Eigenvector Procedure based on Weighted Preference Flows in Multicriteria Outranking Analysis

The outranking analysis has been frequently used to deal with the complex decisions involving qualitative criteria and imprecise data. So far, various versions of ELECTRE have been proposed for ranking alternatives in the outranking analysis. Among others, ELECTRE III has been widely used. A distillation procedure using a qualification index is proposed to rank alternatives from the valued outranking relation. A weakness of ELECTRE III, however, is to involve the arbitrariness in the selection of the discrimination threshold function for the distillation procedure. On the other hand, various variants of PROMETHEE are also proposed for the outranking analysis. PROMETHEE intends to be simple and easy to understand. A deficiency of PROMETHEE is that it does not take into account the preference intensity of alternatives in the in-preference flow and out-preference flow for each alternative. We propose a new preference ranking procedure based on eigenvector using the gweightedh in- and outpreference flows of each alternative in the outranking analysis. The basic idea of the procedure proposed here is that it should be better to outrank a gstrongh alternative than a gweakh one and, conversely, it is less serious to be outranked by a gstrongh alternative than by gweakh one in a PROMETHEE context. It has a completely different interpretation with the AHP (Analytic Hierarchy Process) since the components of the valued outranking relation matrix are neither ratios nor reciprocal as in the AHP.