Winter 2022 Series

Student Presenter: Dr. John Ash, Assistant Professor int he department of Civil Engineering and Construction Management at the University of Cincinnati.

Student Advisor: Professor Yinhai Wang Department of Civil Engineering at University of Washington.

Original Air Date: March 10, 2022

Description

Traditional efforts in traffic safety modeling have focused on prediction of crash frequency and severity. With advances in technology and data availability, real-time crash prediction modeling (RTCPM) has been an area gaining attention over recent years. RTCPM studies the relationship between crash risk and changes in traffic conditions (measured by different sensors) over short-duration time periods; it thus assumes the occurrence of a crash is related to the traffic conditions occurring in some period before the crash takes place. This study applies large-scale probe vehicle trajectory data (supplied by Inrix, Inc.) derived from millions of GPS trace points provided by mobile location services, consumer GPS devices, and commercial vehicle transponders collected over one week across all major freeways in Seattle, WA in 2017 to the RTCPM problem. Such data have not been used in this application before and provide finer spatial/temporal measurement resolution than obtainable through conventional traffic sensing infrastructure (e.g., loop detectors). Due to the size and nature of the data, an efficient data-processing pipeline to handle processes such as map-matching and conflation was developed. A case control study was established from which conditional logit models were developed to provide inference on traffic flow features impacting crash occurrence. Notably, it was observed that variables including, but not limited to, coefficient of variation in speed, average speed, and probe vehicle sample size had an impact on crash likelihood. Such findings can be used to justify traffic control measures to promote safety, such as variable speed limits.