Forecasting by fuzzy double exponential smoothing model

Abstract

The exponential smoothing model is a popular tool in short-term forecasting. However, the smoothing constant is arbitrary and is determined by a decision maker in both the nature and perception of the unknown system structure to make the forecasting of exponential smoothing model ineffective. Therefore, a fuzzy exponential smoothing model is proposed for short-term forecasting where its optimal smoothing constant could be obtained easily and efficiently, whereas the trend for the collected data is yet to be considered. In order to cope with this problem, a fuzzy double exponential smoothing model will be derived to enhance and enlarge the abilities of the short-term fuzzy forecasting tools. Finally, a forecasting example of Taiwan internet users is illustrated to describe the performance of the proposed model.

Keywords:

• Forecasting
• Fuzzy exponential smoothing model
• Fuzzy polynomial trend analysis
• Fuzzy double exponential smoothing model

Acknowledgement

The author acknowledges the financial support of the National Science Council, Taiwan R.O.C., project number NSC 90-2218-E-364-001.