A new method of time series forecasting using intuitionistic fuzzy set based on average-length

Abstract

ABSTRACT The major problem in the field of fuzzy time series (FTS) is the accuracy rate in the forecasted values. To overcome this problem here, we propose a model for intuitionistic FTS forecasting based on average-length of interval, which enhances the forecasting result. The proposed model is focused on how to fuzzify the historical time series data. Here, the fuzzification of each observation is intuitionistic fuzzification, which is based on the maximum degree of score function and also establishes intuitionistic fuzzy logical relationships (IFLR) among all intuitionistic fuzzified data set. Here, we use simple arithmetic computations in defuzzification process with measuring the frequency of IFLR. An illustrative example of enrollments at the University of Alabama is used to verify the effectiveness of the proposed model and comparison in terms of RMSE and AFE with some of the existing forecasting models to show its superiority.