Stock price prediction based on ARIMA-RNN hybrid model
Improving the prediction accuracy of a time series model required a comprehensive understanding of the linear and nonlinear composite characteristics of its data,and this paper used ARIMA and RNN models to model the time series respectively,and mined the linear and nonlinear laws,and finally obtained the comprehensive prediction results of the two models.This paper selected the K-line data of all trading days of the CSI 300 Index(000300)from January 4,2006 to November 26,2021 as the sample,and the analysis results showed that the accuracy of the ARIMA-RNN hybrid model was higher than that of the single recurrent neural network model,and the hybrid model had a higher effect on short-term dynamic and static prediction,which was conducive to investors and enterprises to make more scientific and feasible decisions.
国家科技期刊平台
NSTL
万方数据
