In case of outlier(s) it is inevitable that the performance of the fuzzy time series prediction methods is influenced adversely. Therefore, current prediction methods will not be able to provide satisfactory accuracy rates for defuzzified outputs (predictions) when the data has outlier(s). In this study, not only to be able to sort out this problem but also to be able to improve the forecasting accuracy, we propose a combined robust approach for fuzzy time series by assessing how the prediction performance of the methods will be affected from the
1 Highlights A novel high-order fuzzy time series method is suggested. In the proposed method, intersection operators are utilized to deal with the excessive number of inputs. Feed forward neural network is used to determine fuzzy relations. Fuzzy c-means method is employed for fuzzification. The membership values are used to determine fuzzy relations instead of using cluster numbers. The superior performance of proposed method is demonstrated in real world applications and a simulation study.
ABSTRACTUsing non-stochastic models such as fuzzy time series forecasting models for time series analysis has been attracted attention by researchers in recent years. Fuzzy time series forecasting models do not need strict assumptions. However, conventional stochastic models need to satisfy some assumptions. In addition, fuzzy time series methods can be used if the observations of time series have uncertainty. Fuzzy time series approaches consist of three basic steps. These steps are; fuzzification of the crisp observations, identification of fuzzy relations and defuzzification, respectively. In the literature, there are many methods proposed by researchers that contribute to all these stages to obtain more accurate forecasting results. One of the weak features of fuzzy time series methods is that the membership values are not taken into consideration in forecasting process. This problem was eliminated for first order approaches by taking advantage of artificial neural networks to describe fuzzy relations. While determining of fuzzy relations, if inputs and outputs of the neural networks are membership values for the period t-1 and t respectively, the membership values are not ignored. However, if this approach is extended for high-order models, the number of inputs of neural networks will increase strongly. So, it will be very hard to train such of neural networks. In this study, a novel high-order fuzzy time series approach in which membership values are taken into account and artificial neural networks are employed to identify fuzzy relations is proposed. In the proposed method, intersection operators are utilized to deal with the excessive number of inputs. Also, fuzzy c-means method is employed for fuzzification. The evaluation of forecasting performance is performed by application the proposed method to well-known time series data sets and the obtained results are compared with those produced by other forecasting methods available in the literature. In addition, the superior performance of proposed method is supported by a simulation study.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.