Burglaries Constellations: A New Approach to Deal with Burglaries’ Repeat Patterns in Heterogeneous Environments

Abstract Spatiotemporal patterns of repeat victimization in residential burglaries provide valuable insights for predictive applications, though they display increased complexity in large, heterogeneous environments. When the assumed homogeneous areas near prior burglary sites are diminished, the near-repeat pattern becomes restricted. This suggests that burglars may shift to other similar regions, driven by opportunity criteria aligned with criminal behaviour theories. Under this proposed strategy, certain target areas would remain consistently active throughout a burglary wave, leaving behind a detectable "hot trail" on heat maps. By applying Principal Component Analysis (PCA) with varimax rotation for dimensional reduction, we identified these "hot trails" as weighted groupings of cells, termed burglary constellations. Weekly series within these constellations exhibit non-random patterns, facilitating predictive modelling. The resemblance between cells and burglary profiles in each constellation, along with their spatial representation, offers significant information to improve risk prediction and guide preventive policing strategies in large, diverse areas. These findings are consistent with recent criminological studies, suggesting a novel approach in predictive policing frameworks.