Leveraging random forest regression for informed and risk-aware material selection in supercapacitor fabrication

Abstract The growing socio-economic complexity has intensified the challenge of making informed and reliable decisions, particularly under conditions of uncertainty and imprecise information. While fuzzy set extensions have been extensively applied to model vagueness and incorporate expert cognition, they often overlook the experts’ risk tolerance, a critical aspect of real-world decision-making. To address this limitation, this study introduces a novel decision-making framework that integrates the ( p , q )-fractional linear Diophantine fuzzy set model for capturing linguistic evaluations and applies the TODIM method, which is uniquely grounded in prospect theory and explicitly incorporates the risk appetite of decision-makers. A key enhancement of the proposed framework is the incorporation of a machine learning-based technique, Random Forest Regression (RFR), to objectively derive the weights of decision criteria, thereby minimizing subjectivity in the weighting process. The framework is employed to evaluate twenty-eight composites used in supercapacitor fabrication. To validate the robustness of the derived rankings, Monte Carlo simulations were performed under varying weight perturbations, demonstrating the stability and resilience of the results across 5000 randomized iterations. The findings underscore the reliability of the proposed RFR-TODIM-based hybrid approach and its superior capability in handling both data uncertainty and expert risk behaviour. A comparative assessment with conventional models further highlights its effectiveness and practical utility in material selection for advanced energy storage systems.