Abstract
The Network Kernel Density Estimation (NetKDE) is a useful tool for visualization of point events over a network space, but it lacks in expressing the statistical significance of the mapped phenomenon. In this paper, we discuss the network hotspot detection of street crimes by integrating the NetKDE and the Getis-Ord GI* statistics. We selected four types of network-constrained crimes, i.e., bike theft, car theft, robbery, and snatching. The NetKDE is a useful technique to study the patterns of crimes bounded by the road networks. We used the Spatial Analysis along Networks (SANET) tools for computing the Network Kernel Density Estimation (NetKDE) and utilized the results of the NetKDE as input values for computing the Getis-Ord GI* statistics. The combination of these two methods can detect the network-constrained hotspots that are statistically significant. We also performed the network K-function, the extension of the Ripley’s K-function on networks. The network K-function analysis displays the significant clustering of crime events at different scales. Results demonstrated that the intensity of street crimes are strongly concentrated in the central part of the city. Moreover, the results reflected that the functional nature of different urban land use affects the frequency of crime events. Various urban land uses such as commercial, residential and industrial area seemed to influence the distribution of different types of crimes. The hotspot analysis has real potential, impacting the police patrolling protocols. The methods presented in this study suggest that there is a need to distinguish the planar and network hotspots and crime prevention policies could be enacted according to the type of hotspots.
Original language | English |
---|---|
Pages (from-to) | 599–622 |
Number of pages | 24 |
Journal | Applied Spatial Analysis and Policy |
Volume | 11 |
Issue number | 3 |
Early online date | 18 Sept 2017 |
DOIs | |
Publication status | Published - Sept 2018 |
User-Defined Keywords
- Network-constrained clusters
- network hotspots
- Getis-Ord GI
- NetKDE
- K-function