Crude oil price forecasting based on the ARIMA and BP neural network combinatorial algorithm

The autoregressive integrated moving average (ARIMA) model is always used for price forecasting, but it is often difficult to obtain the nonlinear information. World crude oil prices have nonlinear and time-varying characteristics, so an ARIMA and BP neural network combinatorial algorithm is presented for world crude oil price forecasting. The ARIMA model is put forward for linear information. The nonlinear message is in the residual series, so the BP neural network is used to simulate the residual series and get the nonlinear information. By comparing the crude oil price prediction results of the ARIMA model, the BP neural network model and the combinatorial algorithm, it is shown that the combinatorial algorithm has the highest prediction precision. © 2012 American Society of Civil Engineers.