Generation of a synthetic population for agent-based transportmodelling with small sample travel survey data using statistical raster census data

Samuel Felbermair*, Florian Lammer, Eva Trausinger-Binder, Cornelia Hebenstreit

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

This paper presents a step-by-step method to generate a synthetic population for agent-based transport modelling asinput to MATSim software, which requires an activity chain for each agent. We make use of high spatial resolutionstatistical raster (250 m) census data, applying all calculations at this scale. Due to the small sample, size of travel surveydata an Iterative Proportional Fitting method is not suitable. Therefore, we devise a method utilizing Bayesian networks,maximum likelihood and Markov Chain Monte Carlo simulation to reproduce attribute distribution and fit to rastermargins. Stratified sampling along households is employed to generate activity chains for the synthetic population.
Original languageEnglish
Pages (from-to)9-17
Number of pages9
JournalInternational Journal of Traffic and Transportation Management
Volume2
Issue number2
Publication statusPublished - 2020

Fingerprint

Dive into the research topics of 'Generation of a synthetic population for agent-based transportmodelling with small sample travel survey data using statistical raster census data'. Together they form a unique fingerprint.

Cite this