%0 Journal Article
%T A New Method for Short Multivariate Fuzzy Time Series Based on Genetic Algorithm and Fuzzy Clustering
%A Kamal S. Selim
%A Gihan A. Elanany
%J Advances in Fuzzy Systems
%D 2013
%I Hindawi Publishing Corporation
%R 10.1155/2013/494239
%X Forecasting activities play an important role in our daily life. In recent years, fuzzy time series (FTS) methods were developed to deal with forecasting problems. FTS attracted researchers because of its ability to predict the future values in some critical situations where most standard forecasting models are doubtfully applicable or produce bad fittings. However, some critical issues in FTS are still open; these issues are often subjective and affect the accuracy of forecasting. In this paper, we focus on improving the accuracy of FTS forecasting methods. The new method integrates the fuzzy clustering and genetic algorithm with FTS to reduce subjectivity and improve its accuracy. In the new method, the genetic algorithm is responsible for selecting the proper model. Also, the fuzzy clustering algorithm is responsible for fuzzifying the historical data, based on its membership degrees to each cluster, and using these memberships to defuzzify the results. This method provides better forecasting accuracy when compared with other extant researches. 1. Introduction Time series are widely observed in many aspects of our lives; therefore, the prediction of future values based on the past and present information is very useful. In practice, there are several emergent domains that require dealing with short multivariate time series. As a consequence, the prediction of such time series arises in many situations. There are many existing techniques that are well proven in forecasting with multivariate time series data, but they put constraints on the minimum number of observations and require distribution assumptions to be made regarding the observed time series. Fuzzy time series (FTS) models have become increasingly popular in recent years because of their ability to deal with time series data without the need for validating any theoretical assumptions. However, how to select the proper model, how to partition the universe of discourse and determine effective lengths of intervals objectively to fuzzify the numerical data, and how to defuzzify the results are still open critical issues. These issues are very important and affect the model accuracy. The paper probes into these three questions in the modeling of FTS. The new method incorporates the fuzzy clustering and genetic algorithms (GA) with FTS to reduce its subjectivity and improve its accuracy. More specifically, the new method uses the integer genetic algorithm to search for the optimal model that fits the available data. (In this paper, to solve integer optimization problems, we used the MI-LXPM
%U http://www.hindawi.com/journals/afs/2013/494239/