数据多维处理LSTM股票价格预测模型

Data Multidimensional Processing of LSTM Stock Price

股票的价格具有非线性、随机性等特征,为更精准地预测股票价格,充分利用股票价格数据的时间相关性和数据自身的变化趋势,提出数据多维处理的LSTM股票价格预测模型。通过对股票价格因子数据进行多维处理,提高数据有效信息,形成可以高度反映股票价格的多维数据,在此基础上建立长短期记忆网络组合预测模型,通过收集股市中的股票数据进行实验。实验结果表明,模型预测值与实际股价数据的均方根误差和平均绝对误差仅为0.0132和0.0103,相较于单一长短期记忆网络预测模型,2项误差分别降低90.81%和91.65%。数据多维处理LSTM股票价格预测模型具有较高的预测精度。

Stock prices are characterized by nonlinearity and randomness.In order to predict stock prices more accurately,an LSTM stock price prediction model with multi-dimensional data processing is proposed by making full use of the temporal correlation of stock price data and the variation trend of the data itself.By multidimensional processing the stock price factor data,the effective information of the data is improved,and the multidimensional data that can highly reflect the stock price is formed.On this basis,the combination prediction model of long-term and short-term memory network is established,and the stock data in the stock market is collected for experiments.The experimental results show that the root-mean-square error and the average absolute error between the predicted value of the model and the actual stock price data are only 0.0132 and 0.0103,which are 90.81%and 91.65%lower than the single prediction model of the long and short term memory network.The stock price prediction model of LSTM with data multidimensional processing has high prediction accuracy.