A novel forecasting framework based on hesitant fuzzy time series and multi-objective optimization algorithm for natural gas futures price

Studying and forecasting price fluctuations in natural gas futures can leverage the anticipatory and leading role of futures themselves, thereby assisting policymakers in making better judgments. However, the strong volatility and uncertainty of nature gas futures price series increase the difficulty of price forecasting. Most of the commonly forecasting strategy ignore the fuzziness of series itself and the uncertainty of forecasting results that causes the unsatisfactory results. Therefore, a novel forecasting framework was proposed to realize natural gas futures price forecasting. In the proposed forecasting framework, the hesitant fuzzy time series combined three different interval partition strategies and the optimal weights are determined by multi-objective golden eagle optimizer. Then, the proposed forecasting framework are compared with single interval partition strategy, different optimization algorithms and traditional forecasting models. The results indicate the excellent performance of our proposed forecasting framework with MAPE 1.2778 % and 1.4648 % respectively for closing and opening price forecasting. Moreover, when data fluctuates by 5 %, forecasting accuracy indicators such as MAPE fluctuate by no more than 0.05 % which verified the great robustness of the proposed framework in natural gas futures price forecasting and analysis.