "X_train = X[:6980]\n",
    "y_train = y[:6980]\n",
    "\n",
    "# Validation datasets\n",
    "X_val = X[6980:7856]\n",
    "y_val = y[6980:7856]\n",
    "\n",
    "# Test datasets\n",
    "X_test = X[7856:8732]\n",
    "y_test = y[7856:8732]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "8ac19210",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train: 6980\n",
      "y_train: 6980\n",
      "X_val: 876\n",
      "y_val: 876\n",
      "X_test: 876\n",
      "y_test: 876\n"
     ]
    }
   ],
   "source": [
    "# The number of samples\n",
    "\n",
    "print(f\"X_train: {len(X_train)}\")\n",
    "print(f\"y_train: {len(y_train)}\")\n",
    "print(f\"X_val: {len(X_val)}\")\n",
    "print(f\"y_val: {len(y_val)}\")\n",
    "print(f\"X_test: {len(X_test)}\")\n",
    "print(f\"y_test: {len(y_test)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "36056c14",
   "metadata": {},
   "source": [
    "##### 2.4 Save Datasets to Computer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "cd79fdc6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Pickle library\n",
    "\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "e06a5560",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the datasets to computer\n",
    "\n",
    "datasets = [X_train, y_train, X_val, y_val, X_test, y_test]\n",
    "dataset_names = ['X_train', 'y_train', 'X_val', 'y_val', 'X_test', 'y_test']\n",
    "\n",
    "for i in range(len(datasets)):\n",
    "    with open(r'C:\\Users\\pc\\Desktop\\datasets' + '\\\\' + dataset_names[i] + '.pkl', 'wb') as f:\n",
    "        pickle.dump(datasets[i], f)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}