Guidelines for Evaluating the Accuracy of Travel Time and Speed Data

Abstract

These Guidelines describe a process for evaluating the accuracy of traveler information in three forms: 1) congestion level (qualitative or color‐coded); 2) average link speed; and, 3) average route travel time. Public agencies, private companies, or third‐party evaluators can use these Guidelines to evaluate the accuracy of commercial traveler information services (TIS) prior to purchasing data or to ensure that contract requirements have been met. Also, public agencies and private companies can use these Guidelines internally to evaluate and improve the accuracy of their own TIS. These Guidelines define methods for collecting and processing benchmark data, but do not establish the level of accuracy that a traveler information service provider must achieve. These quality levels must be defined by the user or purchaser of the data based on their intended application of the data. These Guidelines outline and describe three main steps in the accuracy evaluation process: 1) develop an evaluation plan; 2) collect and reduce benchmark data; and 3) compute and report accuracy measures. An evaluation plan outlines basic parameters of an accuracy evaluation (e.g., accuracy measures, benchmark data collection method, the roads and time periods to evaluate, etc.). Once an evaluation plan has been developed, benchmark speed and travel time data are independently collected using reidentification (e.g., Bluetooth, license plate, etc.) or test vehicle (GPS, clipboard/stopwatch, etc.) methods. The benchmark is considered to be the best estimate of the true average travel time and speed for the defined roadway link/route and time period. Once benchmark data are summarized, accuracy measures are selected (based on the evaluation scenario), calculated, and reported. Finally, these Guidelines provide a list of resources and several examples to demonstrate the application of principles. A companion Technical Appendix documents the supporting analyses for these Guidelines.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 7 citations exist upstream.