Robust personnel rostering: How accurate should absenteeism predictions be?

Disruptions to personnel rosters caused by absenteeism often necessitate last-minute adjustments to the employees’ working hours. Acommonstrategytomitigatetheimpactofsuchchangesistoassignemployeestoreserveshifts:specialon-callduties during which an employee can be called in to cover for an absent co-worker. To maximize roster robustness, we assume a predict-then-optimize approach that uses absence predictions from a machine learning model to schedule an adequate number of reserve shifts. In this paper, we propose a methodology to evaluate the robustness of rosters generated by the predictthen-optimize approach, assuming that the machine learning model will make predictions at a predetermined prediction performance level. Instead of training and testing machine learning models, our methodology simulates the predictions based on a characterization of model performance. We show how this methodology can be applied to identify the minimum performancelevel needed forthe modeltooutperformsimplenon-data-driven robust rostering policies. In a computational study for anurserostering problem,wedemonstratehowthepredict-then-optimizeapproachoutperformsnon-data-drivenpolicieseven under not particularly demanding performance requirements, particularly when employees possess interchangeable skills.