Classifying vehicles' behaviors using global positioning systems information

This study presents a solution to enhance the cities’ traffic control by classifying particular vehicles’ behaviors. A Support Vector Machine (SVM) approach is presented, enabling the system to classify cars that are looking to park and those that are simply transiting through a city. Through this paper, we also propose a new way of managing the high density of traffic data using a grid. The results show that the system is able to distinguish the two different behaviors with an accuracy averaging 80%