深度学习Day-14：RNN实现心脏病预测

🍨 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 | 接辅导、项目定制]

要求：

本地读取并加载数据；
了解循环神经网络RNN的构建过程；
测试集accuracy达到87%；

一、基础配置

语言环境：Python3.7
编译器选择：Pycharm
深度学习环境：TensorFlow2.4.1
数据集：私有数据集

二、前期准备

1.设置GPU

import tensorflow as tf

gpus = tf.config.list_physical_devices("GPU")

if gpus:
    tf.config.experimental.set_memory_growth(gpus[0], True)  #设置GPU显存用量按需使用
    tf.config.set_visible_devices([gpus[0]],"GPU")

# 打印显卡信息，确认GPU可用
print(gpus)

根据个人设备情况，选择使用GPU/CPU进行训练，若GPU可用则输出：

[PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]

由于在设备上安装的CUDA版本与TensorFlow版本不一致，故这里直接安装了CPU版的TensorFlow，无上述输出。

2. 导入数据

本项目所采用的数据集未收录于公开数据中，故需要自己在文件目录中导入相应数据集合，并设置对应文件目录，以供后续学习过程中使用。

运行下述代码：

import  pandas as pd

df = pd.read_csv("./data/heart.csv")
print(df)

得到如下输出：

     age  sex  cp  trestbps  chol  fbs  ...  exang  oldpeak  slope  ca  thal  target
0     63    1   3       145   233    1  ...      0      2.3      0   0     1       1
1     37    1   2       130   250    0  ...      0      3.5      0   0     2       1
2     41    0   1       130   204    0  ...      0      1.4      2   0     2       1
3     56    1   1       120   236    0  ...      0      0.8      2   0     2       1
4     57    0   0       120   354    0  ...      1      0.6      2   0     2       1
..   ...  ...  ..       ...   ...  ...  ...    ...      ...    ...  ..   ...     ...
298   57    0   0       140   241    0  ...      1      0.2      1   0     3       0
299   45    1   3       110   264    0  ...      0      1.2      1   0     3       0
300   68    1   0       144   193    1  ...      0      3.4      1   2     3       0
301   57    1   0       130   131    0  ...      1      1.2      1   1     3       0
302   57    0   1       130   236    0  ...      0      0.0      1   1     2       0

[303 rows x 14 columns]

3.检查数据

由于本次所采用的数据集为表格类型，因此，我们需要观察数据中是否有空值、异常值等错误数据的存在，这里我们先观察是否有空值：

dfnull = df.isnull().sum()
print(dfnull)

可以得到如下输出：

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

可见，数据集中无空值存在。

三、数据预处理

1.划分数据集

测试集与验证集的关系：

验证集并没有参与训练过程中的梯度下降过程，即没有参与模型的参数更新过程；
广义上讲，验证集参与了“人工调参”的过程，我们根据每个epoch的训练结果，选择调整对应的超参数
可以认为，验证集参与了训练，但并没有使模型overfit验证集；

from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import  train_test_split

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

X_train,X_test,y_train,y_test = train_test_split(X,y,test_size= 0.1 , random_state= 1)
print(X_train.shape,y_train.shape)

得到如下输出：

(272, 13) (272,)

其中：

X = df.iloc[:,:-1]: 从DataFrame df 中选择所有行和除最后一列外的所有列作为特征矩阵 X；
y = df.iloc[:,-1]: 从DataFrame df 中选择所有行和最后一列作为目标变量 y；
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=1): 将特征矩阵 X 和目标变量 y 划分为训练集和测试集。其中，test_size=0.1 表示测试集占总数据集的比例为10%，random_state=1 是随机种子，保证每次运行代码时划分的结果相同

2.标准化

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

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)

其中：

sc = StandardScaler(): 实例化了StandardScaler类，用于对特征进行标准化处理；
X_train = sc.fit_transform(X_train): 使用fit_transform方法对训练集的特征矩阵进行标准化处理，即将每个特征的数值缩放到均值为0、方差为1的范围内；
X_test = sc.transform(X_test): 使用transform方法对测试集的特征矩阵进行相同的标准化处理，使用的均值和方差参数来自于训练集；
X_train = X_train.reshape(X_train.shape[0], X_train.shape[1], 1): 调整训练集特征矩阵的形状，将其变为三维数组，通常用于适配某些深度学习模型（比如卷积神经网络）的输入要求；
X_test = X_test.reshape(X_test.shape[0], X_test.shape[1], 1): 同样地，调整测试集特征矩阵的形状，使其符合相同的三维格式要求；

四、构建网络

from tensorflow.keras.models import Sequential
from  tensorflow.keras.layers import Dense,LSTM,SimpleRNN

model = Sequential()
model.add(SimpleRNN(200,input_shape= (13,1),activation='relu'))
model.add(Dense(100,activation='relu'))
model.add(Dense(1,activation='sigmoid'))
model.summary()

可以得到如下输出：

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
_________________________________________________________________

我们调用了tf.keras.layers.SimpleRNN这个原型函数：

tf.keras.layers.SimpleRNN(units, activation='tanh', use_bias=True, kernel_initializer='glorot_uniform', recurrent_initializer='orthogonal', bias_initializer='zeros', return_sequences=False)

其中：

units: 整数，表示输出空间的维度（神经元的数量）；
activation: 激活函数的名称，可选，默认为 'tanh'。可以是内置的激活函数，也可以是自定义的激活函数；
use_bias: 布尔值，表示是否使用偏置项，默认为 True；
kernel_initializer: 权重矩阵的初始化方法，默认为 'glorot_uniform'，也称为 Xavier 初始化；
recurrent_initializer: 循环权重矩阵的初始化方法，默认为 'orthogonal'，用于循环连接的权重；
bias_initializer: 偏置项的初始化方法，默认为 'zeros'；
return_sequences: 布尔值，表示在输出序列中是否返回完整的序列，默认为 False。如果为 True，则返回整个输出序列；如果为 False，则只返回最后一个时间步的输出；

五、编译模型

通过下列示例代码：

opt = tf.keras.optimizers.Adam(learning_rate=1e-5)

model.compile(loss='binary_crossentropy',
              optimizer=opt,
              metrics="accuracy")

六、训练模型

通过下列示例代码：

epochs = 100

history = model.fit(X_train,y_train,
                    epochs = epochs,
                    batch_size = 128,
                    validation_data=(X_test,y_test),
                    verbose = 1)

运行得到如下输出：

Epoch 1/100
3/3 [==============================] - 1s 119ms/step - loss: 0.6851 - accuracy: 0.5728 - val_loss: 0.6811 - val_accuracy: 0.6452
Epoch 2/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6831 - accuracy: 0.5796 - val_loss: 0.6795 - val_accuracy: 0.6452
Epoch 3/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6829 - accuracy: 0.5784 - val_loss: 0.6779 - val_accuracy: 0.6452
Epoch 4/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6815 - accuracy: 0.5812 - val_loss: 0.6762 - val_accuracy: 0.6452
Epoch 5/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6796 - accuracy: 0.5947 - val_loss: 0.6745 - val_accuracy: 0.6452
Epoch 6/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6797 - accuracy: 0.6031 - val_loss: 0.6729 - val_accuracy: 0.6774
Epoch 7/100
3/3 [==============================] - 0s 7ms/step - loss: 0.6774 - accuracy: 0.6253 - val_loss: 0.6713 - val_accuracy: 0.7097
Epoch 8/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6760 - accuracy: 0.6394 - val_loss: 0.6697 - val_accuracy: 0.7097
Epoch 9/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6760 - accuracy: 0.6239 - val_loss: 0.6682 - val_accuracy: 0.7097
Epoch 10/100
3/3 [==============================] - 0s 7ms/step - loss: 0.6756 - accuracy: 0.6380 - val_loss: 0.6668 - val_accuracy: 0.7097
Epoch 11/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6743 - accuracy: 0.6564 - val_loss: 0.6655 - val_accuracy: 0.7097
Epoch 12/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6738 - accuracy: 0.6524 - val_loss: 0.6640 - val_accuracy: 0.7419
Epoch 13/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6714 - accuracy: 0.6631 - val_loss: 0.6625 - val_accuracy: 0.7419
Epoch 14/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6715 - accuracy: 0.6524 - val_loss: 0.6610 - val_accuracy: 0.7742
Epoch 15/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6694 - accuracy: 0.6850 - val_loss: 0.6595 - val_accuracy: 0.7419
Epoch 16/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6683 - accuracy: 0.6811 - val_loss: 0.6580 - val_accuracy: 0.7419
Epoch 17/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6688 - accuracy: 0.6598 - val_loss: 0.6565 - val_accuracy: 0.7419
Epoch 18/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6677 - accuracy: 0.6741 - val_loss: 0.6550 - val_accuracy: 0.7419
Epoch 19/100
3/3 [==============================] - 0s 7ms/step - loss: 0.6665 - accuracy: 0.7028 - val_loss: 0.6535 - val_accuracy: 0.7419
Epoch 20/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6665 - accuracy: 0.6949 - val_loss: 0.6520 - val_accuracy: 0.7419
Epoch 21/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6645 - accuracy: 0.7027 - val_loss: 0.6505 - val_accuracy: 0.7742
Epoch 22/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6632 - accuracy: 0.7143 - val_loss: 0.6489 - val_accuracy: 0.7742
Epoch 23/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6619 - accuracy: 0.7084 - val_loss: 0.6474 - val_accuracy: 0.8065
Epoch 24/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6624 - accuracy: 0.6997 - val_loss: 0.6459 - val_accuracy: 0.8065
Epoch 25/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6618 - accuracy: 0.7083 - val_loss: 0.6445 - val_accuracy: 0.8387
Epoch 26/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6601 - accuracy: 0.7169 - val_loss: 0.6431 - val_accuracy: 0.8387
Epoch 27/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6601 - accuracy: 0.7236 - val_loss: 0.6417 - val_accuracy: 0.8387
Epoch 28/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6577 - accuracy: 0.7283 - val_loss: 0.6402 - val_accuracy: 0.8387
Epoch 29/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6567 - accuracy: 0.7282 - val_loss: 0.6388 - val_accuracy: 0.8387
Epoch 30/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6585 - accuracy: 0.7394 - val_loss: 0.6374 - val_accuracy: 0.8387
Epoch 31/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6551 - accuracy: 0.7501 - val_loss: 0.6359 - val_accuracy: 0.8387
Epoch 32/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6549 - accuracy: 0.7462 - val_loss: 0.6345 - val_accuracy: 0.8710
Epoch 33/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6523 - accuracy: 0.7597 - val_loss: 0.6331 - val_accuracy: 0.8710
Epoch 34/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6523 - accuracy: 0.7671 - val_loss: 0.6317 - val_accuracy: 0.9032
Epoch 35/100
3/3 [==============================] - 0s 8ms/step - loss: 0.6512 - accuracy: 0.7749 - val_loss: 0.6303 - val_accuracy: 0.9355
Epoch 36/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6507 - accuracy: 0.7748 - val_loss: 0.6289 - val_accuracy: 0.9355
Epoch 37/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6512 - accuracy: 0.7641 - val_loss: 0.6275 - val_accuracy: 0.9032
Epoch 38/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6495 - accuracy: 0.7641 - val_loss: 0.6261 - val_accuracy: 0.9032
Epoch 39/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6499 - accuracy: 0.7659 - val_loss: 0.6247 - val_accuracy: 0.9032
Epoch 40/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6479 - accuracy: 0.7785 - val_loss: 0.6233 - val_accuracy: 0.9032
Epoch 41/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6483 - accuracy: 0.7640 - val_loss: 0.6218 - val_accuracy: 0.9032
Epoch 42/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6445 - accuracy: 0.7874 - val_loss: 0.6205 - val_accuracy: 0.9032
Epoch 43/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6447 - accuracy: 0.7747 - val_loss: 0.6192 - val_accuracy: 0.9032
Epoch 44/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6459 - accuracy: 0.7718 - val_loss: 0.6179 - val_accuracy: 0.9032
Epoch 45/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6435 - accuracy: 0.7765 - val_loss: 0.6165 - val_accuracy: 0.9032
Epoch 46/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6445 - accuracy: 0.7745 - val_loss: 0.6152 - val_accuracy: 0.9032
Epoch 47/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6443 - accuracy: 0.7695 - val_loss: 0.6138 - val_accuracy: 0.9032
Epoch 48/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6418 - accuracy: 0.7802 - val_loss: 0.6124 - val_accuracy: 0.9032
Epoch 49/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6410 - accuracy: 0.7783 - val_loss: 0.6110 - val_accuracy: 0.9032
Epoch 50/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6410 - accuracy: 0.7763 - val_loss: 0.6096 - val_accuracy: 0.9032
Epoch 51/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6392 - accuracy: 0.7831 - val_loss: 0.6082 - val_accuracy: 0.9032
Epoch 52/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6383 - accuracy: 0.7812 - val_loss: 0.6068 - val_accuracy: 0.9032
Epoch 53/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6387 - accuracy: 0.7657 - val_loss: 0.6053 - val_accuracy: 0.9032
Epoch 54/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6326 - accuracy: 0.8007 - val_loss: 0.6037 - val_accuracy: 0.9032
Epoch 55/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6375 - accuracy: 0.7694 - val_loss: 0.6023 - val_accuracy: 0.9032
Epoch 56/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6389 - accuracy: 0.7800 - val_loss: 0.6008 - val_accuracy: 0.9032
Epoch 57/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6328 - accuracy: 0.7994 - val_loss: 0.5993 - val_accuracy: 0.9032
Epoch 58/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6310 - accuracy: 0.7994 - val_loss: 0.5978 - val_accuracy: 0.9032
Epoch 59/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6297 - accuracy: 0.8034 - val_loss: 0.5962 - val_accuracy: 0.9032
Epoch 60/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6304 - accuracy: 0.7945 - val_loss: 0.5946 - val_accuracy: 0.9032
Epoch 61/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6256 - accuracy: 0.8023 - val_loss: 0.5930 - val_accuracy: 0.9032
Epoch 62/100
3/3 [==============================] - 0s 13ms/step - loss: 0.6312 - accuracy: 0.7883 - val_loss: 0.5915 - val_accuracy: 0.9032
Epoch 63/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6247 - accuracy: 0.8147 - val_loss: 0.5899 - val_accuracy: 0.9032
Epoch 64/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6240 - accuracy: 0.8059 - val_loss: 0.5884 - val_accuracy: 0.9032
Epoch 65/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6280 - accuracy: 0.7981 - val_loss: 0.5869 - val_accuracy: 0.9032
Epoch 66/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6249 - accuracy: 0.8069 - val_loss: 0.5854 - val_accuracy: 0.9032
Epoch 67/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6244 - accuracy: 0.7932 - val_loss: 0.5838 - val_accuracy: 0.9032
Epoch 68/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6209 - accuracy: 0.8137 - val_loss: 0.5822 - val_accuracy: 0.9032
Epoch 69/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6193 - accuracy: 0.8059 - val_loss: 0.5806 - val_accuracy: 0.9032
Epoch 70/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6199 - accuracy: 0.8020 - val_loss: 0.5790 - val_accuracy: 0.9032
Epoch 71/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6185 - accuracy: 0.8097 - val_loss: 0.5774 - val_accuracy: 0.9032
Epoch 72/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6165 - accuracy: 0.8126 - val_loss: 0.5759 - val_accuracy: 0.9032
Epoch 73/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6126 - accuracy: 0.8253 - val_loss: 0.5744 - val_accuracy: 0.9032
Epoch 74/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6141 - accuracy: 0.8214 - val_loss: 0.5729 - val_accuracy: 0.9032
Epoch 75/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6150 - accuracy: 0.8077 - val_loss: 0.5713 - val_accuracy: 0.9032
Epoch 76/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6177 - accuracy: 0.7980 - val_loss: 0.5698 - val_accuracy: 0.9032
Epoch 77/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6156 - accuracy: 0.8048 - val_loss: 0.5682 - val_accuracy: 0.9032
Epoch 78/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6132 - accuracy: 0.8068 - val_loss: 0.5666 - val_accuracy: 0.9032
Epoch 79/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6120 - accuracy: 0.8146 - val_loss: 0.5650 - val_accuracy: 0.9032
Epoch 80/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6077 - accuracy: 0.8136 - val_loss: 0.5635 - val_accuracy: 0.9032
Epoch 81/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6121 - accuracy: 0.8009 - val_loss: 0.5620 - val_accuracy: 0.9032
Epoch 82/100
3/3 [==============================] - 0s 11ms/step - loss: 0.6085 - accuracy: 0.8087 - val_loss: 0.5604 - val_accuracy: 0.9032
Epoch 83/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6060 - accuracy: 0.8136 - val_loss: 0.5587 - val_accuracy: 0.9032
Epoch 84/100
3/3 [==============================] - 0s 12ms/step - loss: 0.6073 - accuracy: 0.8010 - val_loss: 0.5571 - val_accuracy: 0.9032
Epoch 85/100
3/3 [==============================] - 0s 9ms/step - loss: 0.6042 - accuracy: 0.8088 - val_loss: 0.5553 - val_accuracy: 0.9032
Epoch 86/100
3/3 [==============================] - 0s 13ms/step - loss: 0.6068 - accuracy: 0.7950 - val_loss: 0.5535 - val_accuracy: 0.9032
Epoch 87/100
3/3 [==============================] - 0s 10ms/step - loss: 0.6001 - accuracy: 0.8136 - val_loss: 0.5516 - val_accuracy: 0.9032
Epoch 88/100
3/3 [==============================] - 0s 10ms/step - loss: 0.5991 - accuracy: 0.8048 - val_loss: 0.5498 - val_accuracy: 0.9032
Epoch 89/100
3/3 [==============================] - 0s 11ms/step - loss: 0.5977 - accuracy: 0.8107 - val_loss: 0.5481 - val_accuracy: 0.9032
Epoch 90/100
3/3 [==============================] - 0s 10ms/step - loss: 0.5990 - accuracy: 0.7970 - val_loss: 0.5465 - val_accuracy: 0.9032
Epoch 91/100
3/3 [==============================] - 0s 9ms/step - loss: 0.5981 - accuracy: 0.8165 - val_loss: 0.5448 - val_accuracy: 0.9032
Epoch 92/100
3/3 [==============================] - 0s 10ms/step - loss: 0.5960 - accuracy: 0.8213 - val_loss: 0.5431 - val_accuracy: 0.9032
Epoch 93/100
3/3 [==============================] - 0s 10ms/step - loss: 0.5971 - accuracy: 0.8126 - val_loss: 0.5413 - val_accuracy: 0.9032
Epoch 94/100
3/3 [==============================] - 0s 11ms/step - loss: 0.5944 - accuracy: 0.8115 - val_loss: 0.5395 - val_accuracy: 0.9032
Epoch 95/100
3/3 [==============================] - 0s 10ms/step - loss: 0.5964 - accuracy: 0.8066 - val_loss: 0.5376 - val_accuracy: 0.9032
Epoch 96/100
3/3 [==============================] - 0s 9ms/step - loss: 0.5940 - accuracy: 0.8165 - val_loss: 0.5357 - val_accuracy: 0.9032
Epoch 97/100
3/3 [==============================] - 0s 9ms/step - loss: 0.5911 - accuracy: 0.8087 - val_loss: 0.5338 - val_accuracy: 0.9032
Epoch 98/100
3/3 [==============================] - 0s 8ms/step - loss: 0.5889 - accuracy: 0.8125 - val_loss: 0.5319 - val_accuracy: 0.9032
Epoch 99/100
3/3 [==============================] - 0s 9ms/step - loss: 0.5901 - accuracy: 0.8096 - val_loss: 0.5299 - val_accuracy: 0.9032
Epoch 100/100
3/3 [==============================] - 0s 8ms/step - loss: 0.5876 - accuracy: 0.8105 - val_loss: 0.5279 - val_accuracy: 0.9032

Process finished with exit code 0

模型训练结果为：val_accuracy = 90.32%

六、模型评估

1.Loss与Accuracy图

import matplotlib.pyplot as plt

acc = history.history['accuracy']
val_acc = history.history['val_accuracy']

loss = history.history['loss']
val_loss = history.history['val_loss']

epochs_range = range(epochs)

plt.figure(figsize=(14, 4))
plt.subplot(1, 2, 1)

plt.plot(epochs_range, acc, label='Training Accuracy-Adam')
plt.plot(epochs_range, val_acc, label='Validation Accuracy-Adam')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')


plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss, label='Training Loss-Adam')
plt.plot(epochs_range, val_loss, label='Validation Loss-Adam')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')

plt.show()

得到的可视化结果：

七、个人理解

本项目为通过RNN来实现心脏病的预测，需要根据给定的CSV文件来实现该目标。由于CSV为表格类文件，故可能存在数据缺失的情况，这与之前的图片类数据有明显的差异，因此在进入网络模型训练前需要针对表格中的数据做一定的处理，由于初次接触表格数据，故只完成了数据是否为0的检查，之后的学习过程中将完善异常值处理等其他数据操作。