Second Order Intuitionistic Fuzzy Time Series Forecasting Model via Crispification

Abstract

Forecasting is one of the valuable tools in making decisions for proper planning. However, the classical statistical forecasting method cannot cater to data in linguistic variables. The evolvement of the fuzzy set has made forecasting for data in natural language possible. The fuzzy set has been generalized into the intuitionistic fuzzy set (IFS), which can better handle uncertainty. This paper proposes a fuzzy time series (FTS) forecasting model based on the second order intuitionistic fuzzy logical relationships (IFLR). The IFS were converted into crisp values using the crispification method before calculating the forecasted data. The historical data of students’ enrollment at the University of Alabama was adopted to illustrate the proposed model. Two main findings were obtained. Here, the forecasting model based on IFS has shown a better performance than the models based on the fuzzy set, while the second-order IFLR has produced better forecasting results than the first order IFLR.