Data-driven prediction of relevant scenarios for robust combinatorial optimization

Marc Goerigk, Jannis Kurtz

Abstract: Highlights•We consider robust one- and two-stage robust problems with discrete uncertainty.•The scenario recognition problem is to identify the most relevant scenarios.•We derive scenario features to train a machine learning model for this problem.•The model us used to speed up column-and-constraint generation method.•Experiments show improved performance and allow identification of important features.