A simple time variant method for fuzzy time series forecasting

A simple time variant method for fuzzy time series forecasting

0.00 Avg rating0 Votes
Article ID: iaor20084203
Country: United Kingdom
Volume: 38
Issue: 3
Start Page Number: 305
End Page Number: 321
Publication Date: Mar 2007
Journal: Cybernetics and Systems
Authors:
Keywords: fuzzy sets, agriculture & food, education
Abstract:

Forecasting using fuzzy time series models needs computations of fuzzy relations in adjacent observations of time series data. In view of getting better forecasted values, these fuzzy relations have been considered as time invariant and time variant, and have been computed in several ways. However, the complication lies with the various rules developed for obtaining these fuzzy relations and then the defuzzification process. In this article, we propose a simple time variant method for time series forecasting, It uses the difference operator and the values obtained have been used for developing fuzzy rules for forecast. We develop algorithms to forecast enrollments of the University of Alabama and compared them with existing methods. The method has been also implemented to forecast rice production of Pantnagar (farm), India. The computational algorithms of the proposed method are simple and provide higher accuracy in forecasting.

Reviews

Required fields are marked *. Your email address will not be published.