| Popis: |
Fuzzy time series forecasting model was first introduced by Song and Chissom, in 1993. Then, Chen examined this model, in 1996, applying fuzzy rules in the form A i → A j , where A i and A j are fuzzy sets associated with linguistic labels representing their corresponding values in a time series. Since then, many methods have been proposed to improve the accuracy of time series forecasting results or decrease processing times. In this study, we propose another approach dealing directly with human words with their own inherent semantics. That is, we assume that the above rules are of form X i → X j , in which X i and X j are human expert words describing their corresponding time series values. Therefore, we should deal with a so-called linguistic time series forecasting model using the inherent semantics of words and their quantitative semantics based on the hedge algebras formalism. A comparative experiment is made to show the usefulness of the proposed model. |
