Multicriteria decision aiding under uncertainty: challenges, solutions and robustness concerns

Register For The Online Event Here (opens in new link)

Abstract

In a context of Multiple Criteria Decision Aiding (MCDA), which typically involves multiple uncertainties, we address two main concerns. The first is related to robustness, and the second to the temporal dimension. MCAP-mix, a multiple criteria aggregation procedure for mixed evaluations, was designed to deal with different types of uncertainty (more generally called information imperfections). This procedure admits imperfect data (stochastic, fuzzy, evidential or other) for the evaluation of alternatives according to different criteria. It is extended toward a more robust procedure where imperfections affecting the parameters of the procedure are treated. A brief literature review on the issue of robustness in multicriteria decision analysis will be presented, and the link between modeling approaches for different types of information imperfections and the robustness concern will be explored. On the temporal dimension, several studies were conducted to develop a multi-criteria, multi-period outranking method called MUPOM, as well as an extension of the well-known preference disaggregation method UTASTAR called UTASTAR-Tensors (or simply UTASTAR-T). Those studies will be briefly presented, and different decision problems in MCDA under uncertainty, including ranking, sorting and, more recently, ordered clustering, will be discussed.

The Speaker

Professor Ben Amor holds an MSc and a PhD in business administration, specializing in operations and decision support. Her research looks mainly at information imperfections, the modelling languages associated with them, how they are treated in multi-criteria decision analysis (MCDA) and the related treatment of additional information. She has applied her expertise in MCDA in various R&D projects for Defence R&D Canada - Valcartier, particularly with regards to risk analysis.