A New Hesitant Fuzzy-Based Forecasting Method Integrated with Clustering and Modified Smoothing Approach

This study provides a new fuzzy time series method of forecasting integrated with the theory of fuzzy clustering, exponential smoothing approach and hesitant fuzzy information. In most of the existing studies, intervals with equal and unequal lengths have been used individually to handle the complex decision-making problems. This paper adopts the Fuzzy c-means (FCM) clustering approach to partition the universe of discourse into both equal and unequal space intervals to improve the prediction quality. Hesitant fuzzy sets (HFSs) are developed and an aggregation operator is applied to aggregate the hesitant information. In next process, modified smoothing approach is implemented to defuzzify the historical data. To verify and validate the effectiveness of the proposed method, it is applied to forecast the benchmark data set: enrollments of the Alabama University. Empirical results signified the improved accuracy of the proposed method based on statistical accuracy measure techniques and it reveals that proposed method generates better forecasting outputs than all existing models.