tf_keras_DNN_iris_SYJang.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "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": "3.8.8"
    },
    "colab": {
      "name": "tf.keras_DNN_iris_SYJang.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/so02e/ML_python/blob/main/tf_keras_DNN_iris_SYJang.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EYWPrHpHvOaT",
        "outputId": "b296fb96-634d-4f44-8ece-3d4655de0c50"
      },
      "source": [
        " from google.colab import drive\n",
        "\n",
        "drive.mount('/content/gdrive')   # google drive를 google colab에 연결. 처음 실행 시, 인증 필요 "
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "22UZuhWkLtVt"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras import layers  #모듈(변수나 함수를 포함)만 불러오기\n",
        "\n",
        "# iris 데이터를 읽어 들이고 정규화하기\n",
        "dir = \"/content/gdrive/My Drive/Colab Notebooks/AI/\"\n",
        "df = pd.read_csv(dir + \"iris.csv\")"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 417
        },
        "id": "3GAFF1QOk58E",
        "outputId": "9abdabda-558c-4ea6-bd3e-e36e4933cbf8"
      },
      "source": [
        "df"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>sepal_length</th>\n",
              "      <th>sepal_width</th>\n",
              "      <th>petal_length</th>\n",
              "      <th>petal_width</th>\n",
              "      <th>iris_type</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>6.4</td>\n",
              "      <td>3.1</td>\n",
              "      <td>5.5</td>\n",
              "      <td>1.8</td>\n",
              "      <td>Iris-virginica</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>6.5</td>\n",
              "      <td>3.0</td>\n",
              "      <td>5.8</td>\n",
              "      <td>2.2</td>\n",
              "      <td>Iris-virginica</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>4.6</td>\n",
              "      <td>3.1</td>\n",
              "      <td>1.5</td>\n",
              "      <td>0.2</td>\n",
              "      <td>Iris-setosa</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>6.4</td>\n",
              "      <td>2.8</td>\n",
              "      <td>5.6</td>\n",
              "      <td>2.1</td>\n",
              "      <td>Iris-virginica</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>5.0</td>\n",
              "      <td>3.3</td>\n",
              "      <td>1.4</td>\n",
              "      <td>0.2</td>\n",
              "      <td>Iris-setosa</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>145</th>\n",
              "      <td>5.1</td>\n",
              "      <td>3.8</td>\n",
              "      <td>1.9</td>\n",
              "      <td>0.4</td>\n",
              "      <td>Iris-setosa</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>146</th>\n",
              "      <td>5.7</td>\n",
              "      <td>2.8</td>\n",
              "      <td>4.5</td>\n",
              "      <td>1.3</td>\n",
              "      <td>Iris-versicolor</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>147</th>\n",
              "      <td>6.9</td>\n",
              "      <td>3.1</td>\n",
              "      <td>5.4</td>\n",
              "      <td>2.1</td>\n",
              "      <td>Iris-virginica</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>148</th>\n",
              "      <td>7.2</td>\n",
              "      <td>3.0</td>\n",
              "      <td>5.8</td>\n",
              "      <td>1.6</td>\n",
              "      <td>Iris-virginica</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>149</th>\n",
              "      <td>4.9</td>\n",
              "      <td>3.0</td>\n",
              "      <td>1.4</td>\n",
              "      <td>0.2</td>\n",
              "      <td>Iris-setosa</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>150 rows × 5 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "     sepal_length  sepal_width  petal_length  petal_width        iris_type\n",
              "0             6.4          3.1           5.5          1.8   Iris-virginica\n",
              "1             6.5          3.0           5.8          2.2   Iris-virginica\n",
              "2             4.6          3.1           1.5          0.2      Iris-setosa\n",
              "3             6.4          2.8           5.6          2.1   Iris-virginica\n",
              "4             5.0          3.3           1.4          0.2      Iris-setosa\n",
              "..            ...          ...           ...          ...              ...\n",
              "145           5.1          3.8           1.9          0.4      Iris-setosa\n",
              "146           5.7          2.8           4.5          1.3  Iris-versicolor\n",
              "147           6.9          3.1           5.4          2.1   Iris-virginica\n",
              "148           7.2          3.0           5.8          1.6   Iris-virginica\n",
              "149           4.9          3.0           1.4          0.2      Iris-setosa\n",
              "\n",
              "[150 rows x 5 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8vDzDZklFMPE",
        "outputId": "a0e8e8c9-437e-47ff-abb0-6df2b9dc3b68"
      },
      "source": [
        "print(df['iris_type'].unique())"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['Iris-virginica' 'Iris-setosa' 'Iris-versicolor']\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "ZP1AR5S9KNZW",
        "outputId": "2e01d703-5371-4fca-9660-6c0b367528ac"
      },
      "source": [
        "df.describe()"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>sepal_length</th>\n",
              "      <th>sepal_width</th>\n",
              "      <th>petal_length</th>\n",
              "      <th>petal_width</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>150.000000</td>\n",
              "      <td>150.000000</td>\n",
              "      <td>150.000000</td>\n",
              "      <td>150.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>5.843333</td>\n",
              "      <td>3.054000</td>\n",
              "      <td>3.758667</td>\n",
              "      <td>1.198667</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>0.828066</td>\n",
              "      <td>0.433594</td>\n",
              "      <td>1.764420</td>\n",
              "      <td>0.763161</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>4.300000</td>\n",
              "      <td>2.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.100000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>5.100000</td>\n",
              "      <td>2.800000</td>\n",
              "      <td>1.600000</td>\n",
              "      <td>0.300000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>5.800000</td>\n",
              "      <td>3.000000</td>\n",
              "      <td>4.350000</td>\n",
              "      <td>1.300000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>6.400000</td>\n",
              "      <td>3.300000</td>\n",
              "      <td>5.100000</td>\n",
              "      <td>1.800000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>7.900000</td>\n",
              "      <td>4.400000</td>\n",
              "      <td>6.900000</td>\n",
              "      <td>2.500000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       sepal_length  sepal_width  petal_length  petal_width\n",
              "count    150.000000   150.000000    150.000000   150.000000\n",
              "mean       5.843333     3.054000      3.758667     1.198667\n",
              "std        0.828066     0.433594      1.764420     0.763161\n",
              "min        4.300000     2.000000      1.000000     0.100000\n",
              "25%        5.100000     2.800000      1.600000     0.300000\n",
              "50%        5.800000     3.000000      4.350000     1.300000\n",
              "75%        6.400000     3.300000      5.100000     1.800000\n",
              "max        7.900000     4.400000      6.900000     2.500000"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WaC6EJH_kwwz"
      },
      "source": [
        "X = df.iloc[:,0:4]\n",
        "\n",
        "# 레이블\n",
        "bclass = {\"Iris-virginica\":[1,0,0], \"Iris-setosa\":[0,1,0], \"Iris-versicolor\":[0,0,1]}\n",
        "y = np.empty((150,3))     # 150x3 크기의 다차원 벡터 생성\n",
        "for i, v in enumerate(df[\"iris_type\"]):\n",
        "    y[i] = bclass[v]        #\"sepal_length\"이면, y[i]=[1,0,0] 와 같이 할당\n",
        "    \n",
        "# 훈련 전용 데이터와 테스트 전용 데이터로 나누기\n",
        "X_train, y_train = X[0:100], y[0:100]\n",
        "X_test,  y_test  = X[101:151], y[101:151]"
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": true,
        "id": "G2UgYhccLtVu"
      },
      "source": [
        "# 모델 구조 정의하기\n",
        "model = tf.keras.Sequential()  #순차적 계층화 준비\n",
        "model.add(layers.Dense(64, input_shape=(4,)))  #입력 4개로부터 전달받는 64개 노드의 layer 생성\n",
        "model.add(layers.Activation('relu'))  #ReLU 활성화함수 채택\n",
        "model.add(layers.Dropout(0.2))        #dropout ratio=20% (배치 훈련시 20% arc 무시)\n",
        "\n",
        "\n",
        "model.add(layers.Dense(32))         #16개 노드의 layer 생성\n",
        "model.add(layers.Activation('relu')) #ReLU 활성화함수 채택\n",
        "model.add(layers.Dropout(0.1))  #dropout ratio=10% (배치 훈련시 10% arc 무시)\n",
        "\n",
        "\n",
        "model.add(layers.Dense(3))\n",
        "model.add(layers.Activation('softmax')) #분류(classification)을 위해 softmax 함수 사용\n",
        "\n",
        "# 모델 구축하기\n",
        "model.compile(\n",
        "    loss='categorical_crossentropy',  #다중 교차엔트로피\n",
        "    optimizer=\"rmsprop\",   #최적화 기법 중 하나\n",
        "    metrics=['accuracy'])  #정확도 측정"
      ],
      "execution_count": 150,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": true,
        "id": "2IhHchdXLtVv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6844d19a-0d8c-4d4d-e683-b909d9a9f276"
      },
      "source": [
        "# 데이터 훈련하기\n",
        "hist = model.fit(\n",
        "    X_train, y_train,\n",
        "    batch_size=10, #10개에 한 번씩 업데이터 실행\n",
        "    epochs=300, #훈련 데이터셋을 총 300회 반복 실험. 단, 조기중지될 수 있음\n",
        "    validation_split=0.1,  \n",
        "      #validation data 분할 비율. 즉, 150개 중에서 10%인 15개를 validation용으로 분할\n",
        "    callbacks=[tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=5)],  \n",
        "      #'val_loss'를 monitor하여 감소하면 한 번 더(1) 참고 조기중지\n",
        "    verbose=1)  #전 과정을 화면에 출력(1) 또는 미출력(0) 모드\n",
        "\n",
        "# 테스트 데이터로 평가하기\n",
        "score = model.evaluate(X_test, y_test)\n",
        "print('test_loss: ', score[0])\n",
        "print('test_acc: ', score[1])"
      ],
      "execution_count": 151,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/300\n",
            "9/9 [==============================] - 1s 20ms/step - loss: 1.1388 - accuracy: 0.3667 - val_loss: 1.1116 - val_accuracy: 0.1000\n",
            "Epoch 2/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.9988 - accuracy: 0.4778 - val_loss: 0.9785 - val_accuracy: 0.7000\n",
            "Epoch 3/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.9834 - accuracy: 0.5333 - val_loss: 0.8238 - val_accuracy: 0.7000\n",
            "Epoch 4/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.9188 - accuracy: 0.5778 - val_loss: 0.7847 - val_accuracy: 0.7000\n",
            "Epoch 5/300\n",
            "9/9 [==============================] - 0s 5ms/step - loss: 0.8907 - accuracy: 0.5889 - val_loss: 0.7650 - val_accuracy: 0.7000\n",
            "Epoch 6/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.8210 - accuracy: 0.5556 - val_loss: 0.7041 - val_accuracy: 0.7000\n",
            "Epoch 7/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.8013 - accuracy: 0.5667 - val_loss: 0.5863 - val_accuracy: 0.7000\n",
            "Epoch 8/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.7065 - accuracy: 0.7000 - val_loss: 0.5602 - val_accuracy: 0.7000\n",
            "Epoch 9/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.7015 - accuracy: 0.7000 - val_loss: 0.4778 - val_accuracy: 0.7000\n",
            "Epoch 10/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.6502 - accuracy: 0.7111 - val_loss: 0.4040 - val_accuracy: 1.0000\n",
            "Epoch 11/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.6423 - accuracy: 0.7111 - val_loss: 0.3612 - val_accuracy: 1.0000\n",
            "Epoch 12/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.5652 - accuracy: 0.7889 - val_loss: 0.3196 - val_accuracy: 1.0000\n",
            "Epoch 13/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.5498 - accuracy: 0.7667 - val_loss: 0.2868 - val_accuracy: 1.0000\n",
            "Epoch 14/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.4906 - accuracy: 0.7778 - val_loss: 0.2668 - val_accuracy: 1.0000\n",
            "Epoch 15/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.5210 - accuracy: 0.7444 - val_loss: 0.2286 - val_accuracy: 1.0000\n",
            "Epoch 16/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.5107 - accuracy: 0.7556 - val_loss: 0.2093 - val_accuracy: 1.0000\n",
            "Epoch 17/300\n",
            "9/9 [==============================] - 0s 5ms/step - loss: 0.4539 - accuracy: 0.7889 - val_loss: 0.2029 - val_accuracy: 1.0000\n",
            "Epoch 18/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.4549 - accuracy: 0.8000 - val_loss: 0.1945 - val_accuracy: 1.0000\n",
            "Epoch 19/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.4542 - accuracy: 0.7444 - val_loss: 0.1627 - val_accuracy: 1.0000\n",
            "Epoch 20/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.4360 - accuracy: 0.8111 - val_loss: 0.1541 - val_accuracy: 1.0000\n",
            "Epoch 21/300\n",
            "9/9 [==============================] - 0s 5ms/step - loss: 0.3440 - accuracy: 0.8667 - val_loss: 0.1550 - val_accuracy: 1.0000\n",
            "Epoch 22/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.3968 - accuracy: 0.8333 - val_loss: 0.1510 - val_accuracy: 1.0000\n",
            "Epoch 23/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.3471 - accuracy: 0.8444 - val_loss: 0.1477 - val_accuracy: 1.0000\n",
            "Epoch 24/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.4028 - accuracy: 0.8000 - val_loss: 0.1323 - val_accuracy: 1.0000\n",
            "Epoch 25/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.3300 - accuracy: 0.9000 - val_loss: 0.1201 - val_accuracy: 1.0000\n",
            "Epoch 26/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.3232 - accuracy: 0.9000 - val_loss: 0.1440 - val_accuracy: 1.0000\n",
            "Epoch 27/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.3283 - accuracy: 0.8778 - val_loss: 0.1098 - val_accuracy: 1.0000\n",
            "Epoch 28/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.3082 - accuracy: 0.8778 - val_loss: 0.1182 - val_accuracy: 1.0000\n",
            "Epoch 29/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2866 - accuracy: 0.9000 - val_loss: 0.1266 - val_accuracy: 1.0000\n",
            "Epoch 30/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2900 - accuracy: 0.9222 - val_loss: 0.0924 - val_accuracy: 1.0000\n",
            "Epoch 31/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2998 - accuracy: 0.8778 - val_loss: 0.0976 - val_accuracy: 1.0000\n",
            "Epoch 32/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.3165 - accuracy: 0.8889 - val_loss: 0.0961 - val_accuracy: 1.0000\n",
            "Epoch 33/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2557 - accuracy: 0.9333 - val_loss: 0.0663 - val_accuracy: 1.0000\n",
            "Epoch 34/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2695 - accuracy: 0.8889 - val_loss: 0.0695 - val_accuracy: 1.0000\n",
            "Epoch 35/300\n",
            "9/9 [==============================] - 0s 6ms/step - loss: 0.2239 - accuracy: 0.9556 - val_loss: 0.0583 - val_accuracy: 1.0000\n",
            "Epoch 36/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2358 - accuracy: 0.9222 - val_loss: 0.0562 - val_accuracy: 1.0000\n",
            "Epoch 37/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2470 - accuracy: 0.9000 - val_loss: 0.0531 - val_accuracy: 1.0000\n",
            "Epoch 38/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.2235 - accuracy: 0.9333 - val_loss: 0.0493 - val_accuracy: 1.0000\n",
            "Epoch 39/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.2226 - accuracy: 0.9333 - val_loss: 0.0644 - val_accuracy: 1.0000\n",
            "Epoch 40/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2221 - accuracy: 0.9333 - val_loss: 0.0464 - val_accuracy: 1.0000\n",
            "Epoch 41/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2522 - accuracy: 0.8889 - val_loss: 0.0454 - val_accuracy: 1.0000\n",
            "Epoch 42/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2170 - accuracy: 0.9111 - val_loss: 0.0336 - val_accuracy: 1.0000\n",
            "Epoch 43/300\n",
            "9/9 [==============================] - 0s 5ms/step - loss: 0.2251 - accuracy: 0.9111 - val_loss: 0.0376 - val_accuracy: 1.0000\n",
            "Epoch 44/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2198 - accuracy: 0.9111 - val_loss: 0.0414 - val_accuracy: 1.0000\n",
            "Epoch 45/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1774 - accuracy: 0.9667 - val_loss: 0.0276 - val_accuracy: 1.0000\n",
            "Epoch 46/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2124 - accuracy: 0.8889 - val_loss: 0.0356 - val_accuracy: 1.0000\n",
            "Epoch 47/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1939 - accuracy: 0.9222 - val_loss: 0.0279 - val_accuracy: 1.0000\n",
            "Epoch 48/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2074 - accuracy: 0.9222 - val_loss: 0.0366 - val_accuracy: 1.0000\n",
            "Epoch 49/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.2006 - accuracy: 0.9111 - val_loss: 0.0286 - val_accuracy: 1.0000\n",
            "Epoch 50/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.1407 - accuracy: 0.9778 - val_loss: 0.0232 - val_accuracy: 1.0000\n",
            "Epoch 51/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1661 - accuracy: 0.9333 - val_loss: 0.0253 - val_accuracy: 1.0000\n",
            "Epoch 52/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1792 - accuracy: 0.9333 - val_loss: 0.0188 - val_accuracy: 1.0000\n",
            "Epoch 53/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1642 - accuracy: 0.9333 - val_loss: 0.0221 - val_accuracy: 1.0000\n",
            "Epoch 54/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1580 - accuracy: 0.9333 - val_loss: 0.0177 - val_accuracy: 1.0000\n",
            "Epoch 55/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1693 - accuracy: 0.9667 - val_loss: 0.0243 - val_accuracy: 1.0000\n",
            "Epoch 56/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1282 - accuracy: 0.9667 - val_loss: 0.0165 - val_accuracy: 1.0000\n",
            "Epoch 57/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1491 - accuracy: 0.9556 - val_loss: 0.0159 - val_accuracy: 1.0000\n",
            "Epoch 58/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1222 - accuracy: 0.9667 - val_loss: 0.0254 - val_accuracy: 1.0000\n",
            "Epoch 59/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1695 - accuracy: 0.9222 - val_loss: 0.0174 - val_accuracy: 1.0000\n",
            "Epoch 60/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1464 - accuracy: 0.9444 - val_loss: 0.0225 - val_accuracy: 1.0000\n",
            "Epoch 61/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1516 - accuracy: 0.9556 - val_loss: 0.0150 - val_accuracy: 1.0000\n",
            "Epoch 62/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1302 - accuracy: 0.9444 - val_loss: 0.0119 - val_accuracy: 1.0000\n",
            "Epoch 63/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1861 - accuracy: 0.9222 - val_loss: 0.0131 - val_accuracy: 1.0000\n",
            "Epoch 64/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1484 - accuracy: 0.9556 - val_loss: 0.0134 - val_accuracy: 1.0000\n",
            "Epoch 65/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1350 - accuracy: 0.9444 - val_loss: 0.0115 - val_accuracy: 1.0000\n",
            "Epoch 66/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1294 - accuracy: 0.9333 - val_loss: 0.0115 - val_accuracy: 1.0000\n",
            "Epoch 67/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1141 - accuracy: 0.9556 - val_loss: 0.0092 - val_accuracy: 1.0000\n",
            "Epoch 68/300\n",
            "9/9 [==============================] - 0s 5ms/step - loss: 0.1478 - accuracy: 0.9444 - val_loss: 0.0109 - val_accuracy: 1.0000\n",
            "Epoch 69/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1435 - accuracy: 0.9333 - val_loss: 0.0128 - val_accuracy: 1.0000\n",
            "Epoch 70/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1587 - accuracy: 0.9222 - val_loss: 0.0113 - val_accuracy: 1.0000\n",
            "Epoch 71/300\n",
            "9/9 [==============================] - 0s 3ms/step - loss: 0.1328 - accuracy: 0.9444 - val_loss: 0.0104 - val_accuracy: 1.0000\n",
            "Epoch 72/300\n",
            "9/9 [==============================] - 0s 4ms/step - loss: 0.1190 - accuracy: 0.9556 - val_loss: 0.0112 - val_accuracy: 1.0000\n",
            "2/2 [==============================] - 0s 6ms/step - loss: 0.1037 - accuracy: 1.0000\n",
            "test_loss:  0.10372401028871536\n",
            "test_acc:  1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eVZ2sf6Gb5Y-",
        "outputId": "50e22b26-a080-4c75-a6d2-8cfa4f8b4fd7"
      },
      "source": [
        "hist.history.keys()"
      ],
      "execution_count": 156,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])"
            ]
          },
          "metadata": {},
          "execution_count": 156
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": true,
        "id": "Hh1NwcAwLtVw",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "outputId": "4096a35b-4ac8-4f0c-b358-00f1ee7f333a"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "fig, loss_ax = plt.subplots()\n",
        "acc_ax = loss_ax.twinx()\n",
        "\n",
        "loss_ax.plot(hist.history['loss'], 'y', label='train loss')\n",
        "loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')\n",
        "\n",
        "acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')\n",
        "acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')\n",
        "\n",
        "loss_ax.set_xlabel('epoch')\n",
        "loss_ax.set_ylabel('loss')\n",
        "acc_ax.set_ylabel('accuray')\n",
        "\n",
        "loss_ax.legend(loc='upper left')\n",
        "acc_ax.legend(loc='lower left')\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 157,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a0LNN770LtVw",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5a39fea0-816c-4849-c576-edbdb856b549"
      },
      "source": [
        "model.summary()"
      ],
      "execution_count": 158,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_11\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_35 (Dense)             (None, 64)                320       \n",
            "_________________________________________________________________\n",
            "activation_35 (Activation)   (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dropout_24 (Dropout)         (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dense_36 (Dense)             (None, 32)                2080      \n",
            "_________________________________________________________________\n",
            "activation_36 (Activation)   (None, 32)                0         \n",
            "_________________________________________________________________\n",
            "dropout_25 (Dropout)         (None, 32)                0         \n",
            "_________________________________________________________________\n",
            "dense_37 (Dense)             (None, 3)                 99        \n",
            "_________________________________________________________________\n",
            "activation_37 (Activation)   (None, 3)                 0         \n",
            "=================================================================\n",
            "Total params: 2,499\n",
            "Trainable params: 2,499\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ap06VxZI5Aog",
        "outputId": "c6fc1b22-6961-433c-c2da-2fa7bccca81c"
      },
      "source": [
        "model.get_weights()"
      ],
      "execution_count": 159,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[array([[-0.14364223, -0.09170078,  0.20662187,  0.06120177, -0.17792587,\n",
              "          0.08446594,  0.15546623, -0.06156182,  0.06298817, -0.04801513,\n",
              "         -0.22814031,  0.25975114,  0.25974292,  0.18507381,  0.11606517,\n",
              "         -0.15730017, -0.0345535 ,  0.01606034, -0.17482403,  0.0631735 ,\n",
              "          0.08057135,  0.0688    ,  0.02256304, -0.30598953,  0.03605801,\n",
              "         -0.23490278, -0.11449122, -0.23421532,  0.12973721,  0.0874772 ,\n",
              "          0.05354551,  0.01899162,  0.29698864, -0.13798812, -0.06364188,\n",
              "         -0.10097924,  0.09643658, -0.23689193,  0.09288131,  0.06741929,\n",
              "         -0.06151901,  0.13102931, -0.2699818 ,  0.02343902,  0.1824122 ,\n",
              "         -0.14786683, -0.23454258, -0.10374556,  0.00739726,  0.03589683,\n",
              "          0.17727678,  0.1520003 ,  0.18750675, -0.29703224,  0.20185754,\n",
              "         -0.09837657,  0.20752585, -0.02390084, -0.24997559, -0.12361479,\n",
              "          0.12883648,  0.29306912,  0.16372639,  0.09937178],\n",
              "        [-0.23991117, -0.2651923 ,  0.23903784, -0.22173734,  0.19565393,\n",
              "          0.29388314, -0.3067137 , -0.20238128,  0.03470522, -0.27673158,\n",
              "          0.10671121, -0.02797063,  0.24522418, -0.29032373,  0.0868616 ,\n",
              "         -0.28763855, -0.29240856,  0.2513287 ,  0.18060988,  0.00385525,\n",
              "         -0.10595958, -0.12271954, -0.28956935,  0.20349982, -0.253441  ,\n",
              "         -0.1392269 , -0.26641092, -0.08847743,  0.22613223, -0.1474035 ,\n",
              "         -0.2670272 , -0.03030261, -0.12645999,  0.09397569, -0.11134112,\n",
              "          0.09023276, -0.26409036, -0.12146127,  0.23114555, -0.24868488,\n",
              "         -0.09325724, -0.2673173 ,  0.10471647, -0.26785162,  0.15487573,\n",
              "          0.00434515,  0.01528415, -0.04804516,  0.07436057, -0.2771627 ,\n",
              "          0.00906355, -0.0499514 ,  0.08763499,  0.1274384 ,  0.00661903,\n",
              "          0.06569868,  0.24266143, -0.13062564,  0.19046807,  0.00747466,\n",
              "         -0.19054511, -0.03220633, -0.29429793, -0.13276805],\n",
              "        [-0.04142335,  0.08830774, -0.3699135 ,  0.09755793,  0.10515572,\n",
              "         -0.17946318,  0.1547285 , -0.16988781,  0.13360322, -0.13190976,\n",
              "         -0.20959423, -0.288524  , -0.38286117, -0.00845579,  0.13680379,\n",
              "          0.01454079, -0.12751737, -0.01172149,  0.07581013, -0.00570601,\n",
              "         -0.27637577, -0.0463022 , -0.1129355 ,  0.32564947, -0.14709276,\n",
              "          0.24794263, -0.03504458, -0.11189981,  0.1039176 ,  0.01880951,\n",
              "          0.19383153, -0.22736454, -0.28960964,  0.18615702,  0.31934848,\n",
              "          0.13851504,  0.08787164, -0.02029648,  0.16369863, -0.23726529,\n",
              "          0.23427892, -0.1779191 ,  0.26322114, -0.11801088, -0.24689758,\n",
              "         -0.23008853,  0.25092185, -0.02453297, -0.01336099,  0.3640629 ,\n",
              "         -0.08117305, -0.10514396,  0.06681529, -0.01245108,  0.19716606,\n",
              "         -0.03379083, -0.23635021, -0.25163153,  0.16301039, -0.06223419,\n",
              "         -0.26331273, -0.04240328,  0.1839556 ,  0.02520414],\n",
              "        [ 0.29675162, -0.01497975, -0.0598937 ,  0.15868579, -0.2709679 ,\n",
              "         -0.25780836,  0.18955766,  0.22955954,  0.10147751,  0.11574405,\n",
              "         -0.16138442,  0.00556448, -0.12419144,  0.2179568 , -0.333493  ,\n",
              "         -0.12205715,  0.13199565, -0.40463543, -0.011237  ,  0.17551629,\n",
              "         -0.09781829, -0.25073934, -0.02611271,  0.3700922 , -0.04150075,\n",
              "         -0.11234832, -0.0039562 ,  0.26111734, -0.04606202,  0.1616808 ,\n",
              "         -0.13642581,  0.16873449, -0.0535949 ,  0.1341228 ,  0.22445513,\n",
              "         -0.01029954,  0.31392777,  0.04837236, -0.2576316 , -0.04262391,\n",
              "          0.3437531 ,  0.24578851,  0.29029977,  0.18807665, -0.08349811,\n",
              "         -0.14906788, -0.21879981,  0.2901345 ,  0.15894718,  0.25790578,\n",
              "         -0.36278033, -0.003935  , -0.04728084,  0.14755005, -0.13528381,\n",
              "          0.1830813 , -0.38985345,  0.17536288, -0.04464704,  0.02867523,\n",
              "         -0.07119574, -0.10425491,  0.37797186,  0.25327146]],\n",
              "       dtype=float32),\n",
              " array([ 0.        ,  0.        ,  0.07072656, -0.0828547 , -0.01188974,\n",
              "         0.12125116, -0.04546914,  0.        , -0.15672636,  0.        ,\n",
              "         0.        ,  0.13320886,  0.04231279, -0.10457402,  0.04753666,\n",
              "         0.        ,  0.        ,  0.09964358,  0.        , -0.08522082,\n",
              "         0.        ,  0.        ,  0.        , -0.04823308,  0.        ,\n",
              "         0.        ,  0.        ,  0.        , -0.07340725, -0.06851622,\n",
              "        -0.05611772,  0.        ,  0.05556749, -0.08536296, -0.06437982,\n",
              "        -0.09958909, -0.07481578,  0.        , -0.03054872,  0.        ,\n",
              "        -0.06758267,  0.        , -0.07374158,  0.        , -0.02827443,\n",
              "         0.        ,  0.        ,  0.        , -0.08013327,  0.00701076,\n",
              "         0.09329525,  0.01882144, -0.02472069,  0.        , -0.04065948,\n",
              "         0.        ,  0.05629488,  0.        ,  0.        ,  0.        ,\n",
              "         0.        ,  0.03065124, -0.14398548, -0.04802423], dtype=float32),\n",
              " array([[ 0.11248183, -0.18393666, -0.23833317, ..., -0.17133659,\n",
              "          0.00681359, -0.18133223],\n",
              "        [-0.0505017 ,  0.08341688,  0.15268338, ...,  0.09786439,\n",
              "         -0.00126523, -0.01706159],\n",
              "        [-0.0937883 , -0.36025548, -0.03827614, ..., -0.22209859,\n",
              "          0.3470135 ,  0.10935966],\n",
              "        ...,\n",
              "        [ 0.22555424, -0.14856054,  0.18572229, ..., -0.16738884,\n",
              "          0.18176126, -0.1768569 ],\n",
              "        [-0.03430035,  0.06438896, -0.22570942, ..., -0.09367175,\n",
              "         -0.2346227 ,  0.1469156 ],\n",
              "        [ 0.4177367 , -0.00692555, -0.13377169, ..., -0.14770408,\n",
              "         -0.24575977,  0.17766087]], dtype=float32),\n",
              " array([ 0.08131844, -0.05446363,  0.01256962,  0.07616282, -0.06384668,\n",
              "         0.0907747 ,  0.07428314, -0.02884414,  0.11433446,  0.09890234,\n",
              "         0.14632998,  0.05339229, -0.04768599, -0.00235387,  0.07074977,\n",
              "         0.00120164, -0.00946682,  0.15281779, -0.0086075 ,  0.03821617,\n",
              "         0.01080199, -0.08233112,  0.05176098, -0.04177993, -0.06285021,\n",
              "        -0.01206285,  0.03576443,  0.01596363,  0.05651342, -0.02662494,\n",
              "         0.06207152, -0.03967162], dtype=float32),\n",
              " array([[ 0.30739075, -0.30184844,  0.2796789 ],\n",
              "        [ 0.41550526, -0.01091252, -0.40743014],\n",
              "        [-0.00395871,  0.19942045, -0.31347033],\n",
              "        [-0.01507508, -0.37003776,  0.27120256],\n",
              "        [ 0.44695053, -0.26660258, -0.07861991],\n",
              "        [-0.13510753,  0.39677155,  0.4388597 ],\n",
              "        [-0.26974198,  0.35265583, -0.4619305 ],\n",
              "        [ 0.28460464, -0.5983562 , -0.16596927],\n",
              "        [-0.41072753,  0.36763355,  0.28418213],\n",
              "        [-0.5542822 ,  0.4950344 , -0.4512426 ],\n",
              "        [-0.15837857,  0.48561677,  0.34847647],\n",
              "        [-0.3344623 , -0.4254768 , -0.11670851],\n",
              "        [-0.09852629, -0.04580744,  0.3695646 ],\n",
              "        [-0.067922  ,  0.02534706,  0.08875627],\n",
              "        [-0.46358857,  0.145358  ,  0.3262016 ],\n",
              "        [ 0.07043042, -0.69739103,  0.02308412],\n",
              "        [-0.02930449,  0.02593871,  0.27266702],\n",
              "        [-0.32491127,  0.1247346 ,  0.06925768],\n",
              "        [ 0.05987501, -0.63788456, -0.07971676],\n",
              "        [-0.00374913, -0.5595237 ,  0.15666458],\n",
              "        [ 0.09381061, -0.5567702 , -0.13493451],\n",
              "        [ 0.4582494 , -0.19371301, -0.17021859],\n",
              "        [-0.3871142 ,  0.21744047, -0.40445518],\n",
              "        [ 0.36608523, -0.7190112 , -0.04018861],\n",
              "        [-0.28978848,  0.002149  , -0.0286036 ],\n",
              "        [ 0.19492827, -0.15215927,  0.36779132],\n",
              "        [ 0.20249084, -0.17960511,  0.23261702],\n",
              "        [ 0.24670948, -0.00176411, -0.257367  ],\n",
              "        [ 0.16338465,  0.08310977, -0.33604813],\n",
              "        [-0.33226937,  0.36534652, -0.30071625],\n",
              "        [-0.16643086,  0.5619994 , -0.4271915 ],\n",
              "        [ 0.14026146,  0.15943381,  0.31322226]], dtype=float32),\n",
              " array([-0.08775289,  0.00280645,  0.06349376], dtype=float32)]"
            ]
          },
          "metadata": {},
          "execution_count": 159
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": true,
        "id": "nhPK4V6nLtVx",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e6d50d2e-7be8-4793-b8cf-2670705689e4"
      },
      "source": [
        "model.save(dir+\"dnn_iris.h5\")\n",
        "print(\"Saved model to disk.\")\n",
        "\n",
        "#load and evaluate the saved model\n",
        "from numpy import loadtxt\n",
        "from tensorflow.python.keras.models import load_model\n",
        "\n",
        "#load model\n",
        "loaded_model = load_model(dir+\"dnn_iris.h5\")\n",
        "model.summary()\n",
        "\n",
        "score = model.evaluate(X_test, y_test)\n",
        "print('test_loss: ', score[0])\n",
        "print('test_acc: ', score[1])"
      ],
      "execution_count": 160,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Saved model to disk.\n",
            "Model: \"sequential_11\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_35 (Dense)             (None, 64)                320       \n",
            "_________________________________________________________________\n",
            "activation_35 (Activation)   (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dropout_24 (Dropout)         (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dense_36 (Dense)             (None, 32)                2080      \n",
            "_________________________________________________________________\n",
            "activation_36 (Activation)   (None, 32)                0         \n",
            "_________________________________________________________________\n",
            "dropout_25 (Dropout)         (None, 32)                0         \n",
            "_________________________________________________________________\n",
            "dense_37 (Dense)             (None, 3)                 99        \n",
            "_________________________________________________________________\n",
            "activation_37 (Activation)   (None, 3)                 0         \n",
            "=================================================================\n",
            "Total params: 2,499\n",
            "Trainable params: 2,499\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "2/2 [==============================] - 0s 7ms/step - loss: 0.1037 - accuracy: 1.0000\n",
            "test_loss:  0.10372401028871536\n",
            "test_acc:  1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X8sQr21XLtVy",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a50b9c63-9912-441d-f402-49c130eb6cc8"
      },
      "source": [
        "# X_test의 예측 클래스 확인하기\n",
        "y_prob = model.predict(X_test)  # output node의 출력값 확인하기\n",
        "print(y_prob)\n",
        "\n",
        "y_class = y_prob.argmax(axis=-1)  # output node의 예측값(class) 확인하기\n",
        "y_class"
      ],
      "execution_count": 161,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[9.81357753e-01 2.29901964e-08 1.86421946e-02]\n",
            " [9.51393962e-01 3.57366190e-07 4.86057475e-02]\n",
            " [9.13904700e-03 6.23328052e-03 9.84627664e-01]\n",
            " [9.64121580e-01 3.11075006e-08 3.58783416e-02]\n",
            " [3.14289661e-09 9.99704659e-01 2.95279839e-04]\n",
            " [2.04665840e-09 9.99772727e-01 2.27311757e-04]\n",
            " [4.74362077e-11 9.99953985e-01 4.60340598e-05]\n",
            " [8.84346664e-01 2.43018076e-06 1.15650915e-01]\n",
            " [9.18929815e-01 5.46260310e-07 8.10695961e-02]\n",
            " [1.15339205e-01 1.65159494e-04 8.84495616e-01]\n",
            " [1.02999864e-09 9.99811232e-01 1.88807011e-04]\n",
            " [2.44697556e-02 8.51850957e-04 9.74678457e-01]\n",
            " [1.12457890e-10 9.99932528e-01 6.74770345e-05]\n",
            " [3.07105985e-02 1.40851131e-04 9.69148576e-01]\n",
            " [8.03623617e-01 9.79894139e-07 1.96375445e-01]\n",
            " [9.77223873e-01 2.44088881e-07 2.27759462e-02]\n",
            " [5.98061644e-03 1.23963389e-03 9.92779791e-01]\n",
            " [1.33400690e-02 1.04365894e-03 9.85616326e-01]\n",
            " [2.20279639e-09 9.99742687e-01 2.57250096e-04]\n",
            " [1.23006068e-08 9.99426126e-01 5.73924801e-04]\n",
            " [5.16862908e-09 9.99614358e-01 3.85665247e-04]\n",
            " [9.82282817e-01 4.29492069e-08 1.77171268e-02]\n",
            " [8.66566122e-01 1.48163122e-06 1.33432403e-01]\n",
            " [9.26411986e-01 3.88487564e-07 7.35875890e-02]\n",
            " [4.09893772e-08 9.99097824e-01 9.02171538e-04]\n",
            " [6.02270663e-03 5.28137607e-04 9.93449152e-01]\n",
            " [9.95209038e-01 5.94901128e-10 4.79096826e-03]\n",
            " [1.88712571e-02 1.14505005e-03 9.79983628e-01]\n",
            " [1.22502730e-10 9.99927163e-01 7.28690648e-05]\n",
            " [3.29821408e-01 1.65563368e-04 6.70013011e-01]\n",
            " [1.66574091e-01 2.98822968e-04 8.33127081e-01]\n",
            " [4.98471650e-07 9.95911956e-01 4.08758875e-03]\n",
            " [4.99788877e-10 9.99876380e-01 1.23547739e-04]\n",
            " [5.93397021e-01 1.77546899e-05 4.06585187e-01]\n",
            " [4.86188263e-01 2.37472232e-05 5.13788044e-01]\n",
            " [9.48816895e-01 4.78528790e-08 5.11830747e-02]\n",
            " [5.88383079e-01 2.40843583e-05 4.11592871e-01]\n",
            " [9.66841042e-01 4.32654957e-07 3.31584215e-02]\n",
            " [5.57089210e-01 1.49634457e-06 4.42909300e-01]\n",
            " [9.86064553e-01 7.75290516e-08 1.39353303e-02]\n",
            " [1.07876170e-08 9.99436319e-01 5.63712150e-04]\n",
            " [4.48334077e-03 3.61565477e-03 9.91900980e-01]\n",
            " [9.35753584e-01 3.39019920e-07 6.42460585e-02]\n",
            " [2.35544425e-02 1.42283941e-04 9.76303339e-01]\n",
            " [1.04539017e-08 9.99430597e-01 5.69387455e-04]\n",
            " [1.31226435e-01 2.71129014e-04 8.68502438e-01]\n",
            " [8.15125406e-01 1.52539667e-06 1.84873000e-01]\n",
            " [5.47386169e-01 4.52041650e-06 4.52609330e-01]\n",
            " [1.52711941e-08 9.99345720e-01 6.54299336e-04]]\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0, 0, 2, 0, 1, 1, 1, 0, 0, 2, 1, 2, 1, 2, 0, 0, 2, 2, 1, 1, 1, 0,\n",
              "       0, 0, 1, 2, 0, 2, 1, 2, 2, 1, 1, 0, 2, 0, 0, 0, 0, 0, 1, 2, 0, 2,\n",
              "       1, 2, 0, 0, 1])"
            ]
          },
          "metadata": {},
          "execution_count": 161
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "34Xez1yYLtV0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f8e3657d-4387-4508-f8e1-df18979a57c0"
      },
      "source": [
        "# 새로운 샘플의 예측 클래스 확인하기\n",
        "X_new = [5.7, 3.8, 1.7, 0.3]\n",
        "\n",
        "print(X_new)\n",
        "\n",
        "y_prob = model.predict([X_new])\n",
        "y_pred = y_prob.argmax()\n",
        "print(y_prob, y_pred)"
      ],
      "execution_count": 169,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[5.7, 3.8, 1.7, 0.3]\n",
            "[[7.2277356e-10 9.9982733e-01 1.7267295e-04]] 1\n"
          ]
        }
      ]
    }
  ]
}