Improving forecasting accuracy of crude oil prices using decomposition ensemble model with reconstruction of IMFs based on ARIMA model

Abstract

The accuracy of crude oil price forecasting is more important especially for economic development and is considered a lifeblood of the industry. Hence, in this paper, a decomposition-ensemble model with the reconstruction of intrinsic mode functions (IMFs) is proposed for forecasting the crude oil prices based on the well-known autoregressive moving average (ARIMA) model. Essentially, the reconstruction of IMFs enhanced the forecasting accuracy of the existing decomposition ensemble models. The proposed methodology works in four steps: decomposition of the complex data into several IMFs using EEMD, reconstruction of IMFs based on order of ARIMA model, prediction of every reconstructed IMF, and finally ensemble the prediction of every IMF for the final output. A case study is carried out using two crude oil prices time series (i.e. Brent and West Texas Intermediate (WTI)). The empirical results exhibited that the reconstruction of IMFs based on order of ARIMA model was adequate and provided the best forecast. To check the correctness, robustness and generalizability simulations were also carried out.