Planificando Sistemas Territoriales Comerciales en Gran Escala mediante Modelos y Métodos de Programación Entera

Abstract: In this work, a solution method based on mixed-integer programming and cut generation for a very important commercial territory design problem is presented. Particularly, the districting problem under study consists of determining a partition of a set of geographci units into clusters or territories subject to planning requirements such as such as multiple territory balancing, compactness, connectivity, disjoint assignment, and similarity with existing plan. A mixed-integer linear programming model is introduced for this problem. The problem is NP-hard, that is, very hard to solve. Given its complexity, a novel technique based on branch-and-bound and cut generation is proposed for solving the problem. The method is enhanced by several algorithmic strategies. The empirical assessment of the proposed procedure shows its excellent performance by finding optimal and near-optimal solutions to very large-scale real-world instances in a few minutes of computational effort.