An Iterated Local Search Heuristic for the Staff Scheduling Problem for Part-Time Employees in Japan

Abstract

In this paper, we introduce a mathematical programming model for solving a staff scheduling problem based on one-day duties (task patterns) of individual staff members. The model can accommodate various service types, management policies, and staff preferences. We first enumerate all feasible one-day duties and propose an iterated local search approach that incorporates various methodologies, including a size-reduction method and a very large-scale neighborhood search. For the very large-scale neighborhood search, we design a dynamic programming method that aims to find the most improved schedule and can be used in the rescheduling stage. Computational results show that the model and the proposed algorithm perform well for real-world instances in Japan.