Spatio-temporal modeling of destination choice behavior through the Bayesian hierarchical approach

Trip purpose inference is critical in transportation demand management (TDM) as well as traffic congestion alleviation. However, destination choice can be affected by a variety of factors, many of which are difficult to determine (e.g. socio-demographics). Besides, the spatio-temporal variation and correlation inherent in travel patterns further intensify the difficulty of understanding destination choice behavior. To this end, this research proposes a Bayesian hierarchical approach for modeling the destination choice behavior through time and space. The proposed method can take into account both the unavailable factors and spatio-temporal correlations by introducing random fields. Moreover, the implementation of the Integrated Nested Laplace Approximations (INLA) combined with the Stochastic Partial Differential Equation (SPDE) makes it computationally feasible to model large-scale spatio-temporal correlation structures. The model is further applied to two-week data from more than 8000 taxis in Harbin. The empirical results indicate that the proposed approach is capable to capture spatio-temporal variability in destination distribution, and the inclusion of spatial and temporal random effects is of great help to improve the model performance. The case study also examines how the land-use types influence the destination choice. It is believed that the modeling method and the exploratory spatial–temporal analysis of destination distribution in this study can enriches the methodologies for travel demand modeling as well as decision support for transport policy development.