Abstract
As a major social issue during urban development, crime is closely related to socioeconomic, geographic, and environmental factors. Traditional crime prediction models reveal the spatiotemporal dynamics of crime risks, but usually ignore the environmental context of the geographic areas where crimes occur. Therefore, it is difficult to enhance the spatial accuracy of crime prediction. We propose the use of anisotropic diffusion to include environmental factors of the evaluated geographic area in the traditional crime prediction model, thereby aiming to predict crime occurrence at a finer scale regarding spatiotemporal aspects and environmental similarity. Under different evaluation criteria, the average prediction accuracy of the proposed method is 28.8%, improving prediction accuracy by 77.5%, as compared to the traditional methods. The proposed method can provide strong policing support in terms of conducting targeted hotspot policing and fostering sustainable community development.
Highlights
Criminal analytics involves estimating the spatial distribution of crimes, searching the areas inclined to crimes, and predicting crimes
We focused on the crime risk of burglary at the community level and propose crime prediction based on anisotropic diffusion
By analyzing the spatial distribution of burglary risk at the m-th month and comparing the real distribution of burglary cases at the (m + 1)-th month, we evaluated the effectiveness of the anisotropic diffusion model (AnisDM) on crime prediction
Summary
Criminal analytics involves estimating the spatial distribution of crimes, searching the areas inclined to crimes, and predicting crimes. The traditional kernel density estimation algorithm is aimed at determining hotspots in the spatial distribution of crimes and assumes that hotspots might persist until the timestep [7,8]; it is more effective for predicting crimes in areas with stable crime hotspots [2,7,9]. Unlike kernel density estimation, which imposes strict requisites and assumptions, crime prediction based on near-repeat theory does not assume stable crime hotspots. Diffusion models are used for crime prediction that simulate the spatiotemporal transmission of crime risk described by the near-repeat theory and retrieve higher prediction accuracy than kernel density estimation [11,12,13,14,15]. The proposed method introduces environmental factors as background spatial information according to the geographic unit into the diffusion model for improved accuracy of (burglary) crime prediction.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
