Day-to-Day Origin-Destination Tuple Estimation and Prediction with Hierarchical Bayesian Networks Using Multiple Data Sources

Prediction of traffic demand is essential, either for an understanding of the future traffic state or so necessary measures can be taken to alleviate congestion. Usually, an origin-destination (O-D) matrix is used to represent traffic demand between two zones in transportation planning. Vehicles are assumed to be homogeneous; the trips of each vehicle are examined separately. This traditional O-D matrix lacks a behavioral basis and trip-based model structure. Another research stream of travel activity-based research addresses individual travel behaviors. This stream addresses the trip chain for travelers, but the research scope is attributes of trips, which ignores the road network. The concept of the O-D tuple, a sequence of dependent O-D pairs, is proposed for linking these two fields and for predicting traffic demand better. Through advanced monitoring systems that identify and track vehicles in the road network, the additional uncertainties of O-D tuples can be mitigated and thus reduce the underspecification more specifically. The hierarchical Bayesian networks mechanism in Gaussian space with multiprocesses is proposed for gaining the posterior of uncertain parameters. The model includes level and trend components for predicting future traffic volumes. A case study demonstrates that the proposed method can predict demand, and the path flow from cameras can reduce uncertainty in the estimation and prediction process, especially for O-D tuples.