An Application of Enhanced Knowledge Models to Fuzzy Time Series

Abstract

Knowledge is usually employed by domain experts to solve domain-specific problems. Huarng was the first to embed knowledge into forecasting fuzzy time series (2001). His model involved simple calculations and offers better prediction results once more supporting information has been supplied. On the other hand, Chen first proposed a high-order fuzzy time series model to overcome the drawback of existing fuzzy first-order forecasting models. Chen’s model involved limited computing and came with higher accuracy than some other models. For this reason, the study is focused on these two types of models. The first model proposed here, which is referred to as a weighted model, aims to overcome the deficiency of the Huarng’s model. Second, we propose another fuzzy time series model, called knowledge based high-order time series model, to deal with forecasting problems. This model aims to overcome the deficiency of the Chen’s model, which depends strongly on highest-order fuzzy time series to eliminate ambiguities at forecasting and requires a vast memory for data storage. Experimental study of enrollment of University Alabama and the forecasting of a future’s index show that the proposed models reflect fluctuations in fuzzy time series and provide forecast results that are more accurate than the ones obtained when using the to two referenced models.