This study builds on previous efforts found in the field of rural traffic forecasting. The study combines careful statistical analysis with subjective judgement to develop models that are statistically reliable and easy to use. This study developed two different kinds of models - aggregate and disaggregate - to forecast traffic volumes at rural locations in Indiana's state highway network. These models are developed using traffic data from continuous count stations in rural locations as well as data for various county, state, and national level demographic and economic predictor variables. Aggregate models are based on the functional classification of a highway, whereas the disaggregate models are location-specific. These models forecast annual average daily traffic (AADT) for future years as a function of present year AADT, modified by the various predictor variables. The use of both aggregate and disaggregate models will provide more reliable traffic forecasts. The number of predictor variables employed in the models was kept to a minimum. The statistical analysis also found that the predictor variables are statistically significant; no other variables will provide significant predictive power to the models. The models developed in this study provide good r-squared values. More refined statistical techniques reinforce the choice of variables used in the models. This paper appears in Transportation Research Record No. 1203, Demand Forecasting and Trip Generation-Route Choice Dynamics.
Abstract