Daily Traffic Volume Modeling Based on Travel Behaviors

(M)odeling of traffic volume is very important for the estimation and prediction of traffic situations, which is one of the major aspects of time-based intelligent transportation systems. This paper proposes a model of daily traffic volumes based on mixture modeling, in which the components are carrying the information of travel behavior. We first give a general modeling method to show that daily traffic volumes can be approximated by a four-components Gaussian mixture model. Three highway traffic volume sets and six regular and irregular urban traffic volume sets from the Netherlands are verified. Furthermore, we analyze the travel behavior information in the mixture model. Examples of the area of Tanthof in Delft and Escamp in The Hague show the corresponding relation between this model and travel behaviors. All the results show that the mixture model not only fits the traffic volume curve, but also reflects the travel behavior information.