A hybrid diversited attribute group decision-making method based on non-additive measure entropy weight and multi-granularity generalized fuzzy rough set with application in medical diagnosis
Zhang, Xinfang,Chen, Xiangtang,Zhao, Xixuan,Chu, Xiaoli
DOI: https://doi.org/10.1007/s13042-024-02487-8
2024-12-13
International Journal of Machine Learning and Cybernetics
Abstract:In practical scenarios of group decision-making (GDM), a wide range of attribute diversities and varied information structures are encountered. Existing studies have given less attention to scenarios involving multiple decision preferences and attribute weights based on hybrid discrepancy information. Therefore, to address GDM problems characterized by attribute diversity, data variability, and multi-decision scenarios, this research introduces a novel GDM approach based on multigranulation rough set (MGRS). The study introduces the concepts of multi-granular hybrid information systems (MGHIS) and single-granular hybrid information systems (SGHIS). Weight signals are defined at different levels through granular computing to account for varying decision-maker or expert weights in criterion set selection. And the interconnectedness of the attribute indicators is considered in the study, i.e., their non-additivity, and the attribute indicators are not completely independent of each other. Attribute weights are recalculated using the Choquet integral and the generalized Shapley index. Weighted fuzzy relations are then established based on object similarities under different attributes on an SGHIS, providing upper and lower approximations of fuzzy decision objects. A Weighted Fuzzy Rough Set (MWGFRS) is developed on the MGHIS considering expert weights. Subsequently, a GDM method based on the MWGFRS model is proposed for information fusion. The effectiveness and scientific validity of this method are demonstrated through its application to medical diagnosis decision-making in chronic kidney disease, validated through comparative experiments.
computer science, artificial intelligence