Traffic Growth Projection for Traffic Impact Studies

Traffic projection is an important stage in transportation engineering including traffic impact studies and evaluation of traffic concurrency. Accurately forecasted traffic growth is required for transportation planning, highway safety evaluation, traffic operations analysis and design. In view of the importance of accuracy in traffic projection, this study introduces a regression-based traffic forecasting methodology with a tractable prediction function. Five different prediction functions were tested and the best was selected based on the accuracy of projections against the count data. The three-parameter logistic function produced more accurate projections compared to other functions. When validated, the coefficient of correlation was found to be satisfactorily above 80 percent at each of test locations. The t-statistic significance test was performed using the Fisher’s Information Matrix approach. The result suggests all parameters in the logistic function were highly significant. To evaluate reliability of projections, the predictive intervals were calculated at a 95 percent level of confidence using the Delta method. The predictive intervals give the statistical upper and lower bounds in which the observed traffic volumes, if fallen within the bandwidth, are considered reliable. The confidence intervals estimated by the Delta Method were found almost identical to those estimated by statistical software. Predictions by the logistic function were also compared to those by the linear regression suggested by the Florida Department of Transportation (FDOT) based on ten-year historical observations. Linear regression predictions tend to underestimate traffic volumes at the earlier stage of growth and overestimate the traffic volume when the growth is approaching the roadway capacity.