| Title |
Evaluating Street Networks For Predictive Policing |
| ID_Doc |
24622 |
| Authors |
Melo Borges J.C.D.; Cacho N.; Wang L.; Beigl M. |
| Year |
2020 |
| Published |
EAI/Springer Innovations in Communication and Computing |
| DOI |
http://dx.doi.org/10.1007/978-3-030-28925-6_3 |
| Abstract |
The idea that there is some sort of relationship between street networks and some sorts of crime has been researched around for a long time. However, besides from research in environmental criminology, few studies are known in urban computing evaluating the effect of street network features on crime occurrences for predictive and pattern recognition models. In this work, we analyze several street network features—from the perspectives of graph theory—and evaluate their impact on predictive policing models. Our results—based on years of historical crime incidents from San Francisco—suggest that although some street network features are spatially autocorrelated, others have a high mutual dependency to certain sort of crime occurrences. We then leverage selected street networks features in our exemplary use-case scenario for Crime Hotspot Detection, discussing the relationship between street network design and criminal activity. © 2020, Springer Nature Switzerland AG. |
| Author Keywords |
Machine learning; Smart cities; Urban computing |
