理解线性回归keras模型图

Question

I'm using Keras boston dataset, using single feature and trying to perform a linear regression model. I have normalized the input feature. The output plot appears to be a straight line and not aligned to the data distribution:

- what am I missing here? Skipped the lines to load and normalize data below.

model = Sequential()
model.add(Dense(units=1,input_dim=1, activation='sigmoid'))
model.compile(optimizer=optimizers.SGD(lr=0.0001,clipvalue=0.5), loss='binary_crossentropy', metrics=['mae','accuracy'])
history = model.fit(x_train, y_train, validation_split=0.25,batch_size=64, epochs=200,shuffle=True)
loss = model.evaluate(x_test,y_test,batch_size=None)
predict=model.predict(x_test)
plt.plot()
plt.plot(y_test, predict, 'b', x_test , y_test, 'k.')

Answer 1

I implemented simple linear regression model with keras.

from keras.models import Sequential
from keras.layers import Dense, Dropout
import numpy as np
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt

class Train:
    def __init__(self):
        self.__lr = 0.1
        self.x_train, self.y_train, self.x_test, self.y_test, self.y_scaler = self.load_data('train.csv')
        self.train()

This is the part of preprocessing as following:

    def load_data(self, fname):
        data = np.loadtxt(fname, skiprows=1, delimiter=',')

        x_data = data[:, 1:-1]
        x_scaler = MinMaxScaler(feature_range=(0, 1))
        x_data = x_scaler.fit_transform(x_data)

        y_data = data[:, [-1]]
        y_scaler = MinMaxScaler(feature_range=(0, 1))
        y_data = y_scaler.fit_transform(y_data)

        train_size = int(len(x_data)*0.7)
        test_size = len(x_data) - train_size
        x_train, x_test = x_data[0:train_size], x_data[train_size:len(x_data)]
        y_train, y_test = y_data[0:train_size], y_data[train_size:len(y_data)]

        return x_train, y_train, x_test, y_test, y_scaler

And, It's the training part include layers. You must use the relu and mse .
Please refer the code as following:

    def train(self):
        model = Sequential()
        model.add(Dense(128, input_dim=len(self.x_train[0]), activation='relu'))
        model.add(Dropout(0.2))
        model.add(Dense(64, activation='relu'))   
        model.add(Dropout(0.2))
        model.add(Dense(32, activation='relu'))   
        model.add(Dropout(0.2))
        model.add(Dense(1, activation='relu'))                        
        model.compile(loss='mse', optimizer='sgd', metrics=['mse'])
        model.fit(self.x_train, self.y_train, epochs=5000)
        result = model.evaluate(self.x_test, self.y_test)
        predictions = model.predict(self.x_test)
        predictions = self.y_scaler.inverse_transform(predictions)
        print (predictions)

        y_test = self.y_scaler.inverse_transform(self.y_test)

        plt.plot(y_test)
        plt.plot(predictions)
        plt.show()

You know it's the main:

if __name__ == "__main__":
    train = Train()

The result is