Metropolis-Hasting based synthesis of all-day round-trips from OD-matrices and limited population segmentation

Abstract

This work considers the synthesis of coherent and distinguishable round trips from existing travel demand data (OD-matrices) in order to enable person-level mobility simulations with agent-based simulations such as MATSim (Horni et al., 2016). Five distinct population groups are considered and the spatial distribution of their home locations over the study region is reconstructed. Also, the spatial distribution of work locations is reproduced. The round trips are equipped with a plausible temporal structure by reproducing activity durations and departure times derived from a household travel survey. The Metropolis-Hastings (Hastings, 1970) algorithm is used to sample from a probability distribution defined based on input data drawn from an upstream four-step travel demand model. The method is applied to the city of Vienna, resulting in the generation of a population of 50,000 representative round trip makers.