An effective rolling decomposition-ensemble model for gasoline consumption forecasting. (1st May 2021)

• Record Type: Journal Article
• Authors: Yu, Lean; Ma, Yueming; Ma, Mengyao
• Is Part Of: Energy. Volume 222(2021)
• Publication Date: 2021-05-01
• Subjects: Gasoline consumption forecasting -- Decomposition-ensemble model -- Ensemble empirical mode decomposition -- Wavelet decomposition -- Support vector regression -- Rolling mechanism; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2021.119869
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: In this paper, an effective rolling decomposition-ensemble model is proposed for quarterly gasoline consumption forecasting in China. In this model, three steps, data decomposition, component prediction and ensemble output, are involved. In the data decomposition, wavelet decomposition and ensemble empirical mode decomposition are used due to few assumptions and excellent performance. In the component prediction, support vector regression is adopted due to the global approximation capability for data scarcity issue. In the ensemble output, the simple addition strategy is used for final aggregation. In order to solve the illusion of high prediction accuracy caused by the decomposition of the test dataset, the rolling decomposition and forecasting mechanism are adopted in this methodology. For illustration and verification purpose, 30 provincially quarterly gasoline consumption data in China are used. The experimental results demonstrate the effectiveness and robustness of the proposed rolling decomposition-ensemble model for gasoline consumption forecasting in terms of the accuracy of level and directional prediction. Highlights: A rolling decomposition-ensemble model is proposed for gasoline forecasting. The rolling decomposition and forecasting mechanism are adopted in this model. Data decomposition, component prediction and ensemble output are involved. Empirical analysis verifies the effectiveness and robustness of the proposed model.