Estimation and Prediction of Time-Dependent Origin-Destination Flows with a Stochastic Mapping to Path Flows and Link Flows

This paper presents a new suite of models for the estimation and prediction of time-dependent Origin-Destination (O-D) matrices. The key contribution of the proposed approach is the explicit modeling and estimation of the dynamic mapping (the assignment matrix) between time-dependent O-D flows and link volumes. The assignment matrix depends upon underlying travel times and route choice fractions in the network. Since the travel times and route choice fractions are not known with certainty, the assignment matrix is prone to error. The proposed approach provides a systematic way of modeling this uncertainty to address both the offline and real-time versions of the O-D estimation/prediction problem. Preliminary empirical results indicate that generalized models with a stochastic assignment matrix could provide better results compared to conventional models with a fixed matrix.