Multiobjective optimization approach for preference-disaggregation analysis under effects of intensity

Abstract

A widely used approach in Multicriteria Decision Aid is the Preference Disaggregation Analysis (or PDA). This is an indirect approach used to characterize the decision process of a Decision Maker (or DM). By means of a limited set of examples (called a reference set) provided by the DM, the PDA approach estimates the parameter values of a preference model that is characterized by the DM. This paper proposes a new optimization model for PDA, and its solution through an evolutionary algorithm. The novel features in the definition of the model include the use of the effect of the intensity (i. e. the variations among the criteria values used to evaluate decision alternatives), and new ways to combine the number of consistencies and inconsistencies with respect to the reference set. Through an experimental design performed to evaluate the fitness of the new model, it was corroborated its effectiveness to fit the DM preferences, and also it showed comparable results with that provided by an state-of-the-art strategy.