Modeling Crime Density with Population Dynamics in Space and Time: An Application of Assault in Gangnam, South Korea

Abstract

The current study explores populational and environmental factors associated with violent crime. Specifically, it compares ambient and residential populations with regard to their association with assault density at a fine spatial and temporal unit in a city with socio-economic control variables. The results show that the ambient population are consistently associated with the level of assaults throughout the four time periods in a day, while residential population does not contribute much to explaining its variation. In addition, we also find that the percentage of single-member households and the distance to the nearest subway station are constantly associated with assault density, while the proportion of non-residential use and the land price are partially associated.