Fifty years of multiple criteria decision analysis: From classical methods to robust ordinal regression

Multiple Criteria Decision Analysis (MCDA) is a subfield of Operational Research that aims to support Decision-Makers (DMs) in the decision-making process through mathematical models and computational procedures. In this perspective, MCDA employs structured and traceable protocols to identify potential actions and the criteria for evaluating them. MCDA procedures aim to define recommendations consistent with the preferences of DMs for the specific decision problem at hand. These problems are generally formulated in terms of either choosing the best action, classifying actions into pre-defined and ordered decision classes, or ranking actions from best to worst. As the evaluation criteria are generally conflicting, the main challenge is to aggregate them into a mathematical preference model representing the DM value system. We review the development of MCDA over the past fifty years and describe its evolution with examples of distinctive methods. They are distinguished by the type of preference information elicited by DMs, the type of the preference model (criteria aggregation), and the way of converting the preference relation induced by the preference model in the set of potential actions into a decision recommendation. We focus on MCDA methods with a finite set of actions. References to specific application areas will be given. In the conclusion section, some prospective avenues of research will be outlined.