EXTENSIONS OF UNCERTAINTY MEASURES IN MULTI-CRITERIA DECISION-MAKING METHODS AND CLASSIFICATION

In multi-criteria decision-making, there are different approaches to determining the weights before using any techniques (e.g., analytic hierarchy process and fuzzy entropy methods). In this paper, we integrate machine learning techniques with multi-criteria decision-making to address two distinct challenges: the selection of optimal phase transition materials and the classification of diabetes data. For the multi-criteria decision-making problem, we utilize extended entropy functions—including entropy, fractional entropy, and Tsallis entropy—to calculate criteria weights based on non-probabilistic and probabilistic aspects. These weights guide the selection process using multi-objective optimization methods like ratio analysis and complex proportional assessment, aimed at identifying materials with superior thermal performance at minimal cost for latent heat thermal energy storage systems. Empirical results validate the effectiveness of our proposed strategies in phase transition material selection and highlight their advantages when compared to the technique for order performance by similarity to the ideal solution. Additionally, a classification problem for diabetes data is addressed using pattern recognition, demonstrating the synergy between machine learning and multi-criteria decision-making in tackling diverse decision-making challenges.