18
Vol. 2, No.1, Summer 2013 © 2012 Published by JSES.
A NEW FUZZY TIME SERIES ANALYSIS APPROACH BY USING
DIFFERENTIAL EVOLUTION ALGORITHM AND
CHRONOLOGICALLY-DETERMINED WEIGHTS
*
Vedide Rezan USLU
a
, Eren BAS
b
, Ufuk YOLCU
c
, Erol EGRIOGLU
d
Abstract
Fuzzy time series approaches, which do not require the strict assumptions
of traditional time series approaches, generally consist of three stages.
These stages are called as the fuzzification of crisp time series
observations, the identification of fuzzy relationships and the
defuzzification, respectively. All of these stages play an important role on
the forecasting performance of the model. By this study we want to
contribute to the stage of fuzzification so that the interval length is
determined by using the differential evolution algorithm and also we take
into account chronological-determined weights in the stage of
defuzzification.
Keywords: Fuzzy time series, Fuzzification, Differential evolution algorithm, Forecasting
JEL Classification: C53 - Forecasting and Prediction Methods
Authors’ Affiliation
a
- Associate Professor, University of Ondokuz Mayis, Turkey, e-mail: rezzanu@omu.edu.tr(corresponding
author)
b
- Research Assistant, Giresun University, Turkey, e-mail: eren.bas@giresun.edu.tr
c
- Assistant Professor, Giresun University, Turkey, e-mail: ufuk.yolcu@giresun.edu.tr
d
- Associate Professor, University of Ondokuz Mayis, Turkey, e-mail: erole@omu.edu.tr
*An earlier version of this paper was presented at The 6th International Conference on Applied Statistics,
November 2012, Bucharest.