Adapting a local indicator of spatial association to identify hot spots in traffic safety

Abstract

In the last years, traffic safety has become a ‘hot’ topic in the media, for policy makers, for academics and for the broad audience. In general, traffic safety analysis can be split up in four phases: identification of hot spots, ordering, profiling of the hazardous locations and finally selecting the locations that need to be handled. This paper will focus on phase 1. Several spatial measures have been developed to identify hot spots, however, not all measures are equally well suited in the field of traffic safety. E.g. a high-accident location is always of interest for hot spot analysis, but a low-accident location may be of interest if its neighbouring locations show an opposite behaviour regarding the number of accidents. In this paper we will present an adaptation of a local measure of spatial association, more in specific of Moran’s I and apply it to identify hot spots on highways.