Stock price prediction based on ARIMA-PCA-BP hybrid model

Abstract

Stock price prediction has important practical significance and has always been a hot topic. Stock prices are affected by a variety of different factors and are highly dimensional and nonlinear, which makes forecasting difficult. Common stock price prediction models include statistical prediction models and machine learning models. However, the statistical prediction models lack the consideration of nonlinear information. Moreover, the information redundancy of stock price factors will reduce the learning rate of neural network. These defects can reduce the accuracy of prediction. In order to improve the prediction accuracy of stock price, this paper proposes a hybrid model based on autoregressive integrated moving average (ARIMA) and back propagation (BP) neural network, and the BP neural network is optimized by principal component analysis (PCA) algorithm. The empirical analysis of 5 stocks in different industries is carried out. The results show that compared with the single ARIMA model or BP neural network model, the ARIMA-PCA-BP hybrid model improves the prediction accuracy of stock prices and has good learning ability and generalization ability.