A Dynamic Policing Simulation Framework

Abstract

Crime is a serious problem to a society, and its costs are an economic burden. With the help of technology and developed tools, law enforcement agencies are making significant efforts to combat crime, so as to create a safer environment for society, both mentally and physically. The dynamic nature of crime and limited police resources often make their efforts challenging. Although there are numerous crime prediction models found in the policing literature, guidelines for policing strategies based on those models are still lacking. Towards addressing this gap, this dissertation constructs a dynamic policing simulation framework based on the concept of prediction-led policing that combines decision strategy, predictive policing, and simulation modules to enable the study of strategies for dynamic deployment of police resources to reduce crime. The decision strategy dynamically adjusts a policing strategy to try to minimize crime, the predictive policing model is used as a state transition function to predict future crime and the simulation evaluates the policing strategy and produces performance metrics. The main focus of this research is developing the simulation module and integrating it within a framework with the dynamic decision strategy and predictive policing modules. Data provided by Arlington, Texas Police Department (APD) are used to estimate probability distributions for the simulation module and to build a preliminary predictive policing model appropriate for a dynamic policing framework. The developed framework has the flexibility to apply for any city over any time scale, provided an appropriate predictive policing model can be estimated.