A genetic algorithm approach for regrouping service sites

Abstract

We address the problem of regrouping service sites into a smaller number of service centers, where each center serves a region. We propose a two-phase method, based on a weighted-graph problem formulation, for providing good suboptimal solutions. In the first phase, the graph is decomposed into the required number of subgraphs (regions) using a tuned hybrid genetic algorithm. The second phase finds a suitable center within each region by using a heuristic algorithm.

Scope and purpose

Public and private organizations are sometimes interested in consolidating their resources by regrouping their service sites into a smaller number of existing sites, henceforth referred to as service centers. These centers are intended to provide more economic and, perhaps, higher quality services to the clients. Also, the objective is to locate centers with balanced service loads such that the distances that the clients have to travel are kept low. A good example is where educational authorities regroup a number of geographically spread and ill-equipped public schools into a smaller number of schools that are better-equipped in terms of human and physical resources and that also reduces the overall cost.