Crimetracer: activity space based crime location prediction

摘要: Crime reduction and prevention strategies are vital for policymakers law enforcement to face inevitable increases in urban crime rates as a side effect of the projected growth population by year 2030. Studies conclude that does not occur uniformly across landscapes but concentrates certain areas. This phenomenon has drawn attention spatial analysis, primarily focusing on hotspots, areas with disproportionally higher density. In this paper we present CrimeTracer, personalized random walk based approach analysis location prediction outside hotspots. We propose probabilistic model behavior known offenders within their activity space. Pattern Theory concludes offenders, rather than venture into unknown territory, frequently commit opportunistic crimes serial violent taking advantage opportunities they encounter places most familiar part Our experiments large real-world dataset show CrimeTracer outperforms all other methods used recommendation evaluate here.