Integrated Generation of Working Time Models and Staff Schedules in Workforce Management

Our project addresses the question how to automatically and simultaneously assign staff to workstations and generate optimised working time models on the basis of fluctuating personnel demand while taking into account practical constraints. Two fundamentally different solution approaches, a specialized constructive heuristic (commercial) and a hybrid metaheuristic (the evolution strategy) that integrates a repair heuristic to remove contraint violations are compared on a complex real-world problem from a retailer. The hybrid approach clearly outperforms the tailored constructive method. Taken together with our similar findings on a related staff scheduling problem from logistics this result suggests that the evolution strategy, despite its original focus on continuous parameter optimisation, is a powerful tool in combinatorial optimisation and deserves more attention. Moreover, hybridising a metaheuristic with a problem-specific repair heuristic seems a useful approach of resolving the conflict between domain-specific characteristics of a realworld problem and the desire to employ generic optimisation techniques, at least in the domain of workforce management.