keras深度学习笔记

import numpy as np 
import pandas as pd 
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow.keras import models,layers
 
dftrain_raw = pd.read_csv('C:/Users/xxx/Desktop/365/d.txt',header=None)
 
dftest_raw = pd.read_csv('C:/Users/xxx/Desktop/365/tt.csv',header=None)
#dftrain_raw.head(10)
#print(dftrain_raw[22])
 
# 数据预处理
def preprocessing(dfdata):
 
    dfresult= pd.DataFrame()
    dfresult[0] = dfdata[0]
    dfresult[1] = dfdata[1]
    dfresult[2] = dfdata[2]
    dfresult[3] = dfdata[3]
    dfresult[4] = dfdata[4]
    dfresult[5] = dfdata[5]
    dfresult[6] = dfdata[6]
    dfresult[7] = dfdata[7]
    dfresult[8] = dfdata[8]
    dfresult[9] = dfdata[9]
    dfresult[10] = dfdata[10]
    dfresult[11] = dfdata[11]
    dfresult[12] = dfdata[12]
    dfresult[13] = dfdata[13]
    dfresult[14] = dfdata[14]
    dfresult[15] = dfdata[15]
    dfresult[16] = dfdata[16]
    dfresult[17] = dfdata[17]
    dfresult[18] = dfdata[18]
    dfresult[19] = dfdata[19]
    dfresult[20] = dfdata[20]
    dfresult[21] = dfdata[21]
    return(dfresult)
 
 
 
x_train = preprocessing(dftrain_raw)
print(x_train.shape)
y_train = dftrain_raw[22]
 
 
x_test = preprocessing(dftest_raw)
y_test = dftest_raw[22]
 
 
tf.keras.backend.clear_session()
 
model = models.Sequential()
model.add(layers.Dense(25,activation = 'relu',input_shape=(22,)))
model.add(layers.Dense(12,activation = 'relu' ))
model.add(layers.Dense(1,activation = 'sigmoid' ))
 
model.summary()
 
 
# 二分类问题选择二元交叉熵损失函数
model.compile(optimizer='adam',
            loss='binary_crossentropy',
            metrics=['AUC'])
 
history = model.fit(x_train,y_train,
                    batch_size= 64,
                    epochs= 100,
                    validation_split=0.2 #分割一部分训练数据用于验证
                   )
#测试集上的效果：
 
model.evaluate(x = x_test,y = y_test)
 
model.predict(x_test[0:18])