Using Log Transformations to Improve AADT Forecasting Models in Small and Medium Sized Communities

摘要

Due to cost limitations, Annual Average Daily Traffic (AADT) data is not typically collected for every roadway segment, therefore, it is necessary to have a means to estimate the AADT value when the need arises for an uncounted roadway or roadway segment. Often, the methodology used to develop an estimate of the AADT is through a regression based, linear model. This research examines the use logarithmic transformations to improve the relationship between key socio-economic, roadway variables and the estimated AADT. In the study, traffic count, socio-economic, and roadway data were collected and different regression based models were developed and tested to determine if the use of logarithmic transformations improves the model accuracy and the model transferability to other communities of similar size. The results of the paper indicate that a linear-log model produced the best results of the logarithmic transformations, and was an improvement over a traditional linear regression model.