PCA-ARIMA-BP hybrid model, the Shanghai Composite Index, risk-averse
Abstract
A PCA-ARIMA-BP hybrid model is proposed to study the Shanghai Composite Index. The model is based on principle component analysis (PCA), autoregressive integrated moving average model (ARIMA), and backward propagation (BP) neural network. We use data mining methods to select data. BP neural network, PCA-BP model and PCA-ARIMA-BP hybrid model prediction results are compared. The results show that the PCA-ARIMA-BP hybrid model can effectively improve the prediction precision. This can guide investors to avoid risks and improve benefit.