Bayesian Estimation of Origin and Destination from Masked Trip Data

This article introduces a statistical method to estimate trips origin and destination locations from a masked trip data set. The estimation method uses trip features, the graph of the network, and publicly accessible external information on the realtime congestion status to find the most probable trips origin and destination based on a Bayesian approach, Markov Chain rule, and rank aggregation method. A case study of Porto, Portugal assesses the performance of the statistical estimation method by comparing the estimated location with the centroids of reported locations and with the actual trip origin and destination. Despite the limitation of the available data, the method provides better estimates of trips origin and destination compared to the centroids of reported locations.