Prescriptive Analytics in Urban Policing Operations

Problem definition: We consider the case of prescriptive policing, that is, data-driven assignment police cars to different areas a city. analyze key problems with respect prediction, optimization, and evaluation as well trade-offs between quality measures crime types. Academic/practical relevance: Data-driven analytics is gaining substantial attention in operations management research, effective policing at core almost every city world. Given vast amounts data increasingly collected within smart initiatives growing safety challenges faced by urban centers worldwide, our work provides novel insights on development applications an context. Methodology: conduct computational study using auxiliary San Francisco. both strong weak prediction methods along two optimization formulations representing deterrence response time impact vehicle allocations. these effects Results: find even weaker can produce Pareto-efficient outcomes time. identify three archetypes combinations objectives constitute Pareto frontier among configurations we analyze. Furthermore, optimizing for multiple types biases allocations way generally decreases single-type performance one outcome metric but improve it other. Managerial implications: Although integrating all relevant theoretically possible, practically challenging because each type requires collectively consistent weight. present framework combining subset exploring remaining support implementation operations-focused solutions practice.