Hybrid Clustering Strategy for Micro-hubs Location in Newspaper Distribution


  • Karla Alvarez-Uribe1 (Institute for Intelligent Systems Research and Innovation (IISRI) Deakin University, Geelong, VIC 3216, Australia)
  • Eduard Ganan-Cardenas1 (Department of Production Engineering, Instituto Tecnológico Metropolitano, Medellín, Colombia)
  • Diego Perez-Montoya1 (Department of Production Engineering, Instituto Tecnológico Metropolitano, Medellín, Colombia)

Facility location is one of the most critical factors in urban logistics planning, and the newspaper industry is no exception. Given the time-sensitive nature of newspapers and their narrow delivery time windows, efficient distribution network planning becomes essential. This research addresses the micro-hub location problem within the context of newspaper distribution across the Área Metropolitana del Valle de Aburrá in Medellin, Colombia, by developing a novel hybrid clustering strategy. We compare five clustering techniques: K-means, K-medians, K-medoids, Agglomerative Nesting (AGNES), and Density-Based Spatial Clustering of Applications with Noise (DBSCAN). Our strategy first uses AGNES (with single-linkage) to identify high-density regions and subsequently applies K-medoids within these identified areas to form compact clusters. Results demonstrated the superiority of the hybrid clustering strategy over both K-means and the individual clustering techniques. The hybrid approach generates more cohesive clusters, as evidenced by superior silhouette coefficients and within-cluster variance. The clustering proposal allowed 90% of customers to be located within 1.6 kilometers of a micro-hub, improving the distribution of newspapers in the urban areas.

Download full PDF Get metrics Rate article