Spatial adaptive crime event simulation with the RA/CA/ABM computational laboratory.

摘要

An agent-based crime event and crime pattern simulation model is developed in this research. The purpose of the simulation model is to provide a computational laboratory for environmental criminologists to study the interactions among offenders, targets, controllers, and crime places. The simulation model also aims to provide a useful tool for teaching crime event theories.; Routine activity theory and crime pattern theory are the theoretical foundations of the simulation model. Agent-based modeling coupled with cellular automata addresses the complex crime event process of street robbery. A type of spatial autonomous agent is developed with a wayfinding capability on the urban street network. The wayfinding algorithm is based on a reinforcement learning algorithm. Offender agents, target agents and police agents are developed based on the spatial autonomous agent, which can be released on a street network to execute their routine activity schedules. The interactions among offender agents, target agents, police agents, and crime places create crime events and crime patterns for analysis. Offender agents and target agents can learn from their past offending/victimization experience and change their spatial behaviors.; The crime event and crime pattern simulation model is tested to be able to generate credible spatial, temporal, victimization, and offending patterns. The simulation model is then applied to examine the effect of agent adaptations on spatial crime patterns, offending patterns and victimization patterns. The power-function distributions of crime events among crime places and offender population are examined as emphases.; Targeted for MS Windows desktop, the RA/CA/ABM computational laboratory is implemented using Visual C++. The computational laboratory has a graphic user interface that allows users to customize the simulation model, control the simulation process, visualize agent movement and crime patterns during the simulation, and query agent properties and crime patterns during the simulation. The computational laboratory is loosely coupled with Arcview GIS, so that it can take spatial data from Arcview as input, and make use of spatial analytical capabilities of Arcview to analyze the output crime patterns.