LSTM做时间序列拟合预测

参考：LSTM的时序应用
修改了一丁丁点

import numpy as np
import pandas as pd
import math
from keras.models import Sequential
from keras.layers import Dense, LSTM
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_squared_error

def sample_split(df, is_scale=False):  # data processing
    train_size = int(len(df)*0.67)
    test_size = len(df) - train_size
    dataset = df.iloc[:,1:].values
    if is_scale:
        scaler = MinMaxScaler(feature_range=(0,1))
        dataset = scaler.fit_transform(dataset)
    train_set, test_set = dataset[0:train_size,:], dataset[train_size:len(dataset),:]
    train_x, train_y = train_set[:, 1:dataset.shape[1]], train_set[:,0]
    test_x, test_y = test_set[:, 1:dataset.shape[1]], test_set[:,0]
    return train_x, train_y, test_x, test_y

def lstm_predict(df, hidden_number):
    train_x, train_y, test_x, test_y = sample_split(df, is_scale=True)
    train_x = np.reshape(train_x, (train_x.shape[0], 1, train_x.shape[1]))
    test_x = np.reshape(test_x, (test_x.shape[0], 1, test_x.shape[1]))
    model = Sequential()
    model.add(LSTM(hidden_number, input_dim=df.values.shape[1]-2))
    model.add(Dense(1))
    model.compile(loss='mean_squared_error', optimizer='adam')
    model.fit(train_x, train_y, nb_epoch=1750, batch_size=int(train_x.shape[0]/3),verbose=2)
    train_predict, test_predict = model.predict(train_x), model.predict(test_x)
    scaler = MinMaxScaler(feature_range=(0,1))
    scaler_y = scaler.fit_transform(df.values[:,1].reshape(-1,1))
    train_y, test_y = scaler.inverse_transform([train_y]), scaler.inverse_transform([test_y])
    train_predict, test_predict = scaler.inverse_transform(train_predict), scaler.inverse_transform(test_predict)
    train_score = math.sqrt(mean_squared_error(train_y[0], train_predict[:,0]))
    print('Train Score:%.2f RMSE'%(train_score))
    test_score = math.sqrt(mean_squared_error(test_y[0], test_predict[:,0]))
    print('Test Score: %.2f RMSE'%(test_score))
    return np.vstack((train_predict, test_predict))

if __name__ == '__main__':
	np.random.seed(7)
	data = pd.read_csv('lstm_data.csv')
	display(data)   # 查看数据格式
	y_predict = lstm_predict(data, hidden_number=6)
	# 画图
	plt.figure(figsize=(10,4))
	plt.plot(y_predict, label='Pridict', alpha=0.8)
	plt.plot(y, label='Real', alpha=0.8)
	plt.legend()
	plt.ylabel('Price')
	plt.xlabel('Date')
	plt.title('BABA(LSTM Model)')

因变量是close_tomorrow