Explainable Multi-criteria Decision Making for tourism economics: integrating XAI with MCDM for a robust accommodation performance assessment

This paper presents a groundbreaking integration ofMultiple CriteriaDecision Making (MCDM) with explainable artificial intelligence (XAI) for tourism accommodation performance assessment, addressing fundamental limitations in traditional preference elicitation methods. We introduce the XAI-Enhanced MCDM Convergence Theorem that establishes theoretical foundations for combining classical MCDM methods with machine learning explanations, providing objective, data-driven criterion weights that eliminate subjective bias inherent in expert judgments. Ourmethodology extendsTOPSIS, PROMETHEE, and AHP by incorporating Shapley values, Integrated Gradients, and Expected Gradients to derive interpretable multi-criteria rankings. Applied to Lower Aosta Valley accommodation data, our framework demonstrates 18% improvement in ranking accuracy over traditional MCDM approaches while revealing critical sustainability threshold effects previously undetected. The proposed XAI-enhanced framework addresses the longstanding challenge of criterion weight elicitation in MCDM through empirically-derived attribution scores, representing a paradigm shift from subjective to objective multi-criteria analysis in economic decision-making contexts.