Abstract

Many researchers are deploying fuzzy time series model for forecasting purposed including rainfall forecasting. However, the limited division size of intervals, ui to obtain subintervals in that model, influences the accuracy of rainfall forecasting. Hence, in this study the model is enhanced and combined with sliding window algorithm to propose a forecasting model of fuzzy time series sliding window. The proposed model is enhanced in division of intervals, ui, where several types of division groups of intervals are applied to obtain the better subintervals of model which can increase the accuracy of forecasting result. Another enhancement is introduced which identifies temporal prediction values that will produce the final result of forecasting. The function of subintervals is used to categorize temporal prediction values to observe the trend of rainfall forecasting based on rules of forecasting. In order to validate the proposed model, an error measurement of root-mean-squared error is deployed. The RMSE values between several different divisions of intervals in proposed model and previous study of sliding window algorithm are compared. The result shows that the proposed model, with division of interval groups of 5, 4, 3, 2, and remain unchanged, is better among them. The proposed model is suggested to be tested with other types of data for forecasting.