Applications such as route guidance and proactive network management frameworks require not only the currently measured or estimated traffic parameters, but also the forecasts of such parameters in the short-term future. The term short-term is used in this context to refer to time windows of up to one hour into the future. A large number of academic papers are available on the subject of short-term forecasting or prediction of traffic variables. Such papers differ widely in the type of statistical or machine learning tools used for prediction and the type of explanatory variables used to predict the future values of a given traffic variable. This paper examines the