Stochastic Variable Neighborhood Search for a Rotation Assignment Problem

In this paper, we deal with a type of rotation assignment problem motivated by a real-life training program, where a number of trainees are to be assigned to rotations within a planning horizon under a set of constraints. A major type of constraint stipulates that each trainee should perform a rotation in a range of consecutive time periods. Fairness of trainees being assigned to rotations and rest requests from trainees are also considered. We develop a stochastic neighborhood search algorithm for solving this problem. The algorithm uses an iterated stochastic neighborhood search framework with a simple acceptance criterion. The algorithm has no parameters and is very easy to implement. The approach is tested on a set of instances from real-life settings. We compare our approach with another neighborhood search procedure. The results show the effectiveness of our algorithm.