Determining the preference threshold for a multicriteria decision support system through evolutionary modelling.

Abstract

This article uses the concepts of Genetic Algorithms and Transgenetic Algorithms associated to a Multicriteria Decision Support System for discrete alternatives, with the aim of better defining an important parameter of ranking, in this case, the preference threshold, p. This is one of the crucial parameters for the application of multicriteria methods of the French School of Multicriteria Decision Support that are embedded in that system. The use of Genetic Algorithms was shown to be an extremely useful source for helping decision analysts in the choice of values for the thresholds of transitivity. For future studies, an investigation of the possibility of using Genetic Algorithms to study variations in criteria weights in detail and/or the value of discordances is recommended. In addition, studies could be carried out on the pertinences of criteria weights in the context of the use of this fuzzy parameter used by Decision Support Systems.