demo45.py

import numpy as np
from keras.layers import Dense
from keras.models import Sequential
from sklearn.model_selection import train_test_split

dataset1 = np.loadtxt("data\\diabetes.csv", delimiter=",", skiprows=1)
print(type(dataset1))
print(dataset1.shape)

# inputList = dataset1[:, 0:8]
# resultList = dataset1[:, 8]
inputList = dataset1[:, 0:-1]
resultList = dataset1[:, -1]

feature_train, feature_test, label_train, label_test = train_test_split(inputList, resultList, test_size=0.1)
print("input shape={}".format(inputList.shape))
print("result shape={}".format(resultList.shape))
print(np.unique(resultList))  # 0/1 only, classification
model = Sequential()
model.add(Dense(64, input_dim=8, activation='relu'))
model.add(Dense(32, activation='relu'))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
print(model.summary())

model.fit(feature_train, label_train, epochs=200, batch_size=20,
          validation_data=(feature_test, label_test))