A Tabu Search Approach for a Territory Design Problem with Stochastic Demands

Abstract: Territory design can be defined as the problem of grouping small geographic units into larger geographic clusters called territories in such a way that the latter satisfy certain planning criteria. A stochastic version of a commercial territory design problem in presence of uncertain demands is addressed in this work. The objective is to minimize the territory imbalances with respect to the product demand subject to planning criteria such as territory connectivity, compactness, and balance with respect to the number of customers. A tabu search metaheuristic for large instances is developed for this problem. The proposed method incorporates advanced techniques such as strategic oscillation and specific neighborhood exploration strategies. Empirical evidence shows the value of using such advanced techniques. To the best of our knowledge, this is the first study addressing the stochastic nature of the demand in a commercial territory design problem.