Return to the VTRC Home Page
Click here to print the printer friendly version of this page.
Page Title: VTRC Report Detail

The contents of this report reflect the views of the author(s), who is responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of the Virginia Department of Transportation, the Commonwealth Transportation Board, or the Federal Highway Administration. This report does not constitute a standard, specification, or regulation. Any inclusion of manufacturer names, trade names, or trademarks is for identification purposes only and is not to be considered an endorsement.


A Retrospective Evaluation of Traffic Forecasting Techniques
Salwa Anam, Jasmine W. Amanin, Raleigh A. Matteo
John S. Miller
John S. Miller
Year: 2016
VTRC No.: 17-R1
Abstract: Traffic forecasting techniques—such as extrapolation of previous years’ traffic volumes, regional travel demand models, or local trip generation rates—help planners determine needed transportation improvements. Thus, knowing the accuracy of these techniques can help analysts better consider the range of transportation investments for a given location. To determine this accuracy, the forecasts from 39 Virginia studies (published from 1967-2010) were compared to observed volumes for the forecast year. Excluding statewide forecasts, the number of segments in each study ranged from 1 to 240. For each segment, the difference between the forecast volume and the observed volume divided by the observed volume gives a percent error such that a segment with a perfect forecast has an error of 0%. For the 39 studies, the median absolute percent error ranged from 1% to 134%, with an average value of 40%. Slightly more than one-fourth of the error was explained by three factors: the method used to develop the forecast, the length of the duration between the base year and forecast year, and the number of economic recessions between the base year and forecast year. In addition, although data are more limited, studies that forecast a 24-hour volume had a smaller percent error than studies that forecast a peak hour volume (p = 0.04); the reason is that the latter type of forecast requires an additional data element—the peak hour factor—that itself must be forecast. A limitation of this research is that although replication of observed volumes is sought when making a forecast, the observed volumes themselves are not without error; for example, an “observed” traffic count for a given year may in fact be based on a 48-hour count that has been expanded, based on seasonal adjustment factors, to estimate a yearly average traffic volume.

The primary recommendation of this study is that forecasts be presented as a range. For example, based on the 39 studies evaluated, for a study that provides forecasts for multiple links, one would expect the median percent error to be approximately 40%. To be clear, detailed analysis of one study suggests it is possible that even a forecast error will not necessarily alter the decision one would make based on the forecast. Accordingly, considering how a change in a traffic forecast volume (by the expected error) influences decisions can help one better understand the need for a given transportation improvement. A secondary recommendation is to clarify how some of these traffic forecasting techniques can be performed, and supporting details for this clarification are given in Appendix A of this report.