#!/usr/bin/env python
# coding: utf-8

# In[1]:


import warnings
warnings.simplefilter(action='ignore')


# In[2]:


import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd


# ### 1. 数据预处理

# In[3]:


(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()


# In[4]:


x_train_flatten = x_train.reshape(x_train.shape[0], 784).astype('float32')
x_test_flatten = x_test.reshape(x_test.shape[0], 784).astype('float32')


# In[5]:


x_train_normalize = x_train_flatten / 255.0
x_test_normalize = x_test_flatten / 255.0


# In[6]:


y_train_one_hot = tf.keras.utils.to_categorical(y_train)
y_test_one_hot = tf.keras.utils.to_categorical(y_test)


# ### 2. 建立模型

# #### 2.1 建立 Sequential 模型

# In[7]:


model = tf.keras.models.Sequential([
    tf.keras.layers.Dense(units=1000, input_dim=784, kernel_initializer='normal', activation='relu'),  # 输入层-隐藏层(这里隐藏层为1000个神经元)
    tf.keras.layers.Dropout(0.5),  # 添加 Dropout 层
    tf.keras.layers.Dense(units=10, kernel_initializer='normal', activation='softmax')  # 输出层
])


# #### 2.2 查看模型的摘要

# In[8]:


print(model.summary())


# ### 3. 训练模型

# In[9]:


model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])


# In[10]:


train_history = model.fit(x=x_train_normalize, y=y_train_one_hot, validation_split=0.2,
                         epochs=10, batch_size=200, verbose=2)


# ### 4. 以图形显示训练过程

# In[11]:


def show_train_history(train_history, train, validation):
    plt.plot(train_history.history[train])
    plt.plot(train_history.history[validation])
    plt.title('Train History')
    plt.xlabel('Epoch')
    plt.ylabel(train)
    plt.legend(['train', 'validation'], loc='upper left')
    plt.show()


# In[12]:


show_train_history(train_history, 'acc', 'val_acc')


# In[13]:


show_train_history(train_history, 'loss', 'val_loss')


# ### 5. 评估模型的准确率

# In[14]:


scores = model.evaluate(x_test_normalize, y_test_one_hot)
print()
print('accuracy:', scores[1])


# ### 6. 进行预测

# #### 6.1 执行预测

# In[15]:


predictions = model.predict_classes(x_test_normalize)


# #### 6.2 预测结果

# In[16]:


predictions


# #### 6.3 定义函数以显示10项预测结果

# In[17]:


def plot_images_labels_prediction(images, labels, predictions, idx, num=10):
    """
    images: 数字图像数组
    labels: 真实值数组
    predictions: 预测结果数据
    idx: 开始显示的数据index
    num: 要显示的数据项数, 默认为10, 不超过25
    """
    fig = plt.gcf()
    fig.set_size_inches(12, 14)
    if num > 25:
        num = 25
    for i in range(0, num):
        ax = plt.subplot(5, 5, i+1)
        ax.imshow(images[idx], cmap='binary')
        title = 'lable=' + str(labels[idx])
        if len(predictions) > 0:
            title += ',predict=' + str(predictions[idx])
        ax.set_title(title, fontsize=10)
        ax.set_xticks([])
        ax.set_yticks([])
        idx += 1
    plt.show()


# In[18]:


plot_images_labels_prediction(x_test, y_test, predictions, idx=0, num=10)


# ### 7. 显示混淆矩阵

# #### 7.1 建立混淆矩阵

# In[19]:


pd.crosstab(y_test, predictions, rownames=['label'], colnames=['predict'])


# #### 7.2 建立真实值与预测 DataFrame

# In[20]:


df = pd.DataFrame({'label': y_test, 'predict': predictions})
df[:2]


# #### 7.3 查询真实值是 "5" 但预测值是 "2" 的数据

# In[21]:


df[(df.label==5)&(df.predict==2)]


# In[22]:


plot_images_labels_prediction(x_test, y_test, predictions, idx=340, num=1)


# In[23]:


plot_images_labels_prediction(x_test, y_test, predictions, idx=1289, num=1)