Cluster Identifications in Networks

Authored by: Becky P.Y. Loo , Tessa Kate Anderson

Spatial Analysis Methods of Road Traffic Collisions

Print publication date:  September  2015
Online publication date:  December  2015

Print ISBN: 9781439874127
eBook ISBN: 9781439874134
Adobe ISBN:

10.1201/b18937-10

 Download Chapter

 

Abstract

This chapter follows closely to the previous one with regard to outlining network analysis and network autocorrelation. This chapter is concerned with how, once one has established the optimal BSU (basic spatial unit) length or bandwidth, collisions can be measured together to identify hot zones. This issue has been tackled by some researchers in the past, and this chapter seeks to outline how to determine clusters on networks (or spatial contiguity). Many existing geostatistical methods for detecting clusters do not consider the specific nature of the network involved, often leading to biased conclusions. In traffic collision analysis, two techniques are commonly used for determining dangerous locations: the local spatial-autocorrelation method following the link-attribute approach and the kernel method following the event-based approach. Both methods are easily applicable and give comparable results for simplified road segments, such as exclusive highways or hypothetical linear roads, as independent spatial units with no intersections (1D). The analysis of road networks (2D), however, is not so straightforward because the special nature of road networks (notably connectivity) has to be considered. The local autocorrelation method requires the division of the road network into basic statistical units of standard length. There is no unique solution for this task, and it almost inevitably produces a number of statistical units that are often too short and excluded from further analyses. This results in a nonexhaustive coverage of the study area. The use of Euclidian distances in the planar 2D kernel method also disregards the network density: the statistical units are mostly created between, and not across, the intersections. This chapter will be supplemented with case studies from Hong Kong using different approaches to examine traffic collision patterns.

 Cite
Search for more...
Back to top

Use of cookies on this website

We are using cookies to provide statistics that help us give you the best experience of our site. You can find out more in our Privacy Policy. By continuing to use the site you are agreeing to our use of cookies.