A Discrete-Event Simulation Model for Adaptive Allocation of Police Patrol

摘要

A challenge in law enforcement is the dynamic nature of crime, both temporally and spatially. While predictive policing models have attempted to handle these dynamics, the actual allocation of limited police resources still remains a challenge. For example, higher crimes showed up in the North district and at midnight that needed more officers compared to other areas and shifts, but this may change over time. To assist the Arlington Police Department (APD), we developed a discrete-event simulation to explore more adaptive allocations of patrol officers by responding to different situations. Crime calls have been grouped into priorities (E, 1, 2, & 3) per district (North, South, East, & West) and estimated probability distributions of call inter-arrival time and service time, shift (Day, Evening, Midnight) information, number of on-duty officers, service rules are used as the inputs to evolve the model. After reviewing the results of the simulation run, APD can get an idea of how many patrol officers should be deployed in each district and shift that enable dynamic staffing instead of the current annual staffing approach.