Development of a Forecasting Model Based on Fuzzy Time Series Combining Hedge Algebras with Particle Swarm Optimization

Abstract

This paper suggests a novel hybrid fuzzy forecasting model that uses particle swarm optimization (PSO) and hedge algebra to handle the non-determinism and uncertainty related to real-world time series data. In that case, the hedge algebra is utilized as a tool for partitioning the universe of discourse into intervals with different lengths corresponding to the semantic intervals which are determined from the linguistic terms. After partitioning data into intervals, the times series data are fuzzified into fuzzy sets, and fuzzy relationship groups are established based on these fuzzy sets. Finally, the proposed model is combined with PSO to determine the optimum length of each interval with the goal of improving forecasting accuracy. The proposed model's performance is implemented on two datasets from the University of Alabama and data of new confirmed cases of COVID-19. The experimental results show that the proposed model can not only attain lower forecasting errors but also provide good forecasting quality for both the first-order and high-order fuzzy time series, respectively.