Data Mining and Complex Network Algorithms for Traffic Accident Analysis

The field of traffic accident analysis has long been dominated by traditional statistical analysis. With the recent advances in data collection, storage, and archival methods, the size of accident data sets has grown significantly. This result in turn has motivated research on applying data mining and complex network analysis algorithms to traffic accident analysis; the data mining and complex network analysis algorithms are designed specifically to handle data sets with large dimensions. This paper explores the potential for using two such methods namely, a modularity-optimizing community detection algorithm and the association rule learning algorithm to identify important accident characteristics. As a case study, the algorithms were applied to an accident data set compiled for Interstate 190 in the Buffalo Niagara, New York, metropolitan area. Specifically, the community detection algorithm was used to cluster the data to reduce the inherent heterogeneity, and then the association rule learning algorithm was applied to each cluster to discern meaningful patterns within each, related particularly to high accident frequency locations (hot spots) and incident clearance time. To demonstrate the benefits of clustering, the association rule algorithm was also applied to the whole data set (before clustering) and the results were compared with those discovered from the clusters. The study results indicated that (a) the community detection algorithm was quite effective in identifying clusters with discernible characteristics, (b) clustering helped unveil relationships and accident causative factors that remained hidden when the analysis was performed on the whole data set, and (c) the association rule learning algorithm yielded useful insight into accident hot spots and incident clearance time along I-190.