手写数字预测的RNN模型搭建_与训练和评估_LSTM实现.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "d4c08f48-fe23-4ddb-ac46-d97f05397514",
    "_uuid": "f2156d1dd26a1243e18512002e10872c5bd7271e"
   },
   "source": [
    "* **1. Introduction**\n",
    "* **2. Data preparation**\n",
    "    * 2.1 Load data\n",
    "    * 2.2 Check for null and missing values\n",
    "    * 2.3 Normalization\n",
    "    * 2.4 Reshape\n",
    "    * 2.5 Label encoding\n",
    "    * 2.6 Split training and valdiation set\n",
    "* **3. CNN**\n",
    "\n",
    "* **4. Evaluate the model**\n",
    "\n",
    "* **5. Prediction and submition**\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "eb88b372-a6e5-40c8-a1c6-c03799165490",
    "_uuid": "e9aff3cf1bb8daa73bec67b970d12195677679f3"
   },
   "source": [
    "# 1. Introduction\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "_cell_guid": "f67b9393-8ea1-4e23-b856-2ce149cfe421",
    "_execution_state": "idle",
    "_uuid": "72334cb006d02a4bcfc2a2fe622524eba824c6f8"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.image as mpimg\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import itertools\n",
    "\n",
    "from keras.utils.np_utils import to_categorical # convert to one-hot-encoding\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D\n",
    "from keras.optimizers import RMSprop, Adam, SGD\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.callbacks import ReduceLROnPlateau\n",
    "import keras\n",
    "\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.callbacks import ReduceLROnPlateau\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "sns.set(style='white', context='notebook', palette='deep')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "6d2fb3e6-ab71-4974-b5a2-4af1ebdb99f4",
    "_execution_state": "idle",
    "_uuid": "86061d98eccaa02efe0dab0fa3884e71fcf4c310"
   },
   "source": [
    "# 2. Data preparation\n",
    "## 2.1 Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "_cell_guid": "5e51d00e-62fd-4141-bf73-50ac4f2da7d0",
    "_execution_state": "idle",
    "_uuid": "84bbd5ab8d7895bd430d5ecfe2f7ddf77baa7b74"
   },
   "outputs": [],
   "source": [
    "# >>>>>填写<<<< 利用pandas的load_csv函数，读取我们的train 和 test数据集合 变量已经给出 >>>>>填写<<<< ######\n",
    "train = pd.read_csv(\"subset_train.csv\")\n",
    "test = pd.read_csv(\"Small_test.csv\")\n",
    "#####train validation test(完全独立的，与训练过程无关的)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "_cell_guid": "86570a36-5c20-460a-9dfd-2070548532a7",
    "_execution_state": "idle",
    "_uuid": "1213b979d5ed3e0d13824d17d694c79d2ece92fa"
   },
   "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>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>pixel9</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4195</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4196</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4197</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4198</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4199</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4200 rows × 784 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  pixel8  \\\n",
       "0          0       0       0       0       0       0       0       0       0   \n",
       "1          0       0       0       0       0       0       0       0       0   \n",
       "2          0       0       0       0       0       0       0       0       0   \n",
       "3          0       0       0       0       0       0       0       0       0   \n",
       "4          0       0       0       0       0       0       0       0       0   \n",
       "...      ...     ...     ...     ...     ...     ...     ...     ...     ...   \n",
       "4195       0       0       0       0       0       0       0       0       0   \n",
       "4196       0       0       0       0       0       0       0       0       0   \n",
       "4197       0       0       0       0       0       0       0       0       0   \n",
       "4198       0       0       0       0       0       0       0       0       0   \n",
       "4199       0       0       0       0       0       0       0       0       0   \n",
       "\n",
       "      pixel9  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n",
       "0          0  ...         0         0         0         0         0         0   \n",
       "1          0  ...         0         0         0         0         0         0   \n",
       "2          0  ...         0         0         0         0         0         0   \n",
       "3          0  ...         0         0         0         0         0         0   \n",
       "4          0  ...         0         0         0         0         0         0   \n",
       "...      ...  ...       ...       ...       ...       ...       ...       ...   \n",
       "4195       0  ...         0         0         0         0         0         0   \n",
       "4196       0  ...         0         0         0         0         0         0   \n",
       "4197       0  ...         0         0         0         0         0         0   \n",
       "4198       0  ...         0         0         0         0         0         0   \n",
       "4199       0  ...         0         0         0         0         0         0   \n",
       "\n",
       "      pixel780  pixel781  pixel782  pixel783  \n",
       "0            0         0         0         0  \n",
       "1            0         0         0         0  \n",
       "2            0         0         0         0  \n",
       "3            0         0         0         0  \n",
       "4            0         0         0         0  \n",
       "...        ...       ...       ...       ...  \n",
       "4195         0         0         0         0  \n",
       "4196         0         0         0         0  \n",
       "4197         0         0         0         0  \n",
       "4198         0         0         0         0  \n",
       "4199         0         0         0         0  \n",
       "\n",
       "[4200 rows x 784 columns]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# >>>>>填写<<<< 利用pandas的header选择，将label列传递给Y_train >>>>>填写<<<< \n",
    "Y_train = train[\"label\"]\n",
    "\n",
    "Y_test = test['label']\n",
    "# 因为train.csv中，第一列label在上述代码已经传递给Y_label，这里对于x_train 我们不需要训练集的第一列 #####\n",
    "X_train = train.drop(labels = [\"label\"],axis = 1) \n",
    "X_test = test.drop(labels = [\"label\"],axis = 1)\n",
    "# 释放内存\n",
    "\n",
    "X_train\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "5aea4062-1790-4987-b739-c4bebd79030f",
    "_uuid": "b7b1b1d36243c885e57374c8b60c5a7e10abe922"
   },
   "source": [
    "## We have similar counts for the 10 digits."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "5b7d4b66-a140-4fcc-a889-bcef007c880a",
    "_uuid": "5d77934302869925c19128c77e247b3c8ca84d71"
   },
   "source": [
    "## 2.2 Check for null and missing values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "_cell_guid": "ececaa00-2ae3-4d13-b631-438df085b030",
    "_execution_state": "idle",
    "_uuid": "cdf27c27e2a5b15e6d7bfc70de7a18c08f3feb7a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count       784\n",
       "unique        1\n",
       "top       False\n",
       "freq        784\n",
       "dtype: object"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 检查训练数据是否有空值\n",
    "X_train.isnull().any().describe() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "_cell_guid": "92572e80-8543-4e72-8767-5c9be8381b04",
    "_execution_state": "idle",
    "_uuid": "a0089bb7ec9aec76373db475399aea24699ae989"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count       784\n",
       "unique        1\n",
       "top       False\n",
       "freq        784\n",
       "dtype: object"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# >>>>填写<<<< 检查训练数据是否有空值 >>>>填写<<<< ###\n",
    "X_test.isnull().any().describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "09d04cae-4245-4659-85dd-ef48531da295",
    "_uuid": "c0bee59691c2df0b275c78e38e7f9907d02ac038"
   },
   "source": [
    "I check for corrupted images (missing values inside).\n",
    "\n",
    "There is no missing values in the train and test dataset. So we can safely go ahead."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "6812040d-80ad-43d2-a571-275f4f20067b",
    "_uuid": "2954681f25f0dcbe986e6914396cdbce61db591f"
   },
   "source": [
    "## 2.3 Normalization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "159d5854-437a-4d0f-bc1e-fc3f7e43d178",
    "_uuid": "0ecf4b52510ab7957d0d4eb646c0aa1ba5986273"
   },
   "source": [
    "We perform a grayscale normalization to reduce the effect of illumination's differences. \n",
    "\n",
    "Moreover the CNN converg faster on [0..1] data than on [0..255].\n",
    "标准化，将灰度值 0-255 映射到0 - 1区间"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "_cell_guid": "cdc4340b-6e24-4e12-be99-ac806098ff17",
    "_execution_state": "idle",
    "_uuid": "b5d4f8fcf2a967e2c7d57daedf95aa8c5ab7f8cb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4200, 784)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Normalize the data\n",
    "X_train = X_train / 255.0\n",
    "###### >>>填写<<< 标准化测试集合 #######\n",
    "X_test = X_test / 255.0\n",
    "X_train.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "7413df94-bcb9-4f75-b174-c127d4445766",
    "_uuid": "a66741bf1ac597094f3a3166877008feef27c519"
   },
   "source": [
    "## 2.3 Reshape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "_cell_guid": "34b6a5f7-8fd2-4387-8ef4-c9dc19584fed",
    "_execution_state": "idle",
    "_uuid": "f0a6ad80dab8e0f2c2e46165ccd9cd82dd162bc3"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4200, 28, 28)"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# >>>>填写<<<<< 利用 reshape 函数， 将X_train变换成 (height = 28px, width = 28px , canal = 0) ######\n",
    "# CNN (batch=-1 取所有=4200, rows, cols, channels)\n",
    "X_train = X_train.values.reshape(-1,28,28)  # 训练集合是4200个，28*28 通道数为1的输入）\n",
    "                                            # （4200，28，28）\n",
    "#对于RNN 输入为3D 张量，尺寸为 (batch_size, timesteps, input_dim)。\n",
    "#X_train = X_train.values.reshape(-1,28,28,1)    \n",
    "X_test = X_test.values.reshape(-1,28,28) \n",
    "\n",
    "X_train.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "8decd1ce-7b7e-431d-8458-eaca18e0e1f7",
    "_uuid": "f4fb5553e188d9956f5d8b3a5d275ab00ea667ce"
   },
   "source": [
    "Train and test images (28px x 28px) has been stock into pandas.Dataframe as 1D vectors of 784 values. We reshape all data to 28x28x1 3D matrices. \n",
    "\n",
    "Keras requires an extra dimension in the end which correspond to channels. MNIST images are gray scaled so it use only one channel. For RGB images, there is 3 channels, we would have reshaped 784px vectors to 28x28x3 3D matrices. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "bdb422e2-bdec-444f-97a5-283a1e54bf2c",
    "_uuid": "39b7a31e843bac6b705461bcce89da216b91799e"
   },
   "source": [
    "## 2.5 Label encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "_cell_guid": "4b7f3e78-44dc-4561-b1f0-9429ee024cf4",
    "_execution_state": "idle",
    "_uuid": "cabefd1478d5c1bdfe57fd6a34395340916a854c"
   },
   "outputs": [],
   "source": [
    "# 利用0 1编码 将0-9数字标签编码成10维向量 (ex : 9 -> [0,0,0,0,0,0,0,0,0,1])\n",
    "##\n",
    "Y_train = to_categorical(Y_train, num_classes = 10)\n",
    "Y_test = to_categorical(Y_test, num_classes = 10)\n",
    "## one-hot encoding"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "ae068bd8-b12e-4768-8a7e-0fc865dd7562",
    "_uuid": "dcfb688587dfc6feafd27442a3505e35dc01b82d"
   },
   "source": [
    "Labels are 10 digits numbers from 0 to 9. We need to encode these lables to one hot vectors (ex : 2 -> [0,0,1,0,0,0,0,0,0,0])."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "32152fc3-a570-4d64-8a7d-6c689a4acd33",
    "_uuid": "d8abbbf31483b94e1b29d07c4c8253d1311648a7"
   },
   "source": [
    "## 2.6 Split training and valdiation set "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "_cell_guid": "3a698301-9759-4279-ae48-fd980f89ea53",
    "_execution_state": "idle",
    "_uuid": "6e51c925c6e0f1b936679c9649fef345c853555f"
   },
   "outputs": [],
   "source": [
    "# Set the random seed\n",
    "random_seed = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "_cell_guid": "dcd25ebb-d845-4d32-9867-082e352b1396",
    "_execution_state": "idle",
    "_uuid": "b779ac76d8317647db92d5a88b4098d212d72884"
   },
   "outputs": [],
   "source": [
    "# 将训练集合按照9:1 分成训练集合 和验证集合 validation 10折交叉验证 10-fold validation  ####\n",
    "X_train, X_val, Y_train, Y_val = train_test_split(X_train, Y_train, test_size = 0.1, random_state=random_seed)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "adbeacf0-0dc0-4675-b2df-9c9663750f32",
    "_uuid": "60eed15ec5bc0d354385301789ecb8538fc02267"
   },
   "source": [
    "We can get a better sense for one of these examples by visualising the image and looking at the label."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "_cell_guid": "5f76131b-4ba0-45f1-a98c-bd4e7d561793",
    "_execution_state": "idle",
    "_uuid": "e0dae8943d3d35f075dba3d7ba31bde1d4bf2ff4"
   },
   "outputs": [],
   "source": [
    "# Some examples #x-train里面第一个sample的 0:最大 0:最大 0 [:,:,0] \n",
    "#g = plt.imshow(X_train[0][:,:,0],cmap='gray') #plt为什么把灰度可以生"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "d5265777-aeb3-449d-b171-d88cad74c0a4",
    "_uuid": "5fa18b37a9acd9e098bac1d12264b0dd4310fdd3"
   },
   "source": [
    "# 4. RNN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "504fa00e-148c-4364-9b68-218b3aaedfdb",
    "_uuid": "7697570491420f957f6e4d3569d51410b5277250"
   },
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "### RNN 答案 ########\n",
    "batch_size = 100\n",
    "num_classes = 10\n",
    "epochs = 50\n",
    "from keras.layers import SimpleRNN,LSTM\n",
    "# 如果图片是28*32像素\n",
    "# timsteps = 32 / 28 都可以 \n",
    "#(timesteps, input_dim)   28个时间节点的，28个维vector     \n",
    "# keras.layers.LSTM(units, activation='tanh', recurrent_activation='hard_sigmoid', use_bias=True, kernel_initializer='glorot_uniform', recurrent_initializer='orthogonal', bias_initializer='zeros', unit_forget_bias=True, kernel_regularizer=None, recurrent_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, recurrent_constraint=None, bias_constraint=None, dropout=0.0, recurrent_dropout=0.0, implementation=1, return_sequences=False, return_state=False, go_backwards=False, stateful=False, unroll=False)\n",
    "model = Sequential()\n",
    "model.add(LSTM(128, input_shape=(28,28), return_sequences=True))\n",
    "model.add(Dropout(0.2))\n",
    "\n",
    "model.add(LSTM(128))\n",
    "model.add(Dropout(0.1))\n",
    "\n",
    "model.add(Dense(64, activation='relu'))\n",
    "model.add(Dropout(0.2))\n",
    "\n",
    "model.add(Dense(10, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3780, 28, 28)"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "_cell_guid": "a4c55409-6a65-400a-b5e8-a1dc535429c0",
    "_execution_state": "idle",
    "_uuid": "420c704367b397b8255fefe9d882b35ac8929b95"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_4\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lstm_1 (LSTM)                (None, 28, 128)           80384     \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 28, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_2 (LSTM)                (None, 128)               131584    \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 10)                650       \n",
      "=================================================================\n",
      "Total params: 220,874\n",
      "Trainable params: 220,874\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "### 运行model.summary（）回答下列问题 第二天课上一起讨论 ####\n",
    "model.summary()\n",
    "## LSTM中的参数 跟simpleRNN比 什么变化，越多的参数会有什么结果？###\n",
    "# "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "#优化器  尝试使用不同的优化器 至少以下三种 在DL 一个调节的点\n",
    "## 中文参考 https://keras.io/zh/optimizers/\n",
    "\n",
    "## SGD(lr=0.01, momentum=0.0, decay=0.0, nesterov=False)\n",
    "## RMSprop(lr=0.001, rho=0.9, epsilon=None, decay=0.0)\n",
    "## Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)\n",
    "\n",
    "optimizer = RMSprop(lr=0.001, rho=0.9, epsilon=0.0000001, decay=0.0)\n",
    "\n",
    "### 将模型compile 编译\n",
    "### 调节loss 参数，即loss function\n",
    "### mean_squared_error\n",
    "### categorical_crossentropy/为什么不用binary_crossentropy\n",
    "### 尝试用\n",
    "### \n",
    "### mean_absolute_error \n",
    "model.compile(optimizer = optimizer , loss = \"categorical_crossentropy\", metrics=[\"accuracy\"])\n",
    "\n",
    "### training 过程中的 自动调节函数\n",
    "### Reduce LR On Plateau = 减少学习率，当某一个参数达到一个平台期 自动的 把上面优化器中的 lr 减小\n",
    "\n",
    "learning_rate_reduction = ReduceLROnPlateau(monitor='val_accuracy', \n",
    "                                            patience=3, \n",
    "                                            verbose=1, \n",
    "                                            factor=0.5, \n",
    "                                            min_lr=0.00001)\n",
    "\n",
    "\n",
    "\n",
    "#### LSTM 尝试用 sparse_categorical_crossentropy 看看有什么变化结果 ##########"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 3780 samples, validate on 420 samples\n",
      "Epoch 1/50\n",
      "3780/3780 [==============================] - 4s 1ms/step - loss: 1.6891 - accuracy: 0.4108 - val_loss: 1.2944 - val_accuracy: 0.5619\n",
      "Epoch 2/50\n",
      "3780/3780 [==============================] - 3s 743us/step - loss: 1.0831 - accuracy: 0.6201 - val_loss: 0.9908 - val_accuracy: 0.6643\n",
      "Epoch 3/50\n",
      "3780/3780 [==============================] - 3s 765us/step - loss: 0.8485 - accuracy: 0.7071 - val_loss: 1.0563 - val_accuracy: 0.6381\n",
      "Epoch 4/50\n",
      "3780/3780 [==============================] - 3s 764us/step - loss: 0.6834 - accuracy: 0.7667 - val_loss: 0.6043 - val_accuracy: 0.8000\n",
      "Epoch 5/50\n",
      "3780/3780 [==============================] - 3s 761us/step - loss: 0.5438 - accuracy: 0.8198 - val_loss: 0.5852 - val_accuracy: 0.7929\n",
      "Epoch 6/50\n",
      "3780/3780 [==============================] - 3s 863us/step - loss: 0.4386 - accuracy: 0.8556 - val_loss: 0.6277 - val_accuracy: 0.8000\n",
      "Epoch 7/50\n",
      "3780/3780 [==============================] - 3s 860us/step - loss: 0.3684 - accuracy: 0.8770 - val_loss: 0.4610 - val_accuracy: 0.8429\n",
      "Epoch 8/50\n",
      "3780/3780 [==============================] - 3s 865us/step - loss: 0.3219 - accuracy: 0.8987 - val_loss: 0.4078 - val_accuracy: 0.8762\n",
      "Epoch 9/50\n",
      "3780/3780 [==============================] - 3s 823us/step - loss: 0.2623 - accuracy: 0.9161 - val_loss: 0.3544 - val_accuracy: 0.8929\n",
      "Epoch 10/50\n",
      "3780/3780 [==============================] - 3s 871us/step - loss: 0.2204 - accuracy: 0.9312 - val_loss: 0.2894 - val_accuracy: 0.8952\n",
      "Epoch 11/50\n",
      "3780/3780 [==============================] - 3s 845us/step - loss: 0.2143 - accuracy: 0.9336 - val_loss: 0.3163 - val_accuracy: 0.9000\n",
      "Epoch 12/50\n",
      "3780/3780 [==============================] - 3s 887us/step - loss: 0.1906 - accuracy: 0.9402 - val_loss: 0.2430 - val_accuracy: 0.9310\n",
      "Epoch 13/50\n",
      "3780/3780 [==============================] - 3s 839us/step - loss: 0.1515 - accuracy: 0.9513 - val_loss: 0.2799 - val_accuracy: 0.9262\n",
      "Epoch 14/50\n",
      "3780/3780 [==============================] - 3s 859us/step - loss: 0.1420 - accuracy: 0.9550 - val_loss: 0.2375 - val_accuracy: 0.9381\n",
      "Epoch 15/50\n",
      "3780/3780 [==============================] - 3s 881us/step - loss: 0.1268 - accuracy: 0.9601 - val_loss: 0.2348 - val_accuracy: 0.9286\n",
      "Epoch 16/50\n",
      "3780/3780 [==============================] - 3s 868us/step - loss: 0.1198 - accuracy: 0.9616 - val_loss: 0.2919 - val_accuracy: 0.9071\n",
      "Epoch 17/50\n",
      "3780/3780 [==============================] - 3s 850us/step - loss: 0.1141 - accuracy: 0.9643 - val_loss: 0.2506 - val_accuracy: 0.9190\n",
      "Epoch 18/50\n",
      "3780/3780 [==============================] - 3s 872us/step - loss: 0.0872 - accuracy: 0.9759 - val_loss: 0.2541 - val_accuracy: 0.9357\n",
      "Epoch 19/50\n",
      "3780/3780 [==============================] - 3s 875us/step - loss: 0.0834 - accuracy: 0.9735 - val_loss: 0.2208 - val_accuracy: 0.9286\n",
      "Epoch 20/50\n",
      "3780/3780 [==============================] - 3s 862us/step - loss: 0.0859 - accuracy: 0.9738 - val_loss: 0.2566 - val_accuracy: 0.9286\n",
      "Epoch 21/50\n",
      "3780/3780 [==============================] - 3s 859us/step - loss: 0.0723 - accuracy: 0.9780 - val_loss: 0.2132 - val_accuracy: 0.9286\n",
      "Epoch 22/50\n",
      "3780/3780 [==============================] - 3s 863us/step - loss: 0.0677 - accuracy: 0.9802 - val_loss: 0.1942 - val_accuracy: 0.9405\n",
      "Epoch 23/50\n",
      "3780/3780 [==============================] - 4s 1ms/step - loss: 0.0643 - accuracy: 0.9815 - val_loss: 0.5409 - val_accuracy: 0.8762\n",
      "Epoch 24/50\n",
      "3780/3780 [==============================] - 3s 840us/step - loss: 0.0552 - accuracy: 0.9839 - val_loss: 0.2600 - val_accuracy: 0.9310\n",
      "Epoch 25/50\n",
      "3780/3780 [==============================] - 4s 943us/step - loss: 0.0550 - accuracy: 0.9825 - val_loss: 0.2178 - val_accuracy: 0.9452\n",
      "Epoch 26/50\n",
      "3780/3780 [==============================] - 3s 880us/step - loss: 0.0383 - accuracy: 0.9876 - val_loss: 0.2499 - val_accuracy: 0.9333\n",
      "Epoch 27/50\n",
      "3780/3780 [==============================] - 3s 831us/step - loss: 0.0476 - accuracy: 0.9841 - val_loss: 0.3365 - val_accuracy: 0.9238\n",
      "Epoch 28/50\n",
      "3780/3780 [==============================] - 3s 827us/step - loss: 0.0429 - accuracy: 0.9868 - val_loss: 0.3660 - val_accuracy: 0.9214\n",
      "Epoch 29/50\n",
      "3780/3780 [==============================] - 3s 800us/step - loss: 0.0379 - accuracy: 0.9889 - val_loss: 0.2217 - val_accuracy: 0.9381\n",
      "Epoch 30/50\n",
      "3780/3780 [==============================] - 3s 802us/step - loss: 0.0452 - accuracy: 0.9878 - val_loss: 0.2301 - val_accuracy: 0.9500\n",
      "Epoch 31/50\n",
      "3780/3780 [==============================] - 3s 814us/step - loss: 0.0282 - accuracy: 0.9902 - val_loss: 0.2074 - val_accuracy: 0.9476\n",
      "Epoch 32/50\n",
      "3780/3780 [==============================] - 3s 838us/step - loss: 0.0351 - accuracy: 0.9921 - val_loss: 0.3468 - val_accuracy: 0.9262\n",
      "Epoch 33/50\n",
      "3780/3780 [==============================] - 3s 887us/step - loss: 0.0326 - accuracy: 0.9902 - val_loss: 0.2599 - val_accuracy: 0.9452\n",
      "Epoch 34/50\n",
      "3780/3780 [==============================] - 3s 848us/step - loss: 0.0283 - accuracy: 0.9884 - val_loss: 0.2007 - val_accuracy: 0.9452\n",
      "Epoch 35/50\n",
      "3780/3780 [==============================] - 3s 790us/step - loss: 0.0247 - accuracy: 0.9923 - val_loss: 0.2301 - val_accuracy: 0.9500\n",
      "Epoch 36/50\n",
      "3780/3780 [==============================] - 3s 868us/step - loss: 0.0302 - accuracy: 0.9921 - val_loss: 0.2293 - val_accuracy: 0.9476\n",
      "Epoch 37/50\n",
      "3780/3780 [==============================] - 3s 889us/step - loss: 0.0262 - accuracy: 0.9921 - val_loss: 0.1924 - val_accuracy: 0.9548\n",
      "Epoch 38/50\n",
      "3780/3780 [==============================] - 3s 853us/step - loss: 0.0320 - accuracy: 0.9902 - val_loss: 0.2042 - val_accuracy: 0.9524\n",
      "Epoch 39/50\n",
      "3780/3780 [==============================] - 3s 908us/step - loss: 0.0149 - accuracy: 0.9960 - val_loss: 0.2503 - val_accuracy: 0.9405\n",
      "Epoch 40/50\n",
      "3780/3780 [==============================] - 3s 876us/step - loss: 0.0302 - accuracy: 0.9907 - val_loss: 0.2281 - val_accuracy: 0.9476\n",
      "Epoch 41/50\n",
      "3780/3780 [==============================] - 3s 857us/step - loss: 0.0264 - accuracy: 0.9929 - val_loss: 0.1998 - val_accuracy: 0.9524\n",
      "Epoch 42/50\n",
      "3780/3780 [==============================] - 3s 830us/step - loss: 0.0249 - accuracy: 0.9921 - val_loss: 0.2365 - val_accuracy: 0.9429\n",
      "Epoch 43/50\n",
      "3780/3780 [==============================] - 3s 851us/step - loss: 0.0136 - accuracy: 0.9960 - val_loss: 0.3013 - val_accuracy: 0.9405\n",
      "Epoch 44/50\n",
      "3780/3780 [==============================] - 3s 829us/step - loss: 0.0278 - accuracy: 0.9934 - val_loss: 0.2738 - val_accuracy: 0.9405\n",
      "Epoch 45/50\n",
      "3780/3780 [==============================] - 3s 851us/step - loss: 0.0190 - accuracy: 0.9944 - val_loss: 0.3829 - val_accuracy: 0.9500\n",
      "Epoch 46/50\n",
      "3780/3780 [==============================] - 3s 847us/step - loss: 0.0146 - accuracy: 0.9963 - val_loss: 0.2123 - val_accuracy: 0.9524\n",
      "Epoch 47/50\n",
      "3780/3780 [==============================] - 3s 883us/step - loss: 0.0256 - accuracy: 0.9913 - val_loss: 0.2362 - val_accuracy: 0.9405\n",
      "Epoch 48/50\n",
      "3780/3780 [==============================] - 3s 852us/step - loss: 0.0120 - accuracy: 0.9958 - val_loss: 0.2287 - val_accuracy: 0.9548\n",
      "Epoch 49/50\n",
      "3780/3780 [==============================] - 3s 870us/step - loss: 0.0135 - accuracy: 0.9958 - val_loss: 0.2529 - val_accuracy: 0.9500\n",
      "Epoch 50/50\n",
      "3780/3780 [==============================] - 3s 841us/step - loss: 0.0107 - accuracy: 0.9974 - val_loss: 0.2192 - val_accuracy: 0.9571\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train,Y_train, batch_size=batch_size,\n",
    "                              epochs = epochs, validation_data = (X_val,Y_val))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 生成学习曲线 和损失函数 随着epoch的变化曲线\n",
    "# 模型的学习效果怎么样？ 能找到适合的epoch吗？\n",
    "# 简单的评价标准应该用什么？\n",
    "# 尝试改变模型参数 生成不同的学习曲线 比较\n",
    "# 提示 从epoch> 优化器> 损失函数> 学习率> dropout有无 依次调试 \n",
    "\n",
    "fig, ax = plt.subplots(2,1)\n",
    "\n",
    "ax[0].plot(history.history['loss'], color='b', label=\"Training loss\")\n",
    "ax[0].plot(history.history['val_loss'], color='r', label=\"validation loss\",axes =ax[0])\n",
    "legend = ax[0].legend(loc='best', shadow=True)\n",
    "\n",
    "ax[1].plot(history.history['accuracy'], color='b', label=\"Training accuracy\")\n",
    "ax[1].plot(history.history['val_accuracy'], color='r',label=\"Validation accuracy\")\n",
    "legend = ax[1].legend(loc='best', shadow=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 生成10标签混淆矩阵\n",
    "\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, cm[i, j],\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "\n",
    "# Predict the values from the validation dataset\n",
    "Y_pred = model.predict(X_val)\n",
    "# Convert predictions classes to one hot vectors \n",
    "Y_pred_classes = np.argmax(Y_pred,axis = 1) \n",
    "# Convert validation observations to one hot vectors\n",
    "Y_true = np.argmax(Y_val,axis = 1) \n",
    "# compute the confusion matrix\n",
    "confusion_mtx = confusion_matrix(Y_true, Y_pred_classes) \n",
    "# plot the confusion matrix\n",
    "plot_confusion_matrix(confusion_mtx, classes = range(10)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### 打印出认错的数字\n",
    "\n",
    "errors = (Y_pred_classes - Y_true != 0)\n",
    "\n",
    "Y_pred_classes_errors = Y_pred_classes[errors]\n",
    "Y_pred_errors = Y_pred[errors]\n",
    "Y_true_errors = Y_true[errors]\n",
    "X_val_errors = X_val[errors]\n",
    "\n",
    "def display_errors(errors_index,img_errors,pred_errors, obs_errors):\n",
    "    \"\"\" This function shows 6 images with their predicted and real labels\"\"\"\n",
    "    n = 0\n",
    "    nrows = 3\n",
    "    ncols = 3\n",
    "    fig, ax = plt.subplots(nrows,ncols,sharex=True,sharey=True)\n",
    "    for row in range(nrows):\n",
    "        for col in range(ncols):\n",
    "            error = errors_index[n]\n",
    "            ax[row,col].imshow((img_errors[error]).reshape((28,28)))\n",
    "            ax[row,col].set_title(\"Predicted label :{}\\nTrue label :{}\".format(pred_errors[error],obs_errors[error]))\n",
    "            n += 1\n",
    "\n",
    "# Probabilities of the wrong predicted numbers\n",
    "Y_pred_errors_prob = np.max(Y_pred_errors,axis = 1)\n",
    "\n",
    "# Predicted probabilities of the true values in the error set\n",
    "true_prob_errors = np.diagonal(np.take(Y_pred_errors, Y_true_errors, axis=1))\n",
    "\n",
    "# Difference between the probability of the predicted label and the true label\n",
    "delta_pred_true_errors = Y_pred_errors_prob - true_prob_errors\n",
    "\n",
    "# Sorted list of the delta prob errors\n",
    "sorted_dela_errors = np.argsort(delta_pred_true_errors)\n",
    "\n",
    "# Top 9 errors \n",
    "most_important_errors = sorted_dela_errors[-9:]\n",
    "\n",
    "# Show the top 9 errors\n",
    "display_errors(most_important_errors, X_val_errors, Y_pred_classes_errors, Y_true_errors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "#optional 画出roc\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "fpr = dict()\n",
    "tpr = dict()\n",
    "roc_auc = dict()\n",
    "y_score = model.predict(X_test)\n",
    "# 在前天的作业中 y_test Pandas下的DataFrame类型：y_test\n",
    "# 让数据为 Pandas DataFrame类型的话 调用/使用他 第i行第j列的数据:\n",
    "#  y_test.iloc[i,j]\n",
    "\n",
    "# 在今天的作业中，y_test是 numpy的 numarry数据类型\n",
    "# 让数据为numarray 类型的话 调用/使用他 第i行第j列的数据:\n",
    "# y_test[i,j]\n",
    "for i in range(num_classes):\n",
    "     fpr[i], tpr[i], _ = roc_curve(Y_test[:,i], y_score[:,i]) #\n",
    "    # AUC Area Under the Curve\n",
    "     roc_auc[i] = auc(fpr[i], tpr[i])\n",
    "#y_pred_keras = model.predict(X_test).ravel()\n",
    "##fpr_keras, tpr_keras, thresholds_keras = roc_curve(Y_test, y_pred_keras)\n",
    "#y_pred_keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(num_classes):\n",
    "    plt.plot(fpr[i], tpr[i], lw=2, label='ROC curve of class {0} (area = {1:0.2f})'\n",
    "             ''.format(i, roc_auc[i]))\n",
    "\n",
    "plt.plot([0, 1], [0, 1], 'k--', lw=2)\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Some extension of Receiver operating characteristic to multi-class')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from keras.utils import plot_model\n",
    "plot_model(model, to_file='model.png', show_shapes=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(210, 10)"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_score.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": 2
}