Comprehensive Assessment of Third-Party Truck Traffic Volume Data: A Kentucky Case Study

Abstract

Third-party traffic volume data offer a time- and cost-effective alternative to traditional data collection methods using permanent and short-term counters. While many studies have evaluated the accuracy of all vehicle volume data, little attention has been devoted to assessing truck volume data, which is essential for freight planning and ensuring the efficient movement of goods on highway networks. This study addresses this gap by conducting a comprehensive assessment of Streetlight’s truck volume data across various temporal resolutions and roadway groups. The results indicate that Streetlight’s Annual Average Daily Truck Traffic (AADTT) data demonstrated reasonable accuracy at permanent counter locations, with a mean absolute percent error (MAPE) of 16.1%. However, it is important to note that the validation process was not entirely “blind,” as Kentucky’s permanent count data were included in Streetlight’s model training dataset. When compared to short-term counts, the MAPE of Streetlight’s AADTT estimates increased to 41.2%. In addition, when aggregated by day of week, MAPE values for average daily truck traffic estimates ranged from 15.5% to 18% on weekdays but spiked to 95% on weekends. Estimation accuracy further declined when analyzed at daily and hourly temporal resolutions. Significant discrepancies were observed in daily volumes on weekends and major holidays, as well as hourly volumes during nighttime periods on high functional class roads and throughout the day on lower functional class roads. These findings provide valuable insights for transportation agencies considering the use of third-party volume data in their applications.