An improved fuzzy time series forecasting model using variations of data

Abstract

This study proposes an improved fuzzy time series (IFTS) forecasting model using variations of data that can interpolate historical data and forecast the future. The parameters in this model are chosen by algorithms to obtain the most suitable values for each data set. The calculation of the IFTS model can be performed conveniently and efficiently by a procedure within the R statistical software that has been stored in the AnalyseTS package. The proposed model is also used in the forecasting of two real problems in Vietnam: the penetration of salt and the total population. These numerical examples show the advantages of the proposed model in comparison with existing models and illustrate its effectiveness in practical applications.