I. Introduction
One of the applications of the intelligent transportation system is congestion prevention in urban areas which requires continuous prediction of traffic volumes. A prediction for a short-term horizon, around 15 minutes, is sufficient to provide information regarding future traffic conditions and achieve efficient traffic management strategies [1]. In this context, the short-term prediction is to predict traffic volumes on certain roads or road segments of a route at short and sufficient time ahead. Different short-term prediction models have been proposed but their accuracy depends on how these models address traffic volume characteristics, and how they deal with relationships among roads. In this work, we study the problem of how to improve the accuracy of the traffic volume prediction in urban areas.