Developing a Cluster-Based Algorithm for Collision Hotspot Identification

Traffic safety is one of the most important goals for roadway design and traffic system operations. Collision hotspot identification serves as a major fundamental component for traffic safety performance measurements. With identified collision hotpots, limited resources can be better allocated to improve roadway safety. There have been a significant amount of studies on collision hotspots identification over the past decades. However, most studies only considered crash counts as the sole roadway safety performance indicator. In this study, a new cluster-based method is proposed to quantify roadway safety conditions. This method is able to incorporate more heterogeneous safety-related factors for clustering, such as crash fatality, injuries, and average collision duration. Compared with the prevailing Empirical Bayes methods, our cluster-based method demonstrates its improved accuracy and efficiency, and can be easily implemented in practice.