Computing Performance Measures with National Performance Management Research Data Set

Abstract

The National Performance Management Research Data Set (NPMRDS) was procured by the FHWA Office of Operations in 2013. NPMRDS initially served as a research data set for sponsored programs, but rights to use the data set were secured for state departments of transportation and metropolitan planning organizations in anticipation of the performance measure requirements of the Moving Ahead for Progress in the 21st Century Act. NPMRDS is a form of commercial GPS probe data; that is, the traffic conditions are derived from vehicles that periodically self-report speed, position, and heading with GPS electronics. NPMRDS differs from commercially available data feeds in that FHWA specified that no smoothing, outlier detection, or imputation of traffic data be performed. As a result, NPMRDS contains unique characteristics for statistical distribution of reported travel times. These are characteristics such that traditional processing techniques are ineffective in obtaining accurate performance measures. This paper proposes a method for handling the challenges posed by NPMRDS and computing meaningful performance measures from it. The paper first exposes the challenges in processing NPMRDS data and defines a method for overcoming the challenges. The paper compares the results from the proposed method with traffic data from commercial probe data sources and a reference reidentification data source at two case study locations. The case studies indicate that this paper successfully shows the ability to capture performance measures from NPMRDS more accurately with techniques originally developed to accurately reflect travel time and travel time reliability on interrupted-flow facilities.