Comparison between individual and hybrid approaches for estimating fuzzy time series

S.M.T. Fatemi Ghomi¹ S. Jaberi² M.Hajian-Heidary³ Moeen Sammak Jalali⁴

1) Full professor, Faculty of Industrial Engineering & Management Systems, Amirkabir University of Technology-Tehran Polytechnic, Tehran, Iran Email:
2) Faculty of Industrial Engineering & Management Systems, Amirkabir University of Technology-Tehran Polytechnic, Tehran, Iran Email:
3) Faculty of Industrial Engineering & Management Systems, Amirkabir University of Technology-Tehran Polytechnic, Tehran, Iran Email:
4) PhD Candidate, Faculty of Industrial Engineering & Management Systems, Amirkabir University of Technology-Tehran Polytechnic, Tehran, Iran Email:

Publication : 2nd. International Conference on Management & Industrial Engineering(2icmi.com)
Abstract :
Forecasting is one of the methods that helps decision makers to decide better about future. One of the best tools for forecasting is statistical methods. There are several criteria to measure the performance of forecasting methods that one of the most important of them is analysis the forecast error variance. Forecast combination methods are proposed to improve the accuracy of the forecasting. On the other hand, in practical applications, usage of fuzzy logic because of the vagueness and uncertainty is necessary. In the last decade, forecasting based on fuzzy time series has been used but the combination of fuzzy time series techniques to improving forecast accuracy has not been utilized. In current paper, to forecast linear process, fuzzy regression and other forecasting methods such as fuzzy double exponential smoothing are applied. After that combination of these three methods the forecast is utilized. The results imply that the forecast error variance in hybrid method is improved. These results are proved for a numerical example.
Keywords : Fuzzy time series Combination of forecasting methods Fuzzy regression.