Journal of East China Normal University(Natural Sc ›› 2018, Vol. 2018 ›› Issue (3): 136-145.doi: 10.3969/j.issn.1000-5641.2018.03.015

Previous Articles     Next Articles

Detecting spatiotemporal hotspots for vehicle thefts by multi-scale analysis

REN Zhe-hao1,2, ZHANG Hao-tian1,2, LIU Wei-hang1,2, GUO Zhong-yang1,2   

  1. 1. Key Laboratory of Geographic Information Science(Ministry of Education), East China Normal University, Shanghai 200241, China;
    2. School of Geographic Sciences, East China Normal University, Shanghai 200241, China
  • Received:2017-05-17 Online:2018-05-25 Published:2018-05-29

Abstract: The detection of crime hot spots has become increasingly prominent in the conversion from reactive to active policing. There exist many crime analysis methods with good results. This paper focuses on scale effects in analysis. We proposed two multi-scale methods to detect temporal and spatial hotspots for vehicle thefts in a district, whose results were used for policing references. These two methods and their results are stated as follows:① a scaling method is proposed and combined with a rigid process to aggregate temporal data. Through this combination, temporal hotspots can be detected when data are not sufficient under mono-scale. Results showed that daily hot spots (30 days) and hourly hot spots (20 hours) of vehicle thefts are significant at the study site, on which the rearrangement of shift intervals can be based;② on the basis of daily hot spots, we set a median case density of a convex hull as the evaluation function when applying DBSCAN. The optimal scale, verified by the popular Prediction Accuracy Index, was adaptively chosen. We found that several metro line stations and residence zones need key protection.

Key words: multi-scale, crime analysis, wavelet analysis, DBSCAN, convex hull

CLC Number: