A complex networks approach to demographic zonification

Abstract

This paper presents a novel approach for the zone design problem that is based on techniques from the field of complex networks research: community detection by betweenness centrality and label propagation. A new algorithm called Spatial Graph based Clustering by Label Propagation (SGCLAP) is introduced. It can deal with very large spatial clustering problems with time complexity O(nlogn). Besides, we use a parallel version of a betweenness-based community detection algorithm that outputs the graph partitioning that maximizes the so-called modularity metric. Both these methods are put at the centre of an effort to build an open source interactive high performance computing platform to assist researchers working with population data. © 2009 Springer-Verlag Berlin Heidelberg.