Clustering and profiling traffic roads by means of accident data

Abstract

In the first part of this research, model-based clustering is used to cluster 19 central roads of Hasselt into distinct groups based on their similar accident frequencies for 3 consecutive time periods of each 3 years: 1992-1994, 1995-1997, 1998-2000. The observed accident frequencies are assumed to originate from a mixture of density distributions for which the parameters of the distribution, the size and the number of clusters are unknown. It is the objective of latent class clustering to ‘unmix’ the distributions and to find the optimal parameters of the distributions and the number and size of the clusters, given the underlying data. More specifically, we use a multivariate Poisson mixture model with one common covariance term to model the data. A general algebraic modelling system is used to maximise the loglikelihood function. The accident data are obtained from the Belgian “Analysis Form for Traffic Accidents” and contain a rich source of information on the different circumstances in which the accidents have occurred: course of the accident, traffic conditions, environmental conditions, road conditions, human conditions and geographical conditions. In the second part of this paper, the data mining technique of association rules is used to profile each cluster of traffic roads in terms of the available traffic accident data. The strength of this approach lies within the identification of accident circumstances that frequently occur together for each group of traffic roads. This can, in turn, make a strong contribution towards a better understanding of the accident circumstances in these clusters.