CREATING ENHANCED REGRESSION MODELS: STRATEGIES FOR BLENDING FUZZY AND CRISP INPUTS

Abstract

Linear regression models play a crucial role in capturing the linear relationships between response and predictor variables, relying on specific assumptions. These assumptions encompass the availability of sufficient data, the validity of the linear relationship, the exactness of the connection, and the presence of precise data for both variables and coefficients. However, when these assumptions cannot be met, fuzzy regression models provide a practical and flexible alternative. The concept of fuzzy linear regression was initially introduced by Tanaka et al. in 1982 and has since been extended and refined by various researchers. This paper explores the realm of fuzzy regression modeling, tracing its evolution and development through contributions from authors like Tanaka, Lee, Diamond, D’Urso, Yang, Gonzalez-Rodriguez, Choi, Yoon, and Massari. Fuzzy regression offers a robust approach to modeling relationships when traditional linear regression assumptions do not hold, making it a valuable tool in various real-world scenarios.