breath_sound_pilot_model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "2f9ea37f",
   "metadata": {},
   "source": [
    "## Pilot Model Introduction"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ea0e38c3",
   "metadata": {},
   "source": [
    "의사가 청진기를 통해 환자의 호흡음을 평가하는 것은 질병의 진단, 치료에 있어 중요한 작업이다. 현재는 의사1명이 청진기를 통해 호흡음을 듣고 질병의 유무를 판단하지만 이는 정확성, 객관성, 일관성에 있어서 문제가 있을 수 있다. 그러므로 청진기를 통한 호흡음을 듣고 정상, 비정상의 호흡음을 분류할 수 있는 좋은 성능의 딥러닝 모델을 만드는 것이 최종적인 목표이다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "65472090",
   "metadata": {},
   "source": [
    "정상적인 호흡음과 비정상적인 호흡음은 특정한 패턴으로 분류한다. 먼저 정상적인 호흡음과  천명(Wheezing)이라고 불리는 패턴을 가지는 호흡음(비정상적인)을 분류하는(Binary Classification) 모델을 구축하고자 한다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fcdb1ac3",
   "metadata": {},
   "source": [
    "병원 측에서 제공받은 데이터 중 우선적으로 Pilot Model에서 사용할 데이터는 아래와 같다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85ff1f31",
   "metadata": {},
   "source": [
    "정상 호흡음을 정상 호흡음이라고 판단한 Audio Data(91개)\n",
    "\n",
    "Wheezing이라고 판단했지만 정상 호흡음 이었던 Audio Data(24개)\n",
    "\n",
    "정상 호흡음이지만 잡음이 많이 들어간 Audio Data(23개)\n",
    "#### --> 총 178개의 Audio Data = Healthy\n",
    "\n",
    "Wheezing을 Wheezing이라고 판단한 Audio Data(74개)\n",
    "\n",
    "Wheezing을 Wheezing이라고 판단했지만 Wheezing구간을 정확히 맞추지 못한 Audio Data(24개)\n",
    "#### --> 총 85개의 Audio Data = Wheezing\n",
    "\n",
    "현재는 데이터의 수가 굉장히 적기 때문에 쓸 수 있는 데이터는 모두 사용하기로 했다. Healthy와 Wheezing Data간 불균형이 보이므로 추후에는 SMOTE(Syntetic Minority Oversampling Technique)기법으로 불균형을 해소할 예정이다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "1ebb028a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import librosa\n",
    "import pickle\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from tqdm import tqdm\n",
    "from python_speech_features import mfcc\n",
    "from sklearn.preprocessing import LabelEncoder, OneHotEncoder\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from keras.layers import Conv2D, MaxPool2D, Flatten, Dense, Dropout\n",
    "from keras.models import Sequential"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "55e1ab5b",
   "metadata": {},
   "source": [
    "### Data Collection"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a9c7b53",
   "metadata": {},
   "source": [
    "위에서 언급한 것처럼 Binary Classification의 간단한 Task다. fname에 File Name, label에 호흡음 정상여부(Wheezing, healthy)가 기록되어 있다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "74b3cf7e",
   "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>fname</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0001-1.wav</td>\n",
       "      <td>wheezing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0411-2.wav</td>\n",
       "      <td>wheezing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0538-4.wav</td>\n",
       "      <td>wheezing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0581-2.wav</td>\n",
       "      <td>wheezing</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0597-4.wav</td>\n",
       "      <td>wheezing</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        fname     label\n",
       "0  0001-1.wav  wheezing\n",
       "1  0411-2.wav  wheezing\n",
       "2  0538-4.wav  wheezing\n",
       "3  0581-2.wav  wheezing\n",
       "4  0597-4.wav  wheezing"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('data/data1.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "09942010",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df[df['label'].isin(['wheezing','healthy'])]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e99e023e",
   "metadata": {},
   "source": [
    "Numpy array로 바꿔주자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8477e0bb",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 263/263 [00:59<00:00,  4.42it/s]\n",
      "<ipython-input-4-bd871a4eab45>:5: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray\n",
      "  audio_data = np.array(audio_data)\n"
     ]
    }
   ],
   "source": [
    "path = 'data/all_data/'\n",
    "audio_data = list()\n",
    "for i in tqdm(range(df.shape[0])):\n",
    "    audio_data.append(librosa.load(path+df['fname'].iloc[i]))\n",
    "audio_data = np.array(audio_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "60e7a8ff",
   "metadata": {},
   "source": [
    "확인해보자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1ac7a2ee",
   "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>fname</th>\n",
       "      <th>label</th>\n",
       "      <th>audio_waves</th>\n",
       "      <th>samplerate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0001-1.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[-1.960364e-06, -1.6989543e-06, 2.2033912e-06,...</td>\n",
       "      <td>22050</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0411-2.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[0.0079015065, 0.010698691, 0.0077624833, 0.00...</td>\n",
       "      <td>22050</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0538-4.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[0.017012855, 0.026953325, 0.025314119, 0.0284...</td>\n",
       "      <td>22050</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0581-2.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[0.020541016, 0.030520504, 0.02603743, 0.02776...</td>\n",
       "      <td>22050</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0597-4.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[-0.013913074, -0.0219503, -0.020427886, -0.02...</td>\n",
       "      <td>22050</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        fname     label                                        audio_waves  \\\n",
       "0  0001-1.wav  wheezing  [-1.960364e-06, -1.6989543e-06, 2.2033912e-06,...   \n",
       "1  0411-2.wav  wheezing  [0.0079015065, 0.010698691, 0.0077624833, 0.00...   \n",
       "2  0538-4.wav  wheezing  [0.017012855, 0.026953325, 0.025314119, 0.0284...   \n",
       "3  0581-2.wav  wheezing  [0.020541016, 0.030520504, 0.02603743, 0.02776...   \n",
       "4  0597-4.wav  wheezing  [-0.013913074, -0.0219503, -0.020427886, -0.02...   \n",
       "\n",
       "  samplerate  \n",
       "0      22050  \n",
       "1      22050  \n",
       "2      22050  \n",
       "3      22050  \n",
       "4      22050  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['audio_waves'] = audio_data[:,0]\n",
    "df['samplerate'] = audio_data[:,1]\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe9763ef",
   "metadata": {},
   "source": [
    "bit_lengths와 second_lenghts를 생성한다.\n",
    "\n",
    "#### bit_lengths = The number of bits in each audio waves.\n",
    "\n",
    "#### second_lenths = The length of all audio files in seconds.(bit_lenths/samplerate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "5e145baa",
   "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>fname</th>\n",
       "      <th>label</th>\n",
       "      <th>audio_waves</th>\n",
       "      <th>samplerate</th>\n",
       "      <th>bit_lengths</th>\n",
       "      <th>second_lengths</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0001-1.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[-1.960364e-06, -1.6989543e-06, 2.2033912e-06,...</td>\n",
       "      <td>22050</td>\n",
       "      <td>198648</td>\n",
       "      <td>9.00898</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0411-2.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[0.0079015065, 0.010698691, 0.0077624833, 0.00...</td>\n",
       "      <td>22050</td>\n",
       "      <td>158919</td>\n",
       "      <td>7.207211</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0538-4.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[0.017012855, 0.026953325, 0.025314119, 0.0284...</td>\n",
       "      <td>22050</td>\n",
       "      <td>129490</td>\n",
       "      <td>5.872562</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0581-2.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[0.020541016, 0.030520504, 0.02603743, 0.02776...</td>\n",
       "      <td>22050</td>\n",
       "      <td>164805</td>\n",
       "      <td>7.47415</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0597-4.wav</td>\n",
       "      <td>wheezing</td>\n",
       "      <td>[-0.013913074, -0.0219503, -0.020427886, -0.02...</td>\n",
       "      <td>22050</td>\n",
       "      <td>444384</td>\n",
       "      <td>20.153469</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        fname     label                                        audio_waves  \\\n",
       "0  0001-1.wav  wheezing  [-1.960364e-06, -1.6989543e-06, 2.2033912e-06,...   \n",
       "1  0411-2.wav  wheezing  [0.0079015065, 0.010698691, 0.0077624833, 0.00...   \n",
       "2  0538-4.wav  wheezing  [0.017012855, 0.026953325, 0.025314119, 0.0284...   \n",
       "3  0581-2.wav  wheezing  [0.020541016, 0.030520504, 0.02603743, 0.02776...   \n",
       "4  0597-4.wav  wheezing  [-0.013913074, -0.0219503, -0.020427886, -0.02...   \n",
       "\n",
       "  samplerate  bit_lengths second_lengths  \n",
       "0      22050       198648        9.00898  \n",
       "1      22050       158919       7.207211  \n",
       "2      22050       129490       5.872562  \n",
       "3      22050       164805        7.47415  \n",
       "4      22050       444384      20.153469  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bit_lengths = list()\n",
    "for i in range(df.shape[0]):\n",
    "    bit_lengths.append(len(df['audio_waves'].iloc[i]))\n",
    "bit_lengths = np.array(bit_lengths)\n",
    "df['bit_lengths'] = bit_lengths\n",
    "df['second_lengths'] = df['bit_lengths']/df['samplerate']\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fb92e743",
   "metadata": {},
   "source": [
    "### Data Generate"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "02e5ac3e",
   "metadata": {},
   "source": [
    "2 seconds 보다 짝으면 drop."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "4e49f12d",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df[df['second_lengths'] >= 2.0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5c8e1856",
   "metadata": {},
   "source": [
    "min값을 출력해 체크해보자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "fcaf909b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "48113\n",
      "2.1819954648526076\n"
     ]
    }
   ],
   "source": [
    "min_bits = np.min(df['bit_lengths'])\n",
    "print(min_bits)\n",
    "min_seconds = np.min(df['second_lengths'])\n",
    "print(min_seconds)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "38c5f86b",
   "metadata": {},
   "source": [
    "데이터 프레임을 pickle로 저장하자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "4dc8664e",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('audio_df.pickle', 'wb') as f:\n",
    "    pickle.dump(df, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "cf9e45f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('audio_df.pickle', 'rb') as f:\n",
    "    df = pickle.load(f)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b7b6e8d6",
   "metadata": {},
   "source": [
    "아래 그림처럼 Random Sampling을 통해 data generate"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7f779c8e",
   "metadata": {},
   "source": [
    "![1_SQr0VgPZF2pvzVVpYiH-PQ](https://user-images.githubusercontent.com/74411831/122634903-e4ed2100-d11b-11eb-9ede-ee6eb18fbae3.jpeg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "75805f0d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 6000/6000 [00:05<00:00, 1022.73it/s]\n"
     ]
    }
   ],
   "source": [
    "num_samples = 6000\n",
    "generated_audio_waves = list()\n",
    "generated_audio_labels = list()\n",
    "for i in tqdm(range(num_samples)):\n",
    "    try:\n",
    "        chosen_file = np.random.choice(df['fname'].values)\n",
    "        chosen_initial = np.random.choice(np.arange(0,df[df['fname']==chosen_file]['bit_lengths'].values-min_bits))\n",
    "        generated_audio_waves.append(df[df['fname']==chosen_file]['audio_waves'].values[0][chosen_initial:chosen_initial+min_bits])\n",
    "        \n",
    "        generated_audio_labels.append(df[df['fname']==chosen_file]['label'].values)\n",
    "    except ValueError:\n",
    "        continue\n",
    "generated_audio_waves = np.array(generated_audio_waves)\n",
    "generated_audio_labels = np.array(generated_audio_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1dbaf966",
   "metadata": {},
   "source": [
    "MFCC로 Feature를 뽑는다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "0ae55563",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 5974/5974 [00:40<00:00, 147.47it/s]\n"
     ]
    }
   ],
   "source": [
    "mfcc_features = list()\n",
    "for i in tqdm(range(len(generated_audio_waves))):\n",
    "    mfcc_features.append(mfcc(generated_audio_waves[i]))\n",
    "mfcc_features = np.array(mfcc_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "09ef8808",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5974, 48113)\n",
      "(5974, 300, 13)\n"
     ]
    }
   ],
   "source": [
    "print(generated_audio_waves.shape)\n",
    "print(mfcc_features.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "950a484e",
   "metadata": {},
   "source": [
    "Audio Data와 MFCC를 비교. 잘 뽑혔는지 확인하자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "d006c53c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x144 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,2))\n",
    "plt.plot(generated_audio_waves[30])\n",
    "plt.title(generated_audio_labels[30])\n",
    "plt.show()\n",
    "plt.figure(figsize=(12, 2))\n",
    "plt.imshow(mfcc_features[30].T, cmap='hot')\n",
    "plt.title(generated_audio_labels[30])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e1acc406",
   "metadata": {},
   "source": [
    "### Pre-processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "507eec0d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1 0 1 ... 0 1 0]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/nextgen/anaconda3/envs/kbs_coin/lib/python3.8/site-packages/sklearn/utils/validation.py:63: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  return f(*args, **kwargs)\n"
     ]
    }
   ],
   "source": [
    "label_encoder = LabelEncoder()\n",
    "label_encoded = label_encoder.fit_transform(generated_audio_labels)\n",
    "print(label_encoded)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "c4a221b4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1],\n",
       "       [0],\n",
       "       [1],\n",
       "       ...,\n",
       "       [0],\n",
       "       [1],\n",
       "       [0]])"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "label_encoded = label_encoded[:, np.newaxis]\n",
    "label_encoded"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "aaefa9f0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0., 1.],\n",
       "       [1., 0.],\n",
       "       [0., 1.],\n",
       "       ...,\n",
       "       [1., 0.],\n",
       "       [0., 1.],\n",
       "       [1., 0.]])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "one_hot_encoder = OneHotEncoder(sparse=False)\n",
    "one_hot_encoded = one_hot_encoder.fit_transform(label_encoded)\n",
    "one_hot_encoded"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c797a90",
   "metadata": {},
   "source": [
    "### Training(CNN Model)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f37c9f58",
   "metadata": {},
   "source": [
    "MinMax Normalization 및 split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "4f8c579e",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = mfcc_features\n",
    "y = one_hot_encoded\n",
    "X = (X-X.min())/(X.max()-X.min())\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "e36f4ac0",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_shape = (X_train.shape[1], X_train.shape[2], 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5c5856cb",
   "metadata": {},
   "source": [
    "4777개의 training sample, 1195개의 test sample을 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "e76067e0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4779, 300, 13, 1)\n",
      "(1195, 300, 13, 1)\n"
     ]
    }
   ],
   "source": [
    "X_train = X_train.reshape(X_train.shape[0], X_train.shape[1], X_train.shape[2], 1)\n",
    "print(X_train.shape)\n",
    "X_test = X_test.reshape(X_test.shape[0], X_test.shape[1], X_test.shape[2], 1)\n",
    "print(X_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "3df9ea80",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Conv2D(16, (3, 3), activation='relu', strides=(1, 1), \n",
    "    padding='same', input_shape=input_shape))\n",
    "model.add(Conv2D(32, (3, 3), activation='relu', strides=(1, 1), \n",
    "    padding='same'))\n",
    "model.add(MaxPool2D((2, 2)))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(128, activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(64, activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(2, activation='softmax'))\n",
    "\n",
    "model.compile(loss='binary_crossentropy', \n",
    "     optimizer='adam',\n",
    "     metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "dff99134",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 300, 13, 16)       160       \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 300, 13, 32)       4640      \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 150, 6, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 150, 6, 32)        0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 28800)             0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 128)               3686528   \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 2)                 130       \n",
      "=================================================================\n",
      "Total params: 3,699,714\n",
      "Trainable params: 3,699,714\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "c8ec489b",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "150/150 [==============================] - 17s 48ms/step - loss: 0.5817 - acc: 0.6911 - val_loss: 0.3346 - val_acc: 0.9071\n",
      "Epoch 2/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.3825 - acc: 0.8848 - val_loss: 0.2797 - val_acc: 0.9155\n",
      "Epoch 3/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.3369 - acc: 0.8939 - val_loss: 0.2697 - val_acc: 0.9138\n",
      "Epoch 4/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2969 - acc: 0.9022 - val_loss: 0.2639 - val_acc: 0.9155\n",
      "Epoch 5/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2935 - acc: 0.9064 - val_loss: 0.2681 - val_acc: 0.9188\n",
      "Epoch 6/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2951 - acc: 0.9021 - val_loss: 0.2594 - val_acc: 0.9180\n",
      "Epoch 7/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2661 - acc: 0.9111 - val_loss: 0.2605 - val_acc: 0.9180\n",
      "Epoch 8/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2635 - acc: 0.9130 - val_loss: 0.2559 - val_acc: 0.9172\n",
      "Epoch 9/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2583 - acc: 0.9099 - val_loss: 0.2477 - val_acc: 0.9172\n",
      "Epoch 10/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2480 - acc: 0.9102 - val_loss: 0.2544 - val_acc: 0.9172\n",
      "Epoch 11/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2431 - acc: 0.9135 - val_loss: 0.2552 - val_acc: 0.9130\n",
      "Epoch 12/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2250 - acc: 0.9228 - val_loss: 0.2625 - val_acc: 0.9096\n",
      "Epoch 13/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2253 - acc: 0.9195 - val_loss: 0.2470 - val_acc: 0.9163\n",
      "Epoch 14/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2266 - acc: 0.9210 - val_loss: 0.2536 - val_acc: 0.9213\n",
      "Epoch 15/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.2103 - acc: 0.9210 - val_loss: 0.2536 - val_acc: 0.9205\n",
      "Epoch 16/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1981 - acc: 0.9266 - val_loss: 0.2623 - val_acc: 0.9172\n",
      "Epoch 17/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1747 - acc: 0.9353 - val_loss: 0.2311 - val_acc: 0.9180\n",
      "Epoch 18/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1758 - acc: 0.9350 - val_loss: 0.2539 - val_acc: 0.9247\n",
      "Epoch 19/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1784 - acc: 0.9344 - val_loss: 0.2866 - val_acc: 0.9205\n",
      "Epoch 20/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1626 - acc: 0.9380 - val_loss: 0.2790 - val_acc: 0.9046\n",
      "Epoch 21/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1699 - acc: 0.9337 - val_loss: 0.2659 - val_acc: 0.9213\n",
      "Epoch 22/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1495 - acc: 0.9440 - val_loss: 0.2520 - val_acc: 0.9230\n",
      "Epoch 23/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1573 - acc: 0.9375 - val_loss: 0.2515 - val_acc: 0.9238\n",
      "Epoch 24/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1729 - acc: 0.9345 - val_loss: 0.2595 - val_acc: 0.9264\n",
      "Epoch 25/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1485 - acc: 0.9411 - val_loss: 0.2801 - val_acc: 0.9289\n",
      "Epoch 26/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1346 - acc: 0.9485 - val_loss: 0.2561 - val_acc: 0.9280\n",
      "Epoch 27/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1365 - acc: 0.9502 - val_loss: 0.2639 - val_acc: 0.9322\n",
      "Epoch 28/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1375 - acc: 0.9478 - val_loss: 0.2480 - val_acc: 0.9314\n",
      "Epoch 29/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1277 - acc: 0.9524 - val_loss: 0.2608 - val_acc: 0.9222\n",
      "Epoch 30/30\n",
      "150/150 [==============================] - 7s 47ms/step - loss: 0.1285 - acc: 0.9509 - val_loss: 0.3592 - val_acc: 0.9305\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train, y_train, epochs=30, validation_data=(X_test, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a27ec02",
   "metadata": {},
   "source": [
    "### Visualize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "3d7f20c7",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.1232549250125885\n",
      "val_loss: 0.3592245578765869\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc: 0.9531282782554626\n",
      "val_acc: 0.9305439591407776\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,8))\n",
    "plt.title('Loss Value')\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.legend(['loss', 'val_loss'])\n",
    "print('loss:', history.history['loss'][-1])\n",
    "print('val_loss:', history.history['val_loss'][-1])\n",
    "plt.show()\n",
    "plt.figure(figsize=(8,8))\n",
    "plt.title('Accuracy')\n",
    "plt.plot(history.history['acc'])\n",
    "plt.plot(history.history['val_acc'])\n",
    "plt.legend(['acc', 'val_acc'])\n",
    "print('acc:', history.history['acc'][-1])\n",
    "print('val_acc:', history.history['val_acc'][-1])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "33d93658",
   "metadata": {},
   "source": [
    "#### * 결과가 크게 나쁘지는 않지만 에폭이 늘어남에 따라 모델의 성능은 개선되지 않았다.\n",
    "\n",
    "#### 전형적인 오버피팅.\n",
    "\n",
    "#### 이 Pilot Model을 기반으로 좀 더 좋은 성능의 모델을 구축해 보고자 한다."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e44d39ca",
   "metadata": {},
   "source": [
    "### Model Renewal"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "be9f6692",
   "metadata": {},
   "source": [
    "조금 더 층을 단순하게 구성해보자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "d0e22fbb",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Conv2D(32, (3, 3), activation='relu', strides=(1, 1), \n",
    "    padding='same', input_shape=input_shape))\n",
    "model.add(MaxPool2D((2, 2)))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(64, activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(2, activation='softmax'))\n",
    "\n",
    "model.compile(loss='binary_crossentropy', \n",
    "     optimizer='adam',\n",
    "     metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "43858ec8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_2 (Conv2D)            (None, 300, 13, 32)       320       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 150, 6, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 150, 6, 32)        0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 28800)             0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 64)                1843264   \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 2)                 130       \n",
      "=================================================================\n",
      "Total params: 1,843,714\n",
      "Trainable params: 1,843,714\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "1c442778",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "150/150 [==============================] - 4s 21ms/step - loss: 0.6212 - acc: 0.6773 - val_loss: 0.3863 - val_acc: 0.8904\n",
      "Epoch 2/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.4027 - acc: 0.8611 - val_loss: 0.3608 - val_acc: 0.8611\n",
      "Epoch 3/40\n",
      "150/150 [==============================] - 3s 21ms/step - loss: 0.3835 - acc: 0.8656 - val_loss: 0.2900 - val_acc: 0.9105\n",
      "Epoch 4/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3518 - acc: 0.8862 - val_loss: 0.2909 - val_acc: 0.8987\n",
      "Epoch 5/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3236 - acc: 0.8949 - val_loss: 0.2894 - val_acc: 0.8987\n",
      "Epoch 6/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3269 - acc: 0.8997 - val_loss: 0.2807 - val_acc: 0.9113\n",
      "Epoch 7/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3284 - acc: 0.8996 - val_loss: 0.2688 - val_acc: 0.9121\n",
      "Epoch 8/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3157 - acc: 0.8961 - val_loss: 0.2780 - val_acc: 0.9155\n",
      "Epoch 9/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3091 - acc: 0.9065 - val_loss: 0.2640 - val_acc: 0.9121\n",
      "Epoch 10/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3100 - acc: 0.8986 - val_loss: 0.2666 - val_acc: 0.9088\n",
      "Epoch 11/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.3137 - acc: 0.8973 - val_loss: 0.2619 - val_acc: 0.9138\n",
      "Epoch 12/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2967 - acc: 0.9028 - val_loss: 0.2661 - val_acc: 0.9113\n",
      "Epoch 13/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2980 - acc: 0.9015 - val_loss: 0.3001 - val_acc: 0.9013\n",
      "Epoch 14/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2814 - acc: 0.9096 - val_loss: 0.2620 - val_acc: 0.9138\n",
      "Epoch 15/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.3015 - acc: 0.8996 - val_loss: 0.2854 - val_acc: 0.9088\n",
      "Epoch 16/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.2759 - acc: 0.9106 - val_loss: 0.2629 - val_acc: 0.9155\n",
      "Epoch 17/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2744 - acc: 0.9057 - val_loss: 0.2588 - val_acc: 0.9130\n",
      "Epoch 18/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2766 - acc: 0.9009 - val_loss: 0.2619 - val_acc: 0.9113\n",
      "Epoch 19/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.2848 - acc: 0.9025 - val_loss: 0.2864 - val_acc: 0.9105\n",
      "Epoch 20/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.2566 - acc: 0.9096 - val_loss: 0.2703 - val_acc: 0.9155\n",
      "Epoch 21/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.2514 - acc: 0.9124 - val_loss: 0.2678 - val_acc: 0.9121\n",
      "Epoch 22/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.2791 - acc: 0.9027 - val_loss: 0.2528 - val_acc: 0.9113\n",
      "Epoch 23/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.2594 - acc: 0.9121 - val_loss: 0.2548 - val_acc: 0.9079\n",
      "Epoch 24/40\n",
      "150/150 [==============================] - 3s 19ms/step - loss: 0.2684 - acc: 0.8993 - val_loss: 0.2875 - val_acc: 0.9121\n",
      "Epoch 25/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2767 - acc: 0.8940 - val_loss: 0.2718 - val_acc: 0.9105\n",
      "Epoch 26/40\n",
      "150/150 [==============================] - 3s 21ms/step - loss: 0.2620 - acc: 0.9118 - val_loss: 0.2658 - val_acc: 0.9096\n",
      "Epoch 27/40\n",
      "150/150 [==============================] - 3s 21ms/step - loss: 0.2504 - acc: 0.9153 - val_loss: 0.2716 - val_acc: 0.8987\n",
      "Epoch 28/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2544 - acc: 0.9070 - val_loss: 0.2555 - val_acc: 0.9113\n",
      "Epoch 29/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2614 - acc: 0.9077 - val_loss: 0.2610 - val_acc: 0.9088\n",
      "Epoch 30/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2554 - acc: 0.9168 - val_loss: 0.2646 - val_acc: 0.9113\n",
      "Epoch 31/40\n",
      "150/150 [==============================] - 3s 21ms/step - loss: 0.2616 - acc: 0.9066 - val_loss: 0.2685 - val_acc: 0.9138\n",
      "Epoch 32/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2367 - acc: 0.9166 - val_loss: 0.2530 - val_acc: 0.9146\n",
      "Epoch 33/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2575 - acc: 0.9104 - val_loss: 0.2756 - val_acc: 0.9138\n",
      "Epoch 34/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2589 - acc: 0.9071 - val_loss: 0.2564 - val_acc: 0.9113\n",
      "Epoch 35/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2320 - acc: 0.9107 - val_loss: 0.2896 - val_acc: 0.9138\n",
      "Epoch 36/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2381 - acc: 0.9094 - val_loss: 0.3202 - val_acc: 0.9105\n",
      "Epoch 37/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2506 - acc: 0.9059 - val_loss: 0.2620 - val_acc: 0.9138\n",
      "Epoch 38/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2249 - acc: 0.9180 - val_loss: 0.2647 - val_acc: 0.9130\n",
      "Epoch 39/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2371 - acc: 0.9128 - val_loss: 0.2588 - val_acc: 0.9130\n",
      "Epoch 40/40\n",
      "150/150 [==============================] - 3s 20ms/step - loss: 0.2274 - acc: 0.9205 - val_loss: 0.2915 - val_acc: 0.9113\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train, y_train, epochs=40, validation_data=(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "8b72e7e0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 0.23270484805107117\n",
      "val_loss: 0.2915021777153015\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc: 0.9163004755973816\n",
      "val_acc: 0.911297082901001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,8))\n",
    "plt.title('Loss Value')\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.legend(['loss', 'val_loss'])\n",
    "print('loss:', history.history['loss'][-1])\n",
    "print('val_loss:', history.history['val_loss'][-1])\n",
    "plt.show()\n",
    "plt.figure(figsize=(8,8))\n",
    "plt.title('Accuracy')\n",
    "plt.plot(history.history['acc'])\n",
    "plt.plot(history.history['val_acc'])\n",
    "plt.legend(['acc', 'val_acc'])\n",
    "print('acc:', history.history['acc'][-1])\n",
    "print('val_acc:', history.history['val_acc'][-1])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "81739f3e",
   "metadata": {},
   "source": [
    "### 성능이 많이 개선되었다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "4d8391b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "a5fc23d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = np.argmax(predictions, axis=1)\n",
    "y_test = one_hot_encoder.inverse_transform(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "7acd3d7e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAesAAAHgCAYAAACFNEViAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAeZUlEQVR4nO3debxVZb348c/3gAqCGZKYU4hDmvM8ZGJSVCaVlUPdJs0crjlk5a2b9rO0srSyTMvZ65xzat0UNWe9KSKhZVoKamqpqCjgAIfv74+9Dh6O55wOyDr7cfN5v1682Hvt4XlOufmcNey1IjORJEnlamv2BCRJUu+MtSRJhTPWkiQVzlhLklQ4Yy1JUuGMtSRJhRvY7An0ZPAmB/idMqkJnrvrhGZPQVpsDRpIdLfcNWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcMZakqTCGWtJkgpnrCVJKpyxliSpcAObPQG9+aw1cgTn/OiL8+6PWnk4R/3qd2y14SjWWm0FAN66zGCef/Eltv7UDwH4+hc/wB4f24b2uXP52jGXcN0d9zdl7lIr2XHsGJYeMoQBbW0MGDiACy66jBOO/xk33nA9bdHGsOHDOer7RzNixArNnqreoMjMZs+hW4M3OaDMiWk+bW3BQ9d8n+0/fyyPPvncvOU//OrHmT7jJY4+5WrWWf3tnHX0Hmz32R+z4vLL8r8nHcAGOx/J3Ln+X1yi5+46odlTUB/tOHYM5190CcOGLTdv2YwZMxg6dCgA5517Ng8/9He+fcSRzZqiFtCggUR3y90Mrjdkhy3XZso/np4v1ACfHLspF119NwDj3rshF18zkVdnz+GRJ6bx0GPPsMX6qzVhtlLr6wg1wMsvvUREt//2602m1s3gEbF+Zt5X5xhqrl0/uNm8KHfYdtM1+NezL/LQo08DsPLyy/LHe6fOe/zxp55jpRHL9uc0pdYUsN/eexER7LLr7uyy2+4A/OLnx3HVlb9h6NBlOO3Ms5s8SS0Kda9ZnxQRd0bE/hHx1prHUj9bYuAAdtp+Ay679p75lu/2oc25+OoJry3o5jf7Qve+SG8qZ517ARdecjknnnQqF15wHndPuAuAAw8+hPHX38RO4z7Cr88/t8mz1KJQa6wz8z3AZ4BVgQkRcX5EjO3p+RGxT0RMiIgJc575c51T0yLwwfesy6S/PsZTz744b9mAAW18bMxGXHLNxHnLHn/qeVZ5+7B591ceMYwnn57er3OVWlHHgWPDhw9nzPvHct+9k+d7fMedxnHdteObMTUtYrXvs87MvwGHA98AtgeOj4i/RsQnunnuKZm5eWZuPvBt69U9Nb1Bu31o89dtAh+z1do8OPVfPP7U8/OW/e7Gyez6wU1ZcomBjFxpOGu+Y3nuum9q/05WajGzZs1i5swZ827fcfttrLnmWjzyyNR5z7nxhj8watTqTZqhFqW691lvCOwJ7ARcC3wkMydGxErAHcBldY6v+gwetARjtlqHA753wXzLu9uHff/D/+TS8fdwz6WHMad9Ll/54UUeCS69Qc9Om8YhB30ZgDnt7Xx4p3Fsu91ovnrwgUydOoW2tmDFFVfm8CO+2+SZalGo9atbEXEzcBpwcWa+1OWxz2XmOT291q9uSc3hV7ek5unpq1u1rlln5uheHusx1JIk6TV1bwbfFvgOMLIaK4DMTHeiSJLUR3WfbvR04BDgbqC95rEkSWpJdcd6emb+vuYxJElqabXEOiI2rW7eEBHH0jjq+5WOxzNzYrcvlCRJr1PXmvVPutzfvNPtBMbUNK4kSS2nllhn5g4AEbF6Zj7c+bGI8OAySZIWQN1nMLukm2UX1zymJEktpa591usA6wHLdjmt6FuAQXWMKUlSq6prn/XawDjgrcBHOi1/Edi7pjElSWpJde2zvgK4IiK2ycw76hhDkqTFRV2bwX9B46hvIuLTXR/PzIPqGFeSpFZU12bwCTW9ryRJi526NoOfVcf7SpK0OKr7Qh7LA98A1qXTUeCZ6UlRJEnqo7q/Z30ecD8wCvguMBW4q+YxJUlqKXXHenhmng7MzsybMvOLwNY1jylJUkup+6pbs6u/n4yInYAngFVqHlOSpJZSd6y/FxHLAl8DfkHjDGaH1DymJEktpdZYZ+Zvq5vTgR3qHEuSpFZV6z7riHhnRFwfEfdV9zeMiMPrHFOSpFZT9wFmpwL/TbXvOjMnA5+qeUxJklpK3bFeOjPv7LJsTs1jSpLUUuqO9TMRsQavnSd8F+DJmseUJKml1H00+JeBU4B1IuJxYArwmZrHlCSppdQd68eBM4EbgOWAF4AvAEfWPK4kSS2j7lhfATwPTKRxQhRJkrSA6o71Kpn5oZrHkCSppdV9gNntEbFBzWNIktTSalmzjoh7aRwBPhDYMyIeBl4BAsjM3LCOcSVJakV1bQYfV9P7SpK02Kkl1pn5SB3vK0nS4qjufdaSJOkNMtaSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Yy1JEmFM9aSJBXOWEuSVDhjLUlS4Qb29EBEbNrbCzNz4qKfjiRJ6qrHWAM/6eWxBMYs4rlIkqRu9BjrzNyhPyciSZK692/3WUfE0hFxeEScUt1fKyLG1T81SZIEfTvA7EzgVeDd1f1/AN+rbUaSJGk+fYn1Gpl5DDAbIDNfAqLWWUmSpHn6EutXI2IwjYPKiIg1gFdqnZUkSZqnt6PBOxwBXA2sGhHnAdsCe9Q5KUmS9Jp/G+vMvDYiJgJb09j8fXBmPlP7zCRJEtC3NWuA7YH30NgUvgRweW0zkiRJ8+nLV7d+CewH3AvcB+wbESfWPTFJktTQlzXr7YH1M7PjALOzaIRbkiT1g74cDf4A8I5O91cFJtczHUmS1FVvF/K4isY+6mWB+yPizur+VsDt/TM9SZLU22bwH/fbLCRJUo96u5DHTf05EUmS1L2+HA2+dUTcFREzIuLViGiPiBf6Y3KSJKlvB5idAHwa+BswGPhStUySJPWDPp0UJTP/HhEDMrMdODMiPMBMkqR+0pdYz4qIJYFJEXEM8CQwpN5pSZKkDn3ZDP656nkHADNpfM/6E3VOSpIkvaYvF/J4pLr5MvBdgIi4ENi9xnlJkqRKX9asu7PNIp2FJEnq0cLGWpIk9ZOors/x+gciNu3pNcBvM3PF2mYFvDyH7icmqVYXTnqs2VOQFltf2HzV6G55b/usf9LLY399Y9ORJEl91dvpRnfoz4lIkqTuuc9akqTCGWtJkgpnrCVJKlxfrroVEfHZiPh/1f13RMSW9U9NkiRB39asf0njJCifru6/CJxY24wkSdJ8+nIhj60yc9OIuAcgM5+rLuwhSZL6QV/WrGdHxABonKQkIpYH5tY6K0mSNE9fYn08cDkwIiK+D9wK/KDWWUmSpHn6ctWt8yLibuB9NE41unNm3l/7zCRJEtCHWEfEO4BZwFWdl2Xmo3VOTJIkNfTlALPf0dhfHcAgYBTwALBejfOSJEmVvmwG36Dz/epqXPvWNiNJkjSfBT6DWWZOBLaoYS6SJKkbfdln/dVOd9uATYGna5uRJEmaT1/2WS/T6fYcGvuwL61nOpIkqateY12dDGVoZh7aT/ORJEld9LjPOiIGZmY7jc3ekiSpSXpbs76TRqgnRcSVwMXAzI4HM/OymucmSZLo2z7r5YBpwBhe+751AsZakqR+0FusR1RHgt/Ha5HukLXOSpIkzdNbrAcAQ5k/0h2MtSRJ/aS3WD+ZmUf220wkSVK3ejuDWXdr1JIkqZ/1Fuv39dssJElSj3qMdWY+258TkSRJ3VvgC3lIkqT+ZawlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwhlrSZIKZ6wlSSqcsZYkqXDGWpKkwg1s9gT05rfj2DEsPWQIA9raGDBwABdcdBknHP8zbrzhetqijWHDh3PU949mxIgVmj1V6U1tzquvcs5Rh9A+ZzZz29tZZ8vRjN7lC9z/x5u45dKzeeaJR9nzyBNYcfW1AWifM5vfn/4znnz4AaKtjbGf25+R627c3B9CCyUys9lz6NbLcyhzYnqdHceO4fyLLmHYsOXmLZsxYwZDhw4F4Lxzz+bhh/7Ot484sllT1AK4cNJjzZ6CepCZzH7lZZYcNJj2OXM458ivMPZz+7PU0kOIaOP3ZxzH+/5j33mxnjD+Cv455UHG7XsoM6c/x4XHfIs9jzqRaHOjaqm+sPmq0d1y/x9TLTpCDfDySy8R0e1/f5IWQESw5KDBAMxtn0N7+xyI4G0rj2T4Squ+7vnPPP4Iq623CQBDlh3GUkOG8uSUB/t1zlo03AyuNy5gv733IiLYZdfd2WW33QH4xc+P46orf8PQoctw2plnN3mSUmuYO7edMw7bn+f+9Tibjf0YK6/5rh6fu8LI1Xnw7ttZd5sdeGHaU/xzyoO8MO0pVlpjnX6csRaFWmMdEcd3s3g6MCEzr6hzbPWfs869gBEjVmDatGns96U9GbX66my2+RYcePAhHHjwIZx+6sn8+vxz2f+Ag5o9VelNr61tAF86+mRenjmDS447gqcem8KIVUd1+9yNtt+RZx5/lDMO359l3zaCVdZaj7a2Af08Yy0KdW8GHwRsDPyt+rMhsBywV0T8rOuTI2KfiJgQERNOP/WUmqemRaXjwLHhw4cz5v1jue/eyfM9vuNO47ju2vHNmJrUsgYNGcrId23Ew5Pv6vE5bQMGMPZz+/Olo09m168dxcuzZjDs7av04yy1qNS9GXxNYExmzgGIiF8B44GxwL1dn5yZpwCngAeYvVnMmjWLzLkMGTKUWbNmccftt7HvfvvzyCNTGTlyNQBuvOEPjBq1enMnKrWAmS88z4ABAxk0ZCizX32FKX+eyDbjPtXj82e/8jKZyZKDBjPl3rtpaxvA8quM7McZa1GpO9YrA0NobPqmur1SZrZHxCs1j61+8Oy0aRxy0JcBmNPezod3Gse2243mqwcfyNSpU2hrC1ZccWUOP+K7TZ6p9OY38/lnueqkHzF37lwyk3dttT1rbbo1D9x1K+PPOoFZL07nwmMPY4WRa/Dpb/6ImS88z69/9E0i2lhm2HA++p/fbPaPoIVU61e3ImIv4HDgRiCA0cAPgAuA72TmoT291jVrqTn86pbUPD19davWNevMPD0i/hfYkkasv5WZT1QP9xhqSZL0mv74nnUb8DTwLLBmRIzuhzElSWoZdX9160fA7sCfgbnV4gRurnNcSZJaSd0HmO0MrJ2ZHkwmSdJCqnsz+MPAEjWPIUlSS6t7zXoWMCkirgfmrV1npqeykiSpj+qO9ZXVH0mStJDq/urWWXW+vyRJi4NaYh0RF2XmbhFxL7z+5CaZuWEd40qS1IrqWrM+uPp7XE3vL0nSYqOWo8Ez88nq5ueB9sx8pOMP8ME6xpQkqVXV/dWtA4HxEbFDp2X71TymJEktpe5YPw58CPhhRHScC7zbk5RLkqTu1X5u8Mx8FNgeWDciLgYG1z2mJEmtpO5YTwDIzJczc08al8pcsuYxJUlqKbXGOjP3jojBEbF2df/EzFy9zjElSWo1tcY6Ij4CTAKuru5vHBGe0UySpAVQ92bw7wBbAs8DZOYkYFTNY0qS1FLqjvWczJzeZdnrzmgmSZJ6VveFPO6LiP8ABkTEWsBBwO01jylJUkvpj5OirEfj8pjnA9OBr9Q8piRJLaXuq27NAg6LiB9k5sw6x5IkqVXVfTT4uyPiL8D91f2NIuKXdY4pSVKrqXsz+HE0LtwxDSAz/wSMrnlMSZJaSn+cbvSxLova6x5TkqRWUvfR4I9FxLuBjIglaRwNfn/NY0qS1FLqXrPeD/gysDLwD2Dj6r4kSeqjuo8Gfwb4TJ1jSJLU6mqNdUQsD+wNrNZ5rMz8Yp3jSpLUSureZ30FcAtwHR5YJknSQqk71ktn5jdqHkOSpJZW9wFmv42ID9c8hiRJLa2WNeuIeJHXrq71rYh4BZhT3c/MfEsd40qS1IpqiXVmLgMQEefQ2Gd9S2b6/WpJkhZC3ZvBzwTeDhwfEQ9FxCURcXDNY0qS1FLq/p71HyLiJmALYAcaJ0lZH/h5neNKktRK6v6e9fXAEOAOGpvDt8jMp+ocU5KkVlP3ZvDJwKs01qY3BNaPiME1jylJUkupezP4IQARMRTYk9f2YS9V57iSJLWSujeDHwBsB2wGPAKcQWNzuCRJ6qO6z2A2GPgpcHdmzvl3T5YkSa9X92bwY+t8f0mSFgd1H2AmSZLeIGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYUz1pIkFc5YS5JUOGMtSVLhjLUkSYWLzGz2HNSCImKfzDyl2fOQFjd+9lqTa9aqyz7NnoC0mPKz14KMtSRJhTPWkiQVzlirLu4zk5rDz14L8gAzSZIK55q1JEmFM9aaT0SsFhH3LYL32SMiTqhu7xwR63Z67MaI2PyNjiG1qoiYUeN7rxQRl9T1/qqHsVZ/2BlY9989SVL9MvOJzNyl2fPQgjHW6s6AiDg1Iv4cEeMjYnBErBERV0fE3RFxS0SsAxARH4mIP0bEPRFxXUSs0PmNIuLdwEeBYyNiUkSsUT20a0TcGREPRsR21XNviYiNO732tojYsH9+ZKn/RMR/RcRB1e3jIuIP1e33RcS51e3vR8SfIuL/Oj5XEbF8RFwaEXdVf7atlg+JiDOqZfdExMeq5adVn7tJEfF0RBzReetZtQXssuqz/beIOKbTHPeqPp83Vv8enNC//yupM2Ot7qwFnJiZ6wHPA5+kcYTpgZm5GfB14JfVc28Fts7MTYBfA//V+Y0y83bgSuDQzNw4Mx+qHhqYmVsCXwGOqJadBuwBEBHvBJbKzMl1/IBSk90MbFfd3hwYGhFLAO8BbgGGAP+XmRtVz927eu7PgeMycwsan8vTquWHAX+olu9A45fjIZn5pczcGPgYMA34n27msjGwO7ABsHtErBoRKwHfBrYGxgLrLKKfWwtpYLMnoCJNycxJ1e27gdWAdwMXR0THc5aq/l4FuDAiVgSWBKb0cYzLurw/wMXAtyPiUOCLdP8Pi9QK7gY2i4hlgFeAiTSivR1wEPAq8NtOzx1b3X4/sG6nz+Fbqvf4APDRiPh6tXwQ8A7g/ogYROOzdUBmPhIRq3WZy/WZOR0gIv4CjATeBtyUmc9Wyy8G3rmIfnYtBGOt7rzS6XY7sALwfPUbele/AH6amVdGxHuB7yzgGO1U/x1m5qyIuJbGWsBuNP7xklpOZs6OiKnAnsDtwGQaa8RrAPcDs/O179XO+4zQ2Bq6TWa+1Pn9olHvT2bmA90MdxJwWWZe18N0un7eBwLRw3PVJG4GV1+8AEyJiF2h8Q9DRGxUPbYs8Hh1+ws9vP5FYJk+jnUacDxwV8dv9VKLupnGLqWbaWz63g+Y1CnS3RkPHNBxp9MxHtcAB1bRJiI2qf7+MrBMZv5wAed2J7B9RAyLiIE0NrmriYy1+uozwF4R8SfgzzTWfqGxJn1xRNwCPNPDa38NHFod+LJGD88BIDPvpvHLwZmLZNZSuW4BVgTuyMx/AS9Xy3pzELB5REyuNlnvVy0/ClgCmFwdPHZUtfzrwAadDjLb7/Vv+XqZ+TjwA+CPwHXAX4Dpff/RtKh5BjMVpTqw5UZgncyc2+TpSIutiBiamTOqNevLgTMy8/Jmz2tx5Zq1ihERn6fxm/xhhlpquu9ExCTgPhoHjv6mqbNZzLlmLUlS4VyzliSpcMZakqTCGWtJkgpnrKUmioj26is190XExRGx9Bt4r/+JiF2q26d1vtJZN899b3Xe9gUdY2pEvK2vy3t4j3lXZHuj40qLC2MtNddL1TnT16dxisn5vgcbEQMW5k2rc0L/pZenvJfGKWQlvQkYa6kctwBrVmu9N0TE+cC9ETEgIo6trqg0OSL2hXlnkjshIv4SEb8DRnS8UXS6ZnhEfCgiJlZXcLq+Ojf0fsAh1Vr9dr1czWl4NK68dk9EnMwCnIYyIraMiNur194eEWt3enjV6kpPD0TEEZ1e89loXI1tUkScvLC/rEitxnODSwWoTjyxI3B1tWhLYP3MnBIR+wDTM3OLiFgKuC0ixgObAGvTuFrSCjTOMnVGl/ddHjgVGF2913KZ+WxEnATMyMwfV887n8bVnG6NiHfQOH3lu2hcEe3WzDwyInYC9lmAH+uv1bhzIuL9NM6I1XHayi2B9YFZwF3VLxszaVz9advq3Nm/pHHmvLMXYEypJRlrqbkGVyeegMaa9ek0Nk/fmZkdVzD7ALBhx/5oGudjXwsYDVyQme3AE1FdE7mLrYGbO96rl/Ot93Q1p9HAJ6rX/i4inluAn21Z4KyIWAtIGqfD7HBtZk4DiIjLaFwacg6wGY14AwwGnlqA8aSWZayl5nqp69XMqlDN7LyIxrXEr+nyvA/TiGBvog/PgZ6v5kQfX9+do4AbMvPj1ab3Gzs91vU9s5rrWZn53ws5ntSy3Gctle8a4D8jYgmAiHhnRAyhcbWmT1X7tFekcYnFru6gcfWkUdVrl6uWd70SWk9Xc7qZxqZoImJHYNgCzLvzFdn26PLY2IhYLiIGAzsDtwHXA7tExIiOuUbEyAUYT2pZxloq32k09kdPrK6odDKNrWKXA38D7gV+BdzU9YWZ+TSN/cyXVVdMu7B66Crg4x0HmNHz1Zy+C4yOiIk0Nsc/2ss8J0fEP6o/PwWOAY6OiNuArgeK3QqcA0wCLs3MCdXR64cD4yNiMnAtjatSSYs9zw0uSVLhXLOWJKlwxlqSpMIZa0mSCmesJUkqnLGWJKlwxlqSpMIZa0mSCmesJUkq3P8H/fSCITQV/KwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test, predictions)\n",
    "plt.figure(figsize=(8,8))\n",
    "sns.heatmap(cm, annot=True, xticklabels=label_encoder.classes_, yticklabels=label_encoder.classes_, fmt='d', cmap=plt.cm.Blues, cbar=False)\n",
    "plt.xlabel('Predicted Label')\n",
    "plt.ylabel('True Label')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}