1. Home
  2. RIP
  3. View Record
https://trb-annual-meeting.nationalacademies.org/registration

Enhancing Traffic Delay Prediction Utilizing Data-Driven Techniques

A model that accurately predicts both traffic delays and the queues that result from work zones would be a valuable tool to Arizona Department of Transportation (ADOT), helping the agency to manage traffic, enhance work zone planning, reduce congestion, and improve road safety. Currently, ADOT lacks the ability to generate estimates of congestion and delays that result from lane closures and other forms of planned or unplanned roadway capacity reduction. Instead, the agency relies on rough rules of thumb to manage traffic and maintain safe operating conditions around work zones. Integrating a data-driven model—one that is based on roadway capacity and travel demand—into the work-zone management process would help the Traffic Operations Center (TOC) and other ADOT groups respond to both planned and unplanned traffic-delay events. Information that predicts potential problems before they occur could help the TOC prepare more efficiently for closures and other events by anticipating messaging and communication needs to the traveling public. For example, identifying responses to predetermined thresholds of congestion and delay related to work zones—e.g., adjusting signal timing, suggesting drivers use alternate routes, or other strategies—could help the TOC and other groups increase safety and reduce overall congestion. Data-driven models employing machine learning could also aid in the real-time adjustment of those thresholds based on sudden traffic interruptions as a result of incidents.