Designing Intuitionistic Fuzzy Forecasting Model Combined With Information Granules and Weighted Association Reasoning

Abstract

A time series is a sequence of observations that a variable takes with respect to times. It has a wide range of applications in decision making and forecasting in economics, agriculture, medicine, industry, energy sector and other scientific fields. Time series modeling and forecasting contain some of the classical issues that are widely addressed in the literature based on traditional statistical models with low interpretability. Fuzzy time series has become a powerful tool that can counter the problem of prediction of historical data in linguistic terms. This study proposes a new framework for modeling the fuzzy time series approach in the environment of intuitionistic fuzzy set theory to play viable role in ensuring robustness to the uncertainty involved in data series. In order to get the optimized length of intervals, the principles of fuzzy c-means (FCM) clustering and information granules are integrated. To fuzzify the historical data, intuitionistic fuzzy triangular function is practiced to acquire the intuitionistic fuzzy sets. Furthermore, the distance measures between the elements of the intuitionistic fuzzy set of the fuzzified historical data and the centers of the corresponding clusters are computed for all fuzzy sets. Finally, a robust fuzzy time series model is designed by extracting fuzzy logical relationships and employing weighted association reasoning as an exhaustive defuzzification approach. The parameters of accuracy measures such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) are used to identify the strength of the proposed modeling and forecasting. Findings demonstrate that the proposed forecasting method is robust in determining the highly accurate forecasts.

Keywords

• Intuitionistic fuzzy set
• FCM clustering
• information granules
• distance measures
• forecasting