Clustering fuzzy data by hedge algebra and regression approach

Abstract

Fuzzy clustering has been extensively explored across various methodologies, yielding diverse results within the realm of data mining. The plethora of research outcomes underscores the complexity inherent in fuzzy data mining, particularly when confronted with diverse data types aiming to delineate objects' affiliation with specific clusters. This intricacy is further compounded by the ubiquity of incomplete data, commonly referred to as missing data, posing a formidable challenge in this domain. Addressing the missing value predicament becomes imperative for a more nuanced and accurate enhancement of fuzzy clustering. In response to these challenges, a novel approach has emerged, leveraging the synergies between hedging algebra and the linear regression model. This innovative methodology seeks to overcome the intricacies associated with diverse data types and missing values. By integrating algebraic principles with linear regression techniques, the proposed method introduces a robust framework for classifying objects within a cluster. The fusion of these mathematical tools offers a unique solution that not only navigates the complexities of fuzzy data mining but also addresses the pervasive issue of missing data. The paper delves into the advantages of adopting hedging algebra and the linear regression model in tandem, presenting a comprehensive methodology that significantly contributes to the refinement of fuzzy clustering. The collaborative interplay of algebraic principles and regression models not only enhances the accuracy of object classification within clusters but also provides a robust strategy for handling missing values in the dataset. This integrated approach represents a noteworthy advancement in the field of fuzzy clustering, offering a more comprehensive and effective solution to the intricate challenges posed by diverse data types and the prevalent issue of missing data.