Linguistic time series forecasting model for handling numeric time series’ uncertainty in an axiomatic hedge algebra-based formalism

The fuzzy time series forecasting model, introduced by Song and Chissom, aims to utilize the fuzzy sets’ uncertainty to handle the uncertainty of a given numeric time series, denoted by T N , to solve its numeric time series forecasting problem. The model attracts and strongly motivates researchers to develop or extend this model. Since, usually, experts observe T N in terms of its linguistic variable words, it suggests proposing a so-called linguistic time series to utilize the words and their inherent qualitative and quantitative semantics, formalized by the hedge algebras theory, for modeling T N and handling its values and its uncertainty to solve the T N ’s forecasting problem. For a given T N ’s and its declared multi-aspect semantics, the linguistic time series model permits to construct an intervals’ partition of T N ’s numeric universe, whose every subinterval is associated with a unique declared word, and to transform T N into a linguistic time series. Then, the study can develop a forecasting method to compute its desired forecasting values, where the words’ numeric semantics are axiomatically defined by a so-called quantifying mapping of the hedge algebra of the T N ’s linguistic variable. Particularly, the proposed model can also allow the declared word domain scalably growing to increase the exactness of the proposed forecasting method.