A pessimist's approach to one-sided matching

Tom Demeulemeester, Dries R. Goossens, Ben Hermans, Roel Leus

Abstract: Highlights•We study how to implement a probabilistic assignment as a weighted sum of matchings.•We focus on worst-case measures of the matchings in this decomposition.•We find theoretical results when maximizing the worst-case number of assigned agents.•We propose a flexible column generation framework.•We show on real-world data that our framework performs well in practice.