Forecasting with Improved Model of Fuzzy Time Series Based on Hedge Algebras

Abstract

The aim of this paper is to introduce a new model of fuzzy time series (FTS) based on hedge algebras (HA) with three improvements compared with previous ones. Firstly, the new way of partitioning the universe of discourse is applied, secondly, the relationship groups following time points to determine relationships among fuzzy time series values are used, thirdly, weighs following time points are employed to build formula for computing forecasting values. The empirical results indicate that the proposed model outperform the existing models.