Building a Crime Prediction Method Combining Short and Long-Term Risk of Crime —Extending the Concept of Long-Term Risk and Examining How to Introduce Short-Term Risk—

Abstract

Research on geographical crime prediction which combine short-term (explicit) and longterm (implicit) crime risks are found today. However, their combining methodologies do not match criminological theories. In the current study, we try to build a new prediction method which combines two kind of risks according with existing criminological research. Using Risk Terrain Modeling (RTM), which is the most promising crime prediction method in Japan, we consider the shortterm risk elevation (that is, near repeat victimization) associated with the latest crime occurrence. Moreover, we use the residual obtained from a regression analysis of crime occurrence by the RTM risk value. Then, we present a case study using data on thefts from vehicle in Japan to verify prediction accuracy and reliability of the new model. Finally, we discuss the possibility of new prediction methods that combine environments attracting crime and crime itself.