Modelling traffic accident data by cluster analysis approach

Abstract

In recent years, traffic accidents have become an urgent problem due to increasing car ownership and traffic density. One of the most common methods for this problem is determination and analysis of "black spots". The conventional black spot identification method includes marking the location of each accident with a pin and investigation of black spots considering density of the pins on a map. In this study, the traffic accidents data of Denizli city for the years of 2004, 2005 and 2006 have been analyzed using the k-means and the fuzzy clustering methods. The spots that are densely located around the cluster centers are determined as "black spots" and are analyzed. The results of the analysis are evaluated regarding all features of the black spots and recommendations for improving traffic safety are reported.