Need to go further: using INLA to discover limits and chances of burglaries’ spatiotemporal prediction in heterogeneous environments

Abstract Near-repeat victimization patterns have made predictive models for burglaries possible. While the been implemented in different countries, results obtained not always line with initial expectations; to point where their real effectiveness has called into question. The ability predict crime improve preventive policing strategies is still under study. This study aims discover limitations and success of that attempt based on spatiotemporal risk break-ins spreading geographic proximity break-ins. A log-Gaussian Cox process contemplated model generic near-repeat scenario adjusted using Integrated Nested Laplace Approximation (INLA) methodology. approach highly suitable studying describing phenomenon. However, predictions INLA are quite monotonous, low variability do reproduce well local short-term dynamics purposes. conclusion cannot be restricted exclusively distance decay risk, but they must designed detect other types which, among possibilities, open up possibility correlating distant events clusters. Although studies already highlighted this problem, proposal here go one step further clearly extend spatial achieve better prediction results.