A hybrid fuzzy time series forecasting model based on granular computing and bio-inspired optimization approaches

In this article, a novel M-factors fuzzy time series (FTS) forecasting model is presented, which relies upon on the hybridization of two procedures, viz., granular computing and bio-inspired computing. In this investigation, granular computing is utilized to discretize M-factors time series data set to obtain granular intervals. These intervals are additionally used to fuzzify the time series data set. Based on fuzzified time series data set, M-factors fuzzy relations are set-up. These M-factors fuzzy relations are further utilized to acquire forecasting results. Moreover, a novel bio-inspired algorithm is proposed to enhance the forecasting accuracy. The main objective of this algorithm is to adjust the lengths of the intervals (granular and non-granular intervals) in the universe of discourse that are used in forecasting. The proposed model is verified and validated with various real world data sets. Various statistical and comparative analyzes signify that the proposed model can take far better decision with the M-factors time series data sets. Moreover, empirical analysis demonstrates that forecasting accuracy of the proposed model based on granular intervals is better than non-granular intervals.