Casual Employee Scheduling with Constraint Programming and Metaheuristics

Abstract

We consider an employee scheduling problem where many casual employees have to be assigned to shifts defined by the requirement of different work locations. For a given planning horizon, locations specify these requirements by stating the number of employees needed at specific times. Employees place offers for shifts at locations they are willing to serve. The goal is to find an assignment of employees to the locations’ shifts that satisfies certain hard constraints and minimizes an objective function defined as weighted sum of soft constraint violations. The soft constraints consider ideal numbers of employees assigned to shifts, distribution fairness, and preferences of the employees. The specific problem originates in a real-world application at an Austrian association. In this paper, we propose a Constraint Programming (CP) model which we implemented using MiniZinc and tested with different backend solvers. As the application of this exact approach is feasible only for small to medium sized instances, we further consider a hybrid CP/metaheuristic approach where we create an initial feasible solution using a CP solver and then further optimize by means of an ant colony optimization and a variable neighborhood descent. This allows us to create high-quality solutions which are finally tuned by a manual planner.