Traffic Volume Forecasting for Rural Roads in Jordan

Abstract

The accuracy of traffic forecasting is a point of considerable importance to the effective allocation of limited resources. Thus, reasonable and accurate forecasting methods should be developed to help engineers and planners make rational decisions and reduce probable associated risks. This study investigated the performance of three forecasting methods: aggregate regression, disaggregate trend and empirical Bayesian analysis. To accomplish this objective, traffic volumes for major rural roads in 1996 through 2004 were obtained from the Ministry of Public Works and Housing of Jordan. For each city or zone, cross-sectional data on socio-economic and demographic variables were collected. Multivariate regression analysis was carried out to develop mathematical relationships that could have practical applications. The results indicated that the products of populations-to-roadway length ratio, number of employees, fuel consumption, number of buildings and road type significantly influenced traffic interchange between cities or zones. For Jordan conditions, the linear model was recommended. Trend models, having exponential form, were also developed. Performance analysis indicated that aggregate regression and empirical Bayesian analysis provided comparable results. In contrast, the performance of trend method was considered to be poor. Finally, while these results are related to Jordan, they possibly apply elsewhere as well.