Speaker
Description
In multiple criteria decision aid (MCDA), three traditional problem settings are typically distinguished: choice, ranking, and sorting. More recently, attention has turned to the multicriteria clustering problem, in which alternatives are assigned to groups in an unsupervised manner, meaning that categories are not predefined, unlike in sorting approaches. The objective of these methods is to form homogeneous and well-separated clusters.
In this contribution, we investigate the multicriteria anti-clustering (MCAC) problem, which aims at identifying heterogeneous groups that are nevertheless similar in their overall composition. This problem can be viewed as the task of fairly distributing diverse elements into balanced subsets. Typical applications include team formation, balanced delivery routing, and resource allocation scenarios.
Building on preliminary work in MCAC, we develop heuristic procedures based on PROMETHEE II net flow scores to construct balanced and diverse groups. In addition, we introduce new MCAC indicators designed to assess both intra-group diversity and inter-group similarity within a given partition. The proposed heuristics and indicators are evaluated on real-world datasets and compared with the limited number of existing approaches.
| Keywords | Multiple Criteria Decision Aid (MCDA), PROMETHEE, Multicriteria Anti-clustering, Heterogenous Grouping, Balanced Partitioning, Fairness. |
|---|---|
| Participation in the SNSF Early-Stage Researcher Support and Award Scheme | I am submitting a regular presentation and would like to participate in the SNSF Early-Stage Researcher Support and Award Scheme. |