Identification of Influential Factors and Potentially Hazardous Location of Rear-End Collisions with Hard-Braking Events Data

Rear-end collisions on expressways account for approximately 43% of the total collision cases, which is a high proportion. Extensive research has been conducted on rear-end collisions on expressways using traffic collision data. However, because traffic collisions occur rarely and randomly, it is challenging to conduct comprehensive traffic safety-related analyses using collision data only. Surrogate safety measures (SSM) can be used to replace collision data, but they also face challenges in constructing extensive data beyond spatial and temporal constraints. Hard-braking event data used in this study were collected primarily from smartphones or navigation applications installed in vehicles, which allowed us to obtain data from all timeframes on nationwide expressways. Hard-braking events exhibit a strong association with rear-end collisions, and are considered precollision indicators. This research developed negative binomial models for hard-braking events and rear-end collisions, and computed segment-level potential for safety improvement (PSI). High-risk segments are identified by clustering elevated PSI for hard-braking events and, separately, for rear-end collisions, and contributing factors were examined. The findings demonstrate the value of hard-braking events for screening hazardous locations and informing preventive safety strategies on expressways.