Clusterization of Clients for Location Making Decisions of Logistics Facilities

In this paper, we introduce to an application of different statistical techniques is developed, to propose a redesign of the distribution network for a 4PL logistics operator, which manages the supply chain of a telecommunications company. From a database of more than 2.2 million records, it is possible to establish a methodology for grouping clients into clusters, not only because of their geographical proximity, but also because of the similarities in terms of products demanded and service promise. The descriptive analysis, the exploratory factor analysis, the reduction of dimensions, the linear regression and the location of the distribution centers are combined using the centroid technique, to propose a decentralized network design that offers savings in the total annual cost of distribution.