Narrow Results

Various methods have been presented to investigate the length of data interval and partition number of universe of discourse in fuzzy time series to achieve high level forecasting accuracy. However, the interval length is still an issue and thus, influencing the forecasting accuracy. This paper proposes a new approach using the average inter-quartile range to improve the interval length and subsequently the partition number of universe of discourse. Moreover, in forecasting method, the first-differencing data is also considered to obtain better forecast. The enrollment data of Alabama University is used as an example and the efficiency of the proposed method is compared with the previous methods. The result shows that the proposed method improves the accuracy and efficiency of the yearly enrollment forecasting opportunities.

To forecast the non-stationary data is quite difficult when compared with the stationary data time series. Because their variances are not constant and not stable like the second data type. This paper presents the implementation of fuzzy time series (FTS) into the non-stationary time series data forecasting, such as, the electricity load demand, the exchange rates, the enrollment university and others. These data forecasts are derived by implementing of the weightage and linguistic out-sample methods. The result shows that the FTS can be applied in improving the accuracy and efficiency of these non-stationary data forecasting opportunities.