iris(DNN-tensorflow2.0)-Functional.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1) 載入資料集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>SepalLengthCm</th>\n",
       "      <th>SepalWidthCm</th>\n",
       "      <th>PetalLengthCm</th>\n",
       "      <th>PetalWidthCm</th>\n",
       "      <th>Species</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</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",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.7</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</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>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>3.6</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>6.7</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.2</td>\n",
       "      <td>2.3</td>\n",
       "      <td>Iris-virginica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>146</th>\n",
       "      <td>6.3</td>\n",
       "      <td>2.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.9</td>\n",
       "      <td>Iris-virginica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>147</th>\n",
       "      <td>6.5</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.2</td>\n",
       "      <td>2.0</td>\n",
       "      <td>Iris-virginica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>148</th>\n",
       "      <td>6.2</td>\n",
       "      <td>3.4</td>\n",
       "      <td>5.4</td>\n",
       "      <td>2.3</td>\n",
       "      <td>Iris-virginica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>149</th>\n",
       "      <td>5.9</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.1</td>\n",
       "      <td>1.8</td>\n",
       "      <td>Iris-virginica</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>150 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     SepalLengthCm  SepalWidthCm  PetalLengthCm  PetalWidthCm         Species\n",
       "0              5.1           3.5            1.4           0.2     Iris-setosa\n",
       "1              4.9           3.0            1.4           0.2     Iris-setosa\n",
       "2              4.7           3.2            1.3           0.2     Iris-setosa\n",
       "3              4.6           3.1            1.5           0.2     Iris-setosa\n",
       "4              5.0           3.6            1.4           0.2     Iris-setosa\n",
       "..             ...           ...            ...           ...             ...\n",
       "145            6.7           3.0            5.2           2.3  Iris-virginica\n",
       "146            6.3           2.5            5.0           1.9  Iris-virginica\n",
       "147            6.5           3.0            5.2           2.0  Iris-virginica\n",
       "148            6.2           3.4            5.4           2.3  Iris-virginica\n",
       "149            5.9           3.0            5.1           1.8  Iris-virginica\n",
       "\n",
       "[150 rows x 5 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train = pd.read_csv('./data/Iris.csv')\n",
    "df_train = df_train.drop(labels=['Id'],axis=1) # 移除Id\n",
    "df_train"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2) 手動編碼\n",
    "處理名目資料 (Nominal variables) - 資料前處理\n",
    "依據特徵資料的特性，可以選擇手動編碼或自動編碼。\n",
    "\n",
    "### 使用編碼時機?\n",
    "進行深度學習時，神經網路只能處理數值資料。因此我們需要將所有非數字型態的特徵進行轉換。\n",
    "\n",
    "ex:\n",
    "\n",
    "|  Iris-setosa |  Iris-versicolor | Iris-virginica  |\n",
    "|:---:|:---:|:---:|\n",
    "|  1 | 2  | 3  |"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "label_map = {'Iris-setosa': 0, 'Iris-versicolor': 1, 'Iris-virginica': 2}\n",
    "\n",
    "#將編碼後的label map存至df_train['Species']中。\n",
    "df_train['Class'] = df_train['Species'].map(label_map)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>SepalLengthCm</th>\n",
       "      <th>SepalWidthCm</th>\n",
       "      <th>PetalLengthCm</th>\n",
       "      <th>PetalWidthCm</th>\n",
       "      <th>Species</th>\n",
       "      <th>Class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</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",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.7</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</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",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>145</th>\n",
       "      <td>6.7</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.2</td>\n",
       "      <td>2.3</td>\n",
       "      <td>Iris-virginica</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>146</th>\n",
       "      <td>6.3</td>\n",
       "      <td>2.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.9</td>\n",
       "      <td>Iris-virginica</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>147</th>\n",
       "      <td>6.5</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.2</td>\n",
       "      <td>2.0</td>\n",
       "      <td>Iris-virginica</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>148</th>\n",
       "      <td>6.2</td>\n",
       "      <td>3.4</td>\n",
       "      <td>5.4</td>\n",
       "      <td>2.3</td>\n",
       "      <td>Iris-virginica</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>149</th>\n",
       "      <td>5.9</td>\n",
       "      <td>3.0</td>\n",
       "      <td>5.1</td>\n",
       "      <td>1.8</td>\n",
       "      <td>Iris-virginica</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>150 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     SepalLengthCm  SepalWidthCm  PetalLengthCm  PetalWidthCm         Species  \\\n",
       "0              5.1           3.5            1.4           0.2     Iris-setosa   \n",
       "1              4.9           3.0            1.4           0.2     Iris-setosa   \n",
       "2              4.7           3.2            1.3           0.2     Iris-setosa   \n",
       "3              4.6           3.1            1.5           0.2     Iris-setosa   \n",
       "4              5.0           3.6            1.4           0.2     Iris-setosa   \n",
       "..             ...           ...            ...           ...             ...   \n",
       "145            6.7           3.0            5.2           2.3  Iris-virginica   \n",
       "146            6.3           2.5            5.0           1.9  Iris-virginica   \n",
       "147            6.5           3.0            5.2           2.0  Iris-virginica   \n",
       "148            6.2           3.4            5.4           2.3  Iris-virginica   \n",
       "149            5.9           3.0            5.1           1.8  Iris-virginica   \n",
       "\n",
       "     Class  \n",
       "0        0  \n",
       "1        0  \n",
       "2        0  \n",
       "3        0  \n",
       "4        0  \n",
       "..     ...  \n",
       "145      2  \n",
       "146      2  \n",
       "147      2  \n",
       "148      2  \n",
       "149      2  \n",
       "\n",
       "[150 rows x 6 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3) 檢查缺失值\n",
    "使用 numpy 所提供的函式來檢查是否有 NA 缺失值，假設有缺失值使用dropna()來移除。使用的時機在於當只有少量的缺失值適用，若遇到有大量缺失值的情況，或是本身的資料量就很少的情況下建議可以透過機器學習的方法補值來預測缺失值。\n",
    "\n",
    "```python\n",
    "# 移除缺失值\n",
    "train=train.dropna()\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "checked missing data(NAN mount): 0\n"
     ]
    }
   ],
   "source": [
    "df_train = df_train.drop(labels=['Species'],axis=1) # 移除Species\n",
    "# checked missing data\n",
    "print(\"checked missing data(NAN mount):\",len(np.where(np.isnan(df_train))[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4) 將輸出特徵做one-hot encoding\n",
    "如何對欄位class做one-hot encoding？\n",
    "#### class 的值有0,1,2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.]], dtype=float32)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tensorflow.keras.utils import to_categorical\n",
    "y = to_categorical(df_train['Class'])\n",
    "y[0:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4) 切割訓練集與測試集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "X=df_train.drop(labels=['Class'],axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train , X_test , y_train , y_test = train_test_split(X,y , test_size=.3 , random_state=40)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "訓練資料:  (105, 4)\n",
      "測試資料:  (45, 4)\n"
     ]
    }
   ],
   "source": [
    "print('訓練資料: ',X_train.shape)\n",
    "print('測試資料: ',X_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5) 建立網路模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.2.0\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 此範例使用 Tensorflow2.0 Functional API 支援多個輸入、多個輸出，利用傳遞的方式搭建神經網路。\n",
    "def build_model():\n",
    "    model_input = layers.Input(shape=X.shape[-1])\n",
    "    x = layers.Dense(8,activation='relu')(model_input)\n",
    "    x = layers.Dense(16,activation='relu')(x)\n",
    "    model_output = layers.Dense(3,activation='softmax')(x)\n",
    "    \n",
    "    return Model(model_input ,model_output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         [(None, 4)]               0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 8)                 40        \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 16)                144       \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 3)                 51        \n",
      "=================================================================\n",
      "Total params: 235\n",
      "Trainable params: 235\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = build_model()\n",
    "model.summary()   # Weights = (4+1)*8+(8+1)*16+(16+1)*3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "94/94 [==============================] - 0s 4ms/step - loss: 1.3455 - acc: 0.1596 - val_loss: 0.9809 - val_acc: 0.9091\n",
      "Epoch 2/50\n",
      "94/94 [==============================] - 0s 952us/step - loss: 0.9904 - acc: 0.6809 - val_loss: 0.9380 - val_acc: 0.6364\n",
      "Epoch 3/50\n",
      "94/94 [==============================] - 0s 947us/step - loss: 0.8287 - acc: 0.6809 - val_loss: 0.9234 - val_acc: 0.3636\n",
      "Epoch 4/50\n",
      "94/94 [==============================] - 0s 948us/step - loss: 0.6998 - acc: 0.7447 - val_loss: 0.7255 - val_acc: 0.9091\n",
      "Epoch 5/50\n",
      "94/94 [==============================] - 0s 945us/step - loss: 0.6044 - acc: 0.7660 - val_loss: 0.6585 - val_acc: 0.7273\n",
      "Epoch 6/50\n",
      "94/94 [==============================] - 0s 969us/step - loss: 0.5271 - acc: 0.8191 - val_loss: 0.6439 - val_acc: 0.5455\n",
      "Epoch 7/50\n",
      "94/94 [==============================] - 0s 945us/step - loss: 0.4565 - acc: 0.8511 - val_loss: 0.5327 - val_acc: 0.9091\n",
      "Epoch 8/50\n",
      "94/94 [==============================] - 0s 939us/step - loss: 0.4057 - acc: 0.8936 - val_loss: 0.5140 - val_acc: 0.9091\n",
      "Epoch 9/50\n",
      "94/94 [==============================] - 0s 948us/step - loss: 0.3630 - acc: 0.9468 - val_loss: 0.3745 - val_acc: 0.9091\n",
      "Epoch 10/50\n",
      "94/94 [==============================] - 0s 946us/step - loss: 0.3207 - acc: 0.9681 - val_loss: 0.3479 - val_acc: 0.9091\n",
      "Epoch 11/50\n",
      "94/94 [==============================] - 0s 946us/step - loss: 0.2829 - acc: 0.9362 - val_loss: 0.2919 - val_acc: 0.9091\n",
      "Epoch 12/50\n",
      "94/94 [==============================] - 0s 941us/step - loss: 0.2608 - acc: 0.9468 - val_loss: 0.2936 - val_acc: 0.9091\n",
      "Epoch 13/50\n",
      "94/94 [==============================] - 0s 945us/step - loss: 0.2285 - acc: 0.9574 - val_loss: 0.2601 - val_acc: 0.9091\n",
      "Epoch 14/50\n",
      "94/94 [==============================] - 0s 983us/step - loss: 0.2009 - acc: 0.9574 - val_loss: 0.3159 - val_acc: 0.9091\n",
      "Epoch 15/50\n",
      "94/94 [==============================] - 0s 944us/step - loss: 0.1956 - acc: 0.9574 - val_loss: 0.2492 - val_acc: 0.9091\n",
      "Epoch 16/50\n",
      "94/94 [==============================] - 0s 936us/step - loss: 0.1781 - acc: 0.9468 - val_loss: 0.1649 - val_acc: 0.9091\n",
      "Epoch 17/50\n",
      "94/94 [==============================] - 0s 942us/step - loss: 0.1688 - acc: 0.9468 - val_loss: 0.1474 - val_acc: 0.9091\n",
      "Epoch 18/50\n",
      "94/94 [==============================] - 0s 943us/step - loss: 0.1458 - acc: 0.9681 - val_loss: 0.2311 - val_acc: 0.9091\n",
      "Epoch 19/50\n",
      "94/94 [==============================] - 0s 943us/step - loss: 0.1548 - acc: 0.9574 - val_loss: 0.2051 - val_acc: 0.9091\n",
      "Epoch 20/50\n",
      "94/94 [==============================] - 0s 944us/step - loss: 0.1337 - acc: 0.9787 - val_loss: 0.1262 - val_acc: 0.9091\n",
      "Epoch 21/50\n",
      "94/94 [==============================] - 0s 975us/step - loss: 0.1308 - acc: 0.9468 - val_loss: 0.1514 - val_acc: 0.9091\n",
      "Epoch 22/50\n",
      "94/94 [==============================] - 0s 943us/step - loss: 0.1152 - acc: 0.9681 - val_loss: 0.0679 - val_acc: 1.0000\n",
      "Epoch 23/50\n",
      "94/94 [==============================] - 0s 979us/step - loss: 0.1187 - acc: 0.9574 - val_loss: 0.1127 - val_acc: 0.9091\n",
      "Epoch 24/50\n",
      "94/94 [==============================] - 0s 933us/step - loss: 0.1362 - acc: 0.9362 - val_loss: 0.0761 - val_acc: 1.0000\n",
      "Epoch 25/50\n",
      "94/94 [==============================] - 0s 929us/step - loss: 0.1042 - acc: 0.9787 - val_loss: 0.2215 - val_acc: 0.9091\n",
      "Epoch 26/50\n",
      "94/94 [==============================] - 0s 934us/step - loss: 0.1131 - acc: 0.9574 - val_loss: 0.1014 - val_acc: 0.9091\n",
      "Epoch 27/50\n",
      "94/94 [==============================] - 0s 941us/step - loss: 0.1072 - acc: 0.9681 - val_loss: 0.0908 - val_acc: 1.0000\n",
      "Epoch 28/50\n",
      "94/94 [==============================] - 0s 943us/step - loss: 0.1187 - acc: 0.9574 - val_loss: 0.0774 - val_acc: 1.0000\n",
      "Epoch 29/50\n",
      "94/94 [==============================] - 0s 934us/step - loss: 0.1152 - acc: 0.9574 - val_loss: 0.1387 - val_acc: 0.9091\n",
      "Epoch 30/50\n",
      "94/94 [==============================] - 0s 942us/step - loss: 0.1041 - acc: 0.9681 - val_loss: 0.1060 - val_acc: 0.9091\n",
      "Epoch 31/50\n",
      "94/94 [==============================] - 0s 945us/step - loss: 0.1018 - acc: 0.9574 - val_loss: 0.1428 - val_acc: 0.9091\n",
      "Epoch 32/50\n",
      "94/94 [==============================] - 0s 977us/step - loss: 0.1048 - acc: 0.9681 - val_loss: 0.0823 - val_acc: 1.0000\n",
      "Epoch 33/50\n",
      "94/94 [==============================] - 0s 981us/step - loss: 0.1050 - acc: 0.9574 - val_loss: 0.1652 - val_acc: 0.9091\n",
      "Epoch 34/50\n",
      "94/94 [==============================] - 0s 934us/step - loss: 0.1007 - acc: 0.9681 - val_loss: 0.1171 - val_acc: 0.9091\n",
      "Epoch 35/50\n",
      "94/94 [==============================] - 0s 930us/step - loss: 0.0945 - acc: 0.9574 - val_loss: 0.0555 - val_acc: 1.0000\n",
      "Epoch 36/50\n",
      "94/94 [==============================] - 0s 944us/step - loss: 0.0934 - acc: 0.9574 - val_loss: 0.0468 - val_acc: 1.0000\n",
      "Epoch 37/50\n",
      "94/94 [==============================] - 0s 945us/step - loss: 0.1433 - acc: 0.9574 - val_loss: 0.1125 - val_acc: 0.9091\n",
      "Epoch 38/50\n",
      "94/94 [==============================] - 0s 941us/step - loss: 0.0944 - acc: 0.9681 - val_loss: 0.1589 - val_acc: 0.9091\n",
      "Epoch 39/50\n",
      "94/94 [==============================] - 0s 945us/step - loss: 0.1015 - acc: 0.9574 - val_loss: 0.1593 - val_acc: 0.9091\n",
      "Epoch 40/50\n",
      "94/94 [==============================] - 0s 943us/step - loss: 0.1084 - acc: 0.9574 - val_loss: 0.0947 - val_acc: 0.9091\n",
      "Epoch 41/50\n",
      "94/94 [==============================] - 0s 942us/step - loss: 0.0971 - acc: 0.9681 - val_loss: 0.1331 - val_acc: 0.9091\n",
      "Epoch 42/50\n",
      "94/94 [==============================] - 0s 938us/step - loss: 0.0920 - acc: 0.9681 - val_loss: 0.1495 - val_acc: 0.9091\n",
      "Epoch 43/50\n",
      "94/94 [==============================] - 0s 994us/step - loss: 0.0937 - acc: 0.9787 - val_loss: 0.0293 - val_acc: 1.0000\n",
      "Epoch 44/50\n",
      "94/94 [==============================] - 0s 935us/step - loss: 0.1302 - acc: 0.9574 - val_loss: 0.1105 - val_acc: 0.9091\n",
      "Epoch 45/50\n",
      "94/94 [==============================] - 0s 935us/step - loss: 0.0871 - acc: 0.9574 - val_loss: 0.1815 - val_acc: 0.9091\n",
      "Epoch 46/50\n",
      "94/94 [==============================] - 0s 938us/step - loss: 0.0866 - acc: 0.9681 - val_loss: 0.0386 - val_acc: 1.0000\n",
      "Epoch 47/50\n",
      "94/94 [==============================] - 0s 944us/step - loss: 0.0771 - acc: 0.9681 - val_loss: 0.1837 - val_acc: 0.9091\n",
      "Epoch 48/50\n",
      "94/94 [==============================] - 0s 945us/step - loss: 0.0847 - acc: 0.9681 - val_loss: 0.0567 - val_acc: 1.0000\n",
      "Epoch 49/50\n",
      "94/94 [==============================] - 0s 938us/step - loss: 0.0834 - acc: 0.9787 - val_loss: 0.1952 - val_acc: 0.9091\n",
      "Epoch 50/50\n",
      "94/94 [==============================] - 0s 942us/step - loss: 0.0918 - acc: 0.9574 - val_loss: 0.2255 - val_acc: 0.9091\n"
     ]
    }
   ],
   "source": [
    "# 編譯模型\n",
    "optim = Adam(lr=0.001)\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer=optim,\n",
    "              metrics=['acc'])\n",
    "\n",
    "batch_size=1\n",
    "epochs = 50\n",
    "\n",
    "# 訓練模型\n",
    "history = model.fit(X_train, y_train,\n",
    "                    batch_size=batch_size,\n",
    "                    epochs=epochs,\n",
    "                    verbose=1,\n",
    "                    shuffle=True,\n",
    "                    validation_split=0.1)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6) 觀察訓練結果"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "history_dict = history.history\n",
    "history_dict.keys()\n",
    "acc = history_dict['acc']\n",
    "val_acc = history_dict['val_acc']\n",
    "loss = history_dict['loss']\n",
    "val_loss = history_dict['val_loss']\n",
    "epochs_ = range(1,len(acc)+1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(epochs_ , loss , label = 'training loss')\n",
    "plt.plot(epochs_ , val_loss , label = 'val los')\n",
    "plt.title('training and val loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.clf()\n",
    "plt.plot(epochs_ , acc , label='train accuracy')\n",
    "plt.plot(epochs_ , val_acc , label = 'val accuracy')\n",
    "plt.title('train and val acc')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('acc')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 測試"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.96\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "label=np.argmax(y,axis=1)\n",
    "pred =  np.argmax(model.predict(X), axis=1)\n",
    "print(accuracy_score(label, pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}