Bayesian Updating of Transferred Household Travel Data

To address data limitation concerns in small and midsized urban areas, new research areas about the transferability of household travel survey data and synthetic travel data simulation are emerging. Research has suggested that local updating can significantly improve the quality of the transferred data. Among updating techniques, Bayesian updating is intuitively easy to understand and has the advantages of using prior information and successive updating. However, although parameters of interest can take various distributional forms, traditionally in practice only variables with normal distributions have been studied by using Bayesian updating. The application of Bayesian updating to two transferred travel statistics that are shown to be best fitted with gamma distribution was examined. The bootstrap method was used to justify the prior assumptions for the parameters of interest. Then, Markov chain Monte Carlo simulation with Gibbs sampler was used to update the priors of the parameters. The posteriors were then used to derive the updated means and standard deviations, which were later compared with the validation data set. The comparison results prove that significant improvement in quality of transferred travel data can be achieved and the level of improvement increases with the local sample size. A relatively cost-effective sample size was suggested for local updating of the transferred data.