The Golf Tourist Problem

Tourism and travel with the purpose to do sports is gaining in popularity and the golf tourism market is considered to be one of the largest. Motivated by this phenomenon we model and solve the golf tourist problem which generalizes the orienteering problem with time windows. It aims at providing decision support for the traveling golfer by concurrently optimizing two objective functions: travel cost on the one hand and attractiveness of the generated travel plans on the other hand. Travel costs consist of flight cost, hotel cost, car rental cost, green fees as well as petrol cost for traveling between the selected golf courses. Attractiveness is measured by the total par scores of the visited golf courses. We assume that the traveling golfer provides a selection of regions in Europe that he or she is equally inclined to visit on his or her next trip. A feasible travel plan selects one region, contains only golf courses of this region and starts and ends at the respective airport. We solve the golf tourist problem to optimality by means of a recent bi-objective branch-and-bound algorithm and by means of the e-constraint method. Furthermore, we devise a decomposition approach that solves each regional problem separately and then combines the obtained Pareto sets. The proposed methods are applied to several real world instances with up to nine regions and between 57 and 227 golf courses per region. Our results show that the decomposition approach is significantly more efficient than the holistic approach. They also show that the bi-objective branch-and-bound algorithm performs better than the e-constraint scheme.