Dynamic Network Analysis of Spatio- Temporal Crime Incidents in Bogotá - Colombia

Abstract

We use concepts from network theory to describe and understand the topology of a spatio-temporal graph of 4 categories of crime incidents in Bogotá, Colombia for the period 2018 to 2021. Using Page Rank, Betweenness, Closeness, and Eigenvector centrality measures we identify criminal hotspots and describe their role in Bogota's criminal spatio-temporal network. We also evaluate the time persistence of criminal hot spots. For each centrality measure, we identify the central nodes across time by estimating the proportion of times that they remain as hotspots. The result of this exercise suggests us a non-random behavior of the evolution of the network along time. We test this non-random behavior statistically using a “null model”. Finally, we estimate Louvain and Girman Newman communities over a “mean graph” constructed from all the graph timeline, finding well-defined persistent spatio-temporal communities of crime incidents. To the extent of our knowledge, our spatiotemporal crime graphs are a novel way of studying these types of events. <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> Results of the project “Diseño y validacion de modelos de analítica predictiva de fenomenos de seguridad y convivencia para la toma de decisiones en Bogotá” funded by Colciencias with resources from the Sistema General de Regalias, BPIN 2016000100036. The opinions expressed are solely those of the authors.