Previous Chapter Chapter 38: Fuzzy Time Series Models Next Chapter
Chapter 38: Fuzzy Time Series Models
Conventional time series models have long been studied and applied to forecasts. Fuzzy time series models, as a counterpart of the conventional time series models, were proposed by Song and Chissom (1993). Following that study, many relevant studies have been proposed to forecast various problems. Interestingly, many of the forecasting results from these studies have been shown to outperform their conventional counterparts.
Fuzzy time series models can be univariate, bivariate, or multivariate models, which means that we can use one variable, two variables, or multiple variables to forecast. In addition, the models can be autoregressive of order 1 (AR(1)), which means we can use the observation at t − 1 to forecast the one at t. In general, the ...