The two-layer SRU Neural Network Based Analysis of Time Series

Abstract

The traditional measurement model cannot meet the processing requirements of non-linear and high-noise time series data. A two-layer simple recurrent unit (DSRU) prediction model based on BPSO and SO-PMI sentiment analysis algorithm is proposed. First, the model uses the SO-PMI algorithm to calculate the stock market sentiment tendency measurement index. Then the BPSO algorithm is used to extract the characteristics of stock technical indicators with greater stock price correlation. Finally, the DSRU network model is used to predict the stock price. Taking Shanghai Pudong Development Bank stock as an example, the results show that the model has a significant improvement in prediction accuracy, reaching 65.84%, with an average absolute error of 0.0541, which proves the superiority and effectiveness of the method.