This paper presents the findings from an analysis of several Bayesian updating scenarios in the context of data transferability. Bayesian updating has been recognized as having great potential for use in the transportation field, especially in the simulation of travel demand and other transportation-related data. For local areas where comprehensive data collection is too costly and infeasible, Bayesian updating can be used to synthesize travel demand data in a process generally referred to as data transferability. Bayesian updating has been occasionally employed for transferring travel data; however, various aspects and disadvantages of its use have been insufficiently studied. This work addresses issues about Bayesian updating techniques in data transferability, including a comparison of the use of conjugate and nonconjugate formulations in the updating models, their relative effectiveness, and impacts of the quality of the prior information on final results. The study shows that, in general, updating small local samples of travel attribute data with prior information from national data sources provides an improved estimate of local travel attributes compared with using the local sample only. However, this study found that the inclusion of all available historical data in the prior distributions does not necessarily improve the quality of the updating results. Therefore, careful analysis of the applicability of the prior information to the desired context is necessary when a Bayesian updating formulation is used. The 2001 National Household Travel Survey and the 1995 Nationwide Personal Transportation Survey were used for the demonstration exercises in this study.
Effectiveness of Bayesian Updating Attributes in Data Transferability Applications
Transportation Research Record
2013-01-01
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Effectiveness of Bayesian Updating Attributes in Data Transferability Applications
Online Contents | 2013
|Temporal transferability and updating of zonal level accident prediction models
Online Contents | 2006
|Criteria for selecting model updating methods for better temporal transferability
Taylor & Francis Verlag | 2020
|