Law enforcement resource optimization with response time guarantees

In a security-conscious world, and with the rapid increase in the global urbanized population, there is a growing challenge for law enforcement agencies to efficiently respond to emergency calls. We consider the problem of spatially and temporally optimizing the allocation of law enforcement resources such that the quality of service (QoS) in terms of emergency response time can be guaranteed. To solve this problem, we provide a spatio-temporal MILP optimization model, which we learn from a real-world dataset of incidents and dispatching records, and solve by existing solvers. One key feature of our proposed model is the introduction of risk values that allow a planner to flexibly make a tradeoff between their resource budget and the targeted service quality. Experimental results on real-world incident data, and simulations run on learned synthetic data, show a significant reduction in resource requirements over current practice, with violating QoS or abusing resource utilization.