MAXIMUM NUMBER OF DISTINCT RANKINGS IN MCDA PROBLEMS: A COMPARISON BETWEEN LINEAR AND NONLINEAR MODELS

This study investigates the ability of different aggregation models to generate distinct rankings in multi-criteria decision-making (MCDA) problems. Four methods are compared: the linear model, the Choquet integral (complete and 2-additive), and monotonic neural networks. Using Monte Carlo simulations with fixed decision matrices and varying model parameters, the number of attainable distinct rankings is estimated under different configurations of criteria and alternatives. The results indicate that nonlinear models generate a greater diversity of rankings, with emphasis on monotonic neural networks. The approach adopted in this article provides a complementary perspective on the behavior of the models and may contribute to the selection of aggregation methods in specific contexts.