Application of Finite Mixture of Regression Model with Varying Mixing Probabilities to Urban Arterial Travel Time Estimation

摘要

Travel time along an urban arterial is greatly affected by traffic signals. Most studies on urbantravel time employ statistical models to directly obtain the distribution without incorporating theeffects of traffic signal timing (1-6). In this study, a finite mixture of regression model withvarying mixing probabilities (weights) is proposed to gain a better understanding of urban traveltime distribution by considering the signal timing. While the standard finite mixture models withconstant mixing probabilities have limited ability to adapt to the underlying random structuralchanges for the observed travel times, the model developed in this study can capture suchdynamics by 1) modeling the mixing probabilities as a function of the explanatory variablesassociated with signal timing and 2) establishing a linear regression between the mean of eachcomponent and signal timing. The finite mixture of regression model is applied to the travel timedata collected by the Automatic Vehicle Identification (AVI) system on one urban arterial withSydney Coordinated Adaptive Traffic System (SCATS). The results demonstrate that the varyingmixing probabilities can be used to classify the samples of travel time and the mean values ofcomponents can capture the effects of signal timing. By comparing various types of mixturemodels, the proposed approach not only has a better statistical fitting performance but alsoprovides useful information about travel time features.