Bayesian updating of simulated household travel survey data for small/medium metropolitan areas

Abstract

This thesis tests an approach for generating simulated travel survey data that has local characteristics incorporated in it. Travel survey data are generally required for estimating and calibrating travel demand models for a region. The high cost associated with travel surveys puts them beyond the budget of most small/medium MPOs. Therefore simulation of travel survey data provides a viable alternative for these data starved regions to generate data. The simulated data is produced by combining socio-demographic data along with a national survey data set. Updating the simulated data distributions with the distributions obtained by surveying a small sample of local households, adds a local element to the simulated data set. The updating procedure using a small local sample of households is tested for two regions, which had previously conducted household travel surveys. The local sample was drawn from the travel survey and results obtained after updating were compared to those from the travel surveys in order to assess the performance of updating. Comparisons of trip attributes (trip rates, mode shares, departure times and trip lengths) in the two study areas show the updating has succeeded in bringing the updated values closer to the survey values in the majority of cases. The anomalies, which were seen in a few cases, were attributed to the lack of representativeness of the local sample, the inability of the simulation to capture all variations and the contextual differences between the regions. The concept of updating a simulated travel data set using local sample distributions in order to generate an updated simulated travel data set is explained here. While updating in general was found to move the updated trip attributes in the correct direction and towards the survey values, further testing such as comparing the population values estimated from the survey data and the updated simulated data need to be carried out in order to generate conclusive evidence on the benefit of updating. The main beneficiaries of this method are small/medium metropolitan areas who can use this method to produce synthetic travel data for running their travel demand models at a much lower expense.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 8 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.