Predicting residential burglaries based on building elements and offender behavior: Study of a row house area in Seoul, Korea

Current methods for predicting residential burglaries mostly rely on analyses of crime patterns based on a real information. While this model is valid on an urban scale, it fails to consider street-scale environmental factors as well as offender behaviors in response to those factors. To improve the predictability of crime-simulation studies, this study investigated two influential factors in the occurrence of residential burglary: the physical properties of building elements and offender behaviors in response to those properties. First, a prediction algorithm was designed based on analyses of the factors. Next, a prediction method was established by modeling a virtual 3-D environment and a virtual offender using the algorithm. Lastly, the probability of residential burglary was analyzed via a simulation using the prediction method. A comparison of the simulation results with actual residential burglary data confirmed that the proposed method has statistically significant predictability.