基于RNN模型的心脏病预测(tensorflow实现)

• 🍨 本文为🔗365天深度学习训练营 中的学习记录博客
• 🍖 原作者：K同学啊

前言

• RNN是很经典的模型，原理参考：深度学习基础–一文搞懂RNN
• 这个案例是一个基础案例，用RNN模型去做一个二分类问题，心脏病预测，数据集在kaggle上可以找到；
• 欢迎收藏加关注，本人将会持续更新。
  

  文章目录

  • • 1、数据处理
    • • 1、导入库
      • 2、导入数据
      • 3、数据分析
      • • 数据初步分析
        • 缺失值
        • 相关性分析
      • 4、数据划分
      • 5、数据标准化
    • 2、创建模型
    • 3、设置超参数
    • 4、模型训练
    • 5、结果展示
    • 6、模型评估

1、数据处理

1、导入库

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

2、导入数据

data = pd.read_csv('./heart.csv')

data.head()

• age - 年龄
• sex - (1 = male(男性); 0 = (女性))
• cp - chest pain type(胸部疼痛类型)（1：典型的心绞痛-typical，2：非典型心绞痛-atypical，3：没有心绞痛-non-anginal，4：无症状-asymptomatic）
• trestbps - 静息血压 (in mm Hg on admission to the hospital)
• chol - 胆固醇 in mg/dl
• fbs - (空腹血糖 > 120 mg/dl) (1 = true; 0 = false)
• restecg - 静息心电图测量（0：普通，1：ST-T波异常，2：可能左心室肥大）
• thalach - 最高心跳率
• exang - 运动诱发心绞痛 (1 = yes; 0 = no)
• oldpeak - 运动相对于休息引起的ST抑制
• slope - 运动ST段的峰值斜率（1：上坡-upsloping，2：平的-flat，3：下坡-downsloping）
• ca - 主要血管数目(0-4)
• thal - 一种叫做地中海贫血的血液疾病（3 = normal; 6 = 固定的缺陷-fixed defect; 7 = 可逆的缺陷-reversable defect）
• target - 是否患病 (1=yes, 0=no)

3、数据分析

数据初步分析

data.info()   # 数据类型分析

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 303 entries, 0 to 302
Data columns (total 14 columns):
 #   Column    Non-Null Count  Dtype  
---  ------    --------------  -----  
 0   age       303 non-null    int64  
 1   sex       303 non-null    int64  
 2   cp        303 non-null    int64  
 3   trestbps  303 non-null    int64  
 4   chol      303 non-null    int64  
 5   fbs       303 non-null    int64  
 6   restecg   303 non-null    int64  
 7   thalach   303 non-null    int64  
 8   exang     303 non-null    int64  
 9   oldpeak   303 non-null    float64
 10  slope     303 non-null    int64  
 11  ca        303 non-null    int64  
 12  thal      303 non-null    int64  
 13  target    303 non-null    int64  
dtypes: float64(1), int64(13)
memory usage: 33.3 KB

其中分类变量为：sex、cp、fbs、restecg、exang、slope、ca、thal、target

数值型变量：age、trestbps、chol、thalach、oldpeak

data.describe()  # 描述性

• 年纪：均值54，中位数55，标准差9，说明主要是老年人，偏大
• 静息血压：均值131.62， 成年人一般：正常血压：收缩压 < 120 mmHg，偏大
• 胆固醇：均值246.26，理想水平：小于 200 mg/dL，偏大
• 最高心率：均值149.64，一般静息状态下通常是 60 到 100 次每分钟，偏大

最大值和最小值都可能发生，无异常值

缺失值

data.isnull().sum()

age         0
sex         0
cp          0
trestbps    0
chol        0
fbs         0
restecg     0
thalach     0
exang       0
oldpeak     0
slope       0
ca          0
thal        0
target      0
dtype: int64

相关性分析

import seaborn as sns

plt.figure(figsize=(20, 15))

sns.heatmap(data.corr(), annot=True, cmap='Greens')

plt.show()

​

​

相关系数的等级划分

• 非常弱的相关性：
  • 0.00 至 0.19 或 -0.00 至 -0.19
  • 解释：几乎不存在线性关系。
• 弱相关性：
  • 0.20 至 0.39 或 -0.20 至 -0.39
  • 解释：存在一定的线性关系，但较弱。
• 中等相关性：
  • 0.40 至 0.59 或 -0.40 至 -0.59
  • 解释：有明显的线性关系，但不是特别强。
• 强相关性：
  • 0.60 至 0.79 或 -0.60 至 -0.79
  • 解释：两个变量之间有较强的线性关系。
• 非常强的相关性：
  • 0.80 至 1.00 或 -0.80 至 -1.00
  • 解释：几乎完全线性相关，表明两个变量的变化高度一致。

target与chol、没有什么相关性，fbs是分类变量，chol胆固醇是数值型变量，但是从实际角度，这些都有影响，故不剔除特征

4、数据划分

这里先划分为：训练集：测试集 = 9：1

from sklearn.model_selection import train_test_split

X = data.iloc[:, :-1]
y = data.iloc[:, -1]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=42)

5、数据标准化

from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)

# 深度学习、用rnn模型，数据需要3通道，在图片中表示RGB，这里表示1
X_train = X_train.reshape(X_train.shape[0], X_train.shape[1], 1)
X_test = X_test.reshape(X_test.shape[0], X_test.shape[1], 1)

2、创建模型

• RNN的API
  • tf.keras.layers.SimpleRNN(units,activation=‘tanh’,use_bias=True,kernel_initializer=‘glorot_uniform’,
    recurrent_initializer=‘orthogonal’,bias_initializer=‘zeros’,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,return_sequences=False,return_state=False,
    go_backwards=False,stateful=False,unroll=False,**kwargs)
• 参数
  • units: 正整数，表示该层输出空间的维度，即隐藏状态的大小。
  • activation: 激活函数，默认是 ‘tanh’。可以使用其他激活函数如 ‘relu’ 或自定义激活函数。
  • input_shape=(13, 1): 指定输入数据的形状。对于这个 RNN 层，每个样本包含长度为 13 的时间序列，每个时间步有一个特征（即每个时间点的数据维度是 1）。请注意，当你在模型的第一层使用 input_shape 参数时，你不需要指定批量大小（batch size），它默认是 None，意味着批量大小可以是任意值。
  • use_bias: 布尔值，默认为 True，指示是否使用偏置向量。
  • kernel_initializer: 权重矩阵的初始化方法，默认是 ‘glorot_uniform’。
  • recurrent_initializer: 循环核的初始化方法，默认是 ‘orthogonal’。
  • bias_initializer: 偏置向量的初始化方法，默认是 ‘zeros’。
  • kernel_regularizer: 权重矩阵的正则化方法。
  • recurrent_regularizer: 循环核的正则化方法。
  • bias_regularizer: 偏置向量的正则化方法。
  • activity_regularizer: 输出的正则化方法。
  • kernel_constraint: 对权重矩阵施加约束的方法。
  • recurrent_constraint: 对循环核施加约束的方法。
  • bias_constraint: 对偏置向量施加约束的方法。
  • dropout: 浮点数，介于 0 和 1 之间，指输入单元的丢弃比例，默认为 0.0。
  • recurrent_dropout: 浮点数，介于 0 和 1 之间，指循环状态的丢弃比例，默认为 0.0。
  • return_sequences: 布尔值，默认为 False。如果设置为 True，则整个序列会被返回；否则，只返回最后一个输出。
  • return_state: 布尔值，默认为 False。如果设置为 True，则除了输出外还会返回最后一个状态。
  • go_backwards: 布尔值，默认为 False。如果设置为 True，则会反向处理输入序列。
  • stateful: 布尔值，默认为 False。如果设置为 True，则批次间的状态会被保留下来。
  • unroll: 布尔值，默认为 False。如果设置为 True，则网络将被展开。当且仅当输入序列长度有限时适用，可以加速计算但占用更多内存。

import tensorflow as tf  
from tensorflow.keras.models import Sequential 
from tensorflow.keras.layers import SimpleRNN, Dense

# 创建模型
'''
该问题本质是二分类问题，故最后一层全连接层用激活函数为：sigmoid
模型结构：
    RNN：隐藏层200，激活函数：relu
    Dense：--> 100(relu) -> 1(sigmoid)
'''
# 创建模型
model = Sequential()
model.add(SimpleRNN(units=200, input_shape=(13, 1), activation='relu'))
model.add(Dense(100, activation='relu'))
model.add(Dense(1, activation='sigmoid'))

model.summary()

Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 simple_rnn (SimpleRNN)      (None, 200)               40400     
                                                                 
 dense (Dense)               (None, 100)               20100     
                                                                 
 dense_1 (Dense)             (None, 1)                 101       
                                                                 
=================================================================
Total params: 60,601
Trainable params: 60,601
Non-trainable params: 0
_________________________________________________________________

3、设置超参数

opt = tf.keras.optimizers.Adam(learning_rate=1e-4)

model.compile(
    optimizer=opt,
    loss='binary_crossentropy',  # 二分类问题
    metrics=['accuracy']
)

4、模型训练

epochs = 100

history = model.fit(
    X_train, y_train,
    epochs=epochs,
    batch_size=32,
    validation_data=(X_test, y_test),
    verbose=1
)

Epoch 1/100
9/9 [==============================] - 1s 46ms/step - loss: 0.6821 - accuracy: 0.5551 - val_loss: 0.6679 - val_accuracy: 0.6774
Epoch 2/100
9/9 [==============================] - 0s 8ms/step - loss: 0.6549 - accuracy: 0.7059 - val_loss: 0.6460 - val_accuracy: 0.7097
Epoch 3/100
9/9 [==============================] - 0s 8ms/step - loss: 0.6299 - accuracy: 0.7904 - val_loss: 0.6265 - val_accuracy: 0.7419
Epoch 4/100
9/9 [==============================] - 0s 8ms/step - loss: 0.6051 - accuracy: 0.7978 - val_loss: 0.6062 - val_accuracy: 0.7097
Epoch 5/100
9/9 [==============================] - 0s 8ms/step - loss: 0.5784 - accuracy: 0.8015 - val_loss: 0.5835 - val_accuracy: 0.7097
Epoch 6/100
9/9 [==============================] - 0s 8ms/step - loss: 0.5484 - accuracy: 0.8051 - val_loss: 0.5573 - val_accuracy: 0.7097
Epoch 7/100
9/9 [==============================] - 0s 8ms/step - loss: 0.5103 - accuracy: 0.8125 - val_loss: 0.5264 - val_accuracy: 0.7419
Epoch 8/100
9/9 [==============================] - 0s 8ms/step - loss: 0.4676 - accuracy: 0.8162 - val_loss: 0.5022 - val_accuracy: 0.7742
Epoch 9/100
9/9 [==============================] - 0s 8ms/step - loss: 0.4247 - accuracy: 0.8088 - val_loss: 0.4968 - val_accuracy: 0.8065
Epoch 10/100
9/9 [==============================] - 0s 8ms/step - loss: 0.4020 - accuracy: 0.8088 - val_loss: 0.5068 - val_accuracy: 0.7742
Epoch 11/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3937 - accuracy: 0.8051 - val_loss: 0.5095 - val_accuracy: 0.7742
Epoch 12/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3824 - accuracy: 0.8235 - val_loss: 0.5062 - val_accuracy: 0.7742
Epoch 13/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3706 - accuracy: 0.8162 - val_loss: 0.5138 - val_accuracy: 0.7742
Epoch 14/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3667 - accuracy: 0.8199 - val_loss: 0.5076 - val_accuracy: 0.7742
Epoch 15/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3528 - accuracy: 0.8346 - val_loss: 0.5169 - val_accuracy: 0.7742
Epoch 16/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3472 - accuracy: 0.8272 - val_loss: 0.5167 - val_accuracy: 0.7742
Epoch 17/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3414 - accuracy: 0.8493 - val_loss: 0.5150 - val_accuracy: 0.7742
Epoch 18/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3462 - accuracy: 0.8235 - val_loss: 0.5171 - val_accuracy: 0.7742
Epoch 19/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3344 - accuracy: 0.8566 - val_loss: 0.5133 - val_accuracy: 0.7742
Epoch 20/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3226 - accuracy: 0.8529 - val_loss: 0.5268 - val_accuracy: 0.8065
Epoch 21/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3192 - accuracy: 0.8566 - val_loss: 0.5237 - val_accuracy: 0.7742
Epoch 22/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3127 - accuracy: 0.8676 - val_loss: 0.5270 - val_accuracy: 0.7742
Epoch 23/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3071 - accuracy: 0.8640 - val_loss: 0.5354 - val_accuracy: 0.8065
Epoch 24/100
9/9 [==============================] - 0s 8ms/step - loss: 0.3029 - accuracy: 0.8713 - val_loss: 0.5337 - val_accuracy: 0.8065
Epoch 25/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2931 - accuracy: 0.8824 - val_loss: 0.5310 - val_accuracy: 0.8065
Epoch 26/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2906 - accuracy: 0.8897 - val_loss: 0.5291 - val_accuracy: 0.8065
Epoch 27/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2833 - accuracy: 0.8934 - val_loss: 0.5333 - val_accuracy: 0.8065
Epoch 28/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2777 - accuracy: 0.8897 - val_loss: 0.5417 - val_accuracy: 0.8065
Epoch 29/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2725 - accuracy: 0.8897 - val_loss: 0.5342 - val_accuracy: 0.8065
Epoch 30/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2696 - accuracy: 0.9044 - val_loss: 0.5417 - val_accuracy: 0.8065
Epoch 31/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2626 - accuracy: 0.8897 - val_loss: 0.5420 - val_accuracy: 0.8065
Epoch 32/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2552 - accuracy: 0.9007 - val_loss: 0.5424 - val_accuracy: 0.8065
Epoch 33/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2506 - accuracy: 0.9044 - val_loss: 0.5456 - val_accuracy: 0.8065
Epoch 34/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2482 - accuracy: 0.9044 - val_loss: 0.5500 - val_accuracy: 0.8065
Epoch 35/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2437 - accuracy: 0.9044 - val_loss: 0.5552 - val_accuracy: 0.8065
Epoch 36/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2425 - accuracy: 0.9191 - val_loss: 0.5511 - val_accuracy: 0.8065
Epoch 37/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2383 - accuracy: 0.9081 - val_loss: 0.5523 - val_accuracy: 0.8065
Epoch 38/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2253 - accuracy: 0.9191 - val_loss: 0.5765 - val_accuracy: 0.8065
Epoch 39/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2265 - accuracy: 0.9191 - val_loss: 0.5664 - val_accuracy: 0.8065
Epoch 40/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2197 - accuracy: 0.9265 - val_loss: 0.5732 - val_accuracy: 0.8065
Epoch 41/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2151 - accuracy: 0.9338 - val_loss: 0.5716 - val_accuracy: 0.8065
Epoch 42/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2199 - accuracy: 0.9154 - val_loss: 0.5718 - val_accuracy: 0.8065
Epoch 43/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2208 - accuracy: 0.9338 - val_loss: 0.5830 - val_accuracy: 0.8065
Epoch 44/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2010 - accuracy: 0.9412 - val_loss: 0.5761 - val_accuracy: 0.8065
Epoch 45/100
9/9 [==============================] - 0s 8ms/step - loss: 0.2038 - accuracy: 0.9265 - val_loss: 0.5897 - val_accuracy: 0.8065
Epoch 46/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1971 - accuracy: 0.9412 - val_loss: 0.5865 - val_accuracy: 0.8065
Epoch 47/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1941 - accuracy: 0.9412 - val_loss: 0.5939 - val_accuracy: 0.8065
Epoch 48/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1917 - accuracy: 0.9375 - val_loss: 0.5984 - val_accuracy: 0.8065
Epoch 49/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1890 - accuracy: 0.9449 - val_loss: 0.5874 - val_accuracy: 0.8065
Epoch 50/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1880 - accuracy: 0.9449 - val_loss: 0.5964 - val_accuracy: 0.8065
Epoch 51/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1799 - accuracy: 0.9485 - val_loss: 0.6004 - val_accuracy: 0.8065
Epoch 52/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1762 - accuracy: 0.9449 - val_loss: 0.6068 - val_accuracy: 0.8065
Epoch 53/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1719 - accuracy: 0.9485 - val_loss: 0.6046 - val_accuracy: 0.8065
Epoch 54/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1720 - accuracy: 0.9522 - val_loss: 0.6117 - val_accuracy: 0.8065
Epoch 55/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1691 - accuracy: 0.9485 - val_loss: 0.6201 - val_accuracy: 0.8065
Epoch 56/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1614 - accuracy: 0.9522 - val_loss: 0.6132 - val_accuracy: 0.8387
Epoch 57/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1597 - accuracy: 0.9559 - val_loss: 0.6295 - val_accuracy: 0.8065
Epoch 58/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1583 - accuracy: 0.9559 - val_loss: 0.6428 - val_accuracy: 0.8065
Epoch 59/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1542 - accuracy: 0.9632 - val_loss: 0.6260 - val_accuracy: 0.8387
Epoch 60/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1515 - accuracy: 0.9559 - val_loss: 0.6527 - val_accuracy: 0.8065
Epoch 61/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1495 - accuracy: 0.9596 - val_loss: 0.6550 - val_accuracy: 0.8065
Epoch 62/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1468 - accuracy: 0.9559 - val_loss: 0.6562 - val_accuracy: 0.8065
Epoch 63/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1452 - accuracy: 0.9596 - val_loss: 0.6574 - val_accuracy: 0.8387
Epoch 64/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1463 - accuracy: 0.9522 - val_loss: 0.6606 - val_accuracy: 0.8065
Epoch 65/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1406 - accuracy: 0.9632 - val_loss: 0.6614 - val_accuracy: 0.8387
Epoch 66/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1322 - accuracy: 0.9706 - val_loss: 0.6803 - val_accuracy: 0.8065
Epoch 67/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1306 - accuracy: 0.9669 - val_loss: 0.6647 - val_accuracy: 0.8387
Epoch 68/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1239 - accuracy: 0.9706 - val_loss: 0.6856 - val_accuracy: 0.8387
Epoch 69/100
9/9 [==============================] - 0s 10ms/step - loss: 0.1195 - accuracy: 0.9743 - val_loss: 0.6805 - val_accuracy: 0.8387
Epoch 70/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1164 - accuracy: 0.9743 - val_loss: 0.7036 - val_accuracy: 0.8387
Epoch 71/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1154 - accuracy: 0.9706 - val_loss: 0.7068 - val_accuracy: 0.8387
Epoch 72/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1107 - accuracy: 0.9706 - val_loss: 0.7011 - val_accuracy: 0.8387
Epoch 73/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1081 - accuracy: 0.9706 - val_loss: 0.7218 - val_accuracy: 0.8387
Epoch 74/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1031 - accuracy: 0.9706 - val_loss: 0.7341 - val_accuracy: 0.8387
Epoch 75/100
9/9 [==============================] - 0s 8ms/step - loss: 0.1045 - accuracy: 0.9706 - val_loss: 0.7233 - val_accuracy: 0.8387
Epoch 76/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0986 - accuracy: 0.9669 - val_loss: 0.7459 - val_accuracy: 0.8387
Epoch 77/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0955 - accuracy: 0.9743 - val_loss: 0.7471 - val_accuracy: 0.8387
Epoch 78/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0900 - accuracy: 0.9743 - val_loss: 0.7459 - val_accuracy: 0.8387
Epoch 79/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0916 - accuracy: 0.9743 - val_loss: 0.7714 - val_accuracy: 0.8387
Epoch 80/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0845 - accuracy: 0.9743 - val_loss: 0.7712 - val_accuracy: 0.8387
Epoch 81/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0817 - accuracy: 0.9743 - val_loss: 0.7707 - val_accuracy: 0.8387
Epoch 82/100
9/9 [==============================] - 0s 10ms/step - loss: 0.0827 - accuracy: 0.9779 - val_loss: 0.7993 - val_accuracy: 0.8387
Epoch 83/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0750 - accuracy: 0.9779 - val_loss: 0.7947 - val_accuracy: 0.8387
Epoch 84/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0738 - accuracy: 0.9743 - val_loss: 0.8213 - val_accuracy: 0.8387
Epoch 85/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0713 - accuracy: 0.9779 - val_loss: 0.8187 - val_accuracy: 0.8387
Epoch 86/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0670 - accuracy: 0.9816 - val_loss: 0.8190 - val_accuracy: 0.8387
Epoch 87/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0643 - accuracy: 0.9816 - val_loss: 0.8394 - val_accuracy: 0.8387
Epoch 88/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0623 - accuracy: 0.9816 - val_loss: 0.8506 - val_accuracy: 0.8387
Epoch 89/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0569 - accuracy: 0.9890 - val_loss: 0.8615 - val_accuracy: 0.8387
Epoch 90/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0551 - accuracy: 0.9890 - val_loss: 0.8653 - val_accuracy: 0.8387
Epoch 91/100
9/9 [==============================] - 0s 12ms/step - loss: 0.0518 - accuracy: 0.9890 - val_loss: 0.8789 - val_accuracy: 0.8387
Epoch 92/100
9/9 [==============================] - 0s 11ms/step - loss: 0.0506 - accuracy: 0.9890 - val_loss: 0.8979 - val_accuracy: 0.8387
Epoch 93/100
9/9 [==============================] - 0s 10ms/step - loss: 0.0475 - accuracy: 0.9853 - val_loss: 0.9083 - val_accuracy: 0.8387
Epoch 94/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0458 - accuracy: 0.9926 - val_loss: 0.8964 - val_accuracy: 0.8387
Epoch 95/100
9/9 [==============================] - 0s 9ms/step - loss: 0.0430 - accuracy: 0.9926 - val_loss: 0.9234 - val_accuracy: 0.8387
Epoch 96/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0422 - accuracy: 0.9926 - val_loss: 0.9358 - val_accuracy: 0.8387
Epoch 97/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0390 - accuracy: 0.9890 - val_loss: 0.9299 - val_accuracy: 0.8387
Epoch 98/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0367 - accuracy: 0.9926 - val_loss: 0.9745 - val_accuracy: 0.8387
Epoch 99/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0348 - accuracy: 0.9926 - val_loss: 0.9798 - val_accuracy: 0.8387
Epoch 100/100
9/9 [==============================] - 0s 8ms/step - loss: 0.0343 - accuracy: 0.9963 - val_loss: 0.9618 - val_accuracy: 0.8387

5、结果展示

train_acc = history.history['accuracy']
train_loss = history.history['loss']

test_acc = history.history['val_accuracy']
test_loss = history.history['val_loss']

epochs_range = range(epochs)

plt.figure(figsize=(15, 5))

plt.subplot(1, 2, 1)
plt.plot(epochs_range, train_acc, label='Train_acc')
plt.plot(epochs_range, test_acc, label='Test_acc')
plt.legend(loc='lower right')
plt.title("Accuracy")

plt.subplot(1, 2, 2)
plt.plot(epochs_range, train_loss, label='Train_loss')
plt.plot(epochs_range, test_loss, label='Test_loss')
plt.legend(loc='upper right')
plt.title("Loss")

plt.show()

​

​

6、模型评估

# 评估：返回的是自己在model.compile中设置，这里为accuracy
score = model.evaluate(X_test, y_test, verbose=0)
print("socre[loss, accuracy]: ", score) # 返回为两个，一个是loss，一个是accuracy

socre[loss, accuracy]:  [0.9617615938186646, 0.8387096524238586]