Ubiquitous Traffic Volume Estimation through Machine-Learning Procedure

Research output: NRELPresentation


Traffic volume data is one of the most important metrics for accurate assessment of the performance of a transportation system. Quality volume data is required to effectively assess extent of delay and congestion, detect real-time perturbations to the network, and understand traffic patterns during major weather events. Traffic volume on freeways are typically collected through continuous count stations installed by state DOTs, while there is lack of traffic volume observability on off-freeway roads. The National Renewable Energy Laboratory (NREL), in Collaboration with the I-95 Corridor Coalition and the University of Maryland, extended its research into estimating volumes anywhere anytime from industry probe based data for off-freeway roads. NREL combined vehicle probe count data with several other data sets (speed, whether, roadway geometry, time-of-day, day-of-week, etc.) to estimate hourly volumes as well as AADTs. The research validated and demonstrated the machine learning model, namely XGBoost, using data collected from Pennsylvania, North Carolina, and Tennessee.
Original languageAmerican English
Number of pages25
StatePublished - 2020

Publication series

NamePresented at the 2020 Vehicle Technologies Office Annual Merit Review and Peer Evaluation Meeting, 1-4 June 2020

NREL Publication Number

  • NREL/PR-5400-76617


  • machine learning
  • traffic volume estimation


Dive into the research topics of 'Ubiquitous Traffic Volume Estimation through Machine-Learning Procedure'. Together they form a unique fingerprint.

Cite this