Handling imperfect information in multiple criteria decision-making through a comprehensive interval outranking approach

Abstract

In multi-criteria decision-making, the elicitation of the decision model parameters and the assessment of criteria performance levels usually cannot avoid uncertainty, imprecision, ill-determination, and arbitrariness. This imperfect knowledge can be modeled by using interval numbers. This paper presents a wide extension of the outranking approach to the interval framework. It is a comprehensive methodological proposal which makes it possible to combine interval information in some criteria, weights, veto and majority thresholds, and information as pseudo-criteria based on discriminating thresholds (including ordinal information) for the remaining criteria. The generalized approach keeps the advantages of both models of imperfect knowledge on criteria performance levels, also allowing the interval modeling of the outranking model's parameters. Two multi-criteria ordinal classification methods that can handle imprecise model parameters, interval criteria and pseudo criteria, are presented on this integrated framework. The potential of the proposed methodological advance is illustrated in a decision problem where the public-sector performance of the Organization for Economic Co-operation and Development (OECD) countries is assigned to a set of ordered classes.