Abstract
The U.S. Congress has mandated that the National Performance Management Research Data Set (NPMRDS) be used by federal and state transport agencies and metropolitan planning organizations for performance analyses. The NPMRDS contains archived space mean speed (SMS) from probe vehicles representing Trucks, Passenger Vehicles, and All-Traffic for every 5-min time period of each day across predefined roadway links in the United States. This paper analyzes approximately 41 million SMS estimates from 4,385 spatial segments disaggregated at 5-min intervals on Alabama highways for all days in October 2023. There was strong evidence found that the NPMRDS SMS estimates for All-Traffic were biased toward Truck SMS estimates. The authors subsequently developed a statistically valid approach for removing the oversampling bias. It was found that the maximum mean average percentage difference between the proposed approach and the uncorrected All-Traffic SMS was approximately 25%. To improve the accuracy of traffic metrics that depend on the NPMRDS, it is recommended that users should consider the proposed correction method if their NPMRDS data show similar oversampling bias. It is also recommended that the NPMRDS should explicitly define how the All-Traffic SMS data were estimated and provide the actual number of probes or sample size instead of a quasi-representation of actual probes.
Get full access to this article
View all access options for this article.