Energy Price Index Prediction via Information Granulation and Fuzzy Time Series Model

Capturing the trend and main variation ranges of the energy price index is crucial for energy management and decision-making. This study designs a forecasting method of the energy price index based on information granule and fuzzy time series model. Firstly, the principle of justifiable granularity is applied to a collection of interval-based information granules for characterizing the main variation interval of the energy price index. Next, the fuzzy c-means clustering algorithm based on Hausdorff distance is employed to fuzzify the granular time series. Then, the fuzzy rules based on granular time series are established, and a fuzzy inference system is constructed to forecast the future variation ranges. The experimental results on four energy price index data demonstrate the practicality and effectiveness of the proposed granular fuzzy time series prediction method.