An Explainable Fuzzy Multi-Criteria Decision-Making Framework with SHAP-Guided Rule Extraction for Transparent Decision Support Under Uncertainty

Conventional fuzzy multi-criteria decision-making (MCDM) methods support ranking under uncertainty but often provide limited explanation of why alternatives are preferred. This study proposes an explainable fuzzy decision-making framework that integrates the Fuzzy Analytic Hierarchy Process (FAHP) and Fuzzy TOPSIS with surrogate modeling, SHAP-based analysis, and linguistic rule extraction. The main contribution is an explanation layer that preserves the original FAHP–FTOPSIS ranking structure while decomposing ranking scores into criterion-level contributions and transforming recurrent attribution patterns into IF–THEN rules. The framework is evaluated through a supplier-selection case study using expert fuzzy evaluations, local perturbation analysis, leave-one-supplier-out cross-validation, and a synthetic benchmark. The results show that the fuzzy MCDM layer produces discriminative rankings and that the top-ranked supplier remains comparatively stable under perturbations. Among the tested surrogates, the Random Forest Regressor achieved the strongest local fidelity, outperforming linear regression and a shallow decision tree. SHAP analysis showed ordinal alignment between FAHP weights and global criterion importance, while the extracted rules achieved high coverage, consistency, and threshold stability. The framework is useful for researchers, decision analysts, procurement managers, and supply chain professionals who require transparent, interpretable, and auditable multicriteria decisions under uncertainty.