Decision Rule Preference Model

Abstract

Abstract We present the methodology of Multiple‐Criteria Decision Aiding (MCDA) based on preference modeling in terms of “ if , then ” decision rules. The basic assumption of the decision rule approach is that the decision maker (DM) accepts to give preferential information in terms of examples of decisions and looks for simple rules justifying her decisions. An important advantage of this approach is the possibility of handling inconsistencies in the preferential information, resulting from hesitations of the DM. The proposed methodology is based on the elementary, natural and rational principle of dominance. It says that if action is at least as good as action on each criterion from a considered family, then is also comprehensively at least as good as . From a methodological point of view, the set of decision rules constituting the preference model is induced from the preferential information using a knowledge discovery technique properly modified, so as to handle the dominance principle. The mathematical basis of the decision rule approach to MCDA is the Dominance‐based Rough Set Approach (DRSA) developed by the authors. We present some basic applications of this approach, starting by multiple‐criteria classification problems, and then going through decision under uncertainty, hierarchical decision making, classification problems with partially missing information, problems with imprecise information modeled by fuzzy sets, until multiple‐criteria choice and ranking problems.