Urban junctions usually present significant safety concerns, and the majority of all crashes in urban areas occur in or near junctions. Factors contributing to crash severity at junctions have been explored, but crash risk levels and crash severity patterns of different junction types have hardly been investigated. In order to fill this gap, this study analyzed the safety performance of six junction types and the factors contributing to crash severity, in order to assist city transportation authorities to implement effective countermeasures. Fault tree analysis (FTA) was applied for the risk evaluation of urban junctions and association rules (AR) algorithm was employed for the crash severity pattern analysis based on data from the U.K. STATS19 database from 2012 to 2016. Overall, four types of urban junctions with high crash risk level and over 4,000 AR contributing to crash severity are identified in the present paper. The results show that: (a) roundabouts and mini-roundabouts have the lowest fatality and casualty rates while T-junctions or staggered junctions and crossroads have the highest crash risk levels; (b) FTA may produce inaccurate outcomes because of incorrect logic gates, but AR can generate real potential relationships between crash severity and risk factors; (c) crash severity patterns are quite complex and the interdependence between risk factors is different for each junction type; (d) risk factors such as male driver, no physical crossing facilities within 50 meters, and give way or uncontrolled junction are common in high-risk junctions at night.
Traffic Safety Region Estimation Based on SFS-PCA-LSSVM: An Application to Highway Crash Risk Evaluation
