Creating a synthetic population: A comparison of tools

S. Jain, N. Ronald, S. Winter, “Creating a synthetic population: A comparison of tools”, inĀ Proceedings of 3rdĀ Conference of Transportation Research Group, Dec. 2015

Shubham Jain will present his work titled “Creating a synthetic population: A comparison of tools” in Conference of Transportation Research Group in Kolkata, India. In his MPhil research, he plans to use population characteristics and current activity-travel pattern of Melbourne from population census data and existing, self-completed household travel survey (VISTA) in conjunction with insights from usage pattern of some of the existing collaborative transport services in different regions of the world to identify demand of these services here. This kind of demand modelling requires microdata of population at household and personal level as a key input. Unfortunately due to privacy constraints, such data is often not available. To fulfill this lack of data, population can be synthesised to represent actual demographics of study area as per population census. This paper presents the process of creating a synthetic population for a metropolis for the case of Greater Melbourne using 2011 Population Census. Microdata was created for households and persons in Greater Melbourne at statistical area level 1 (SA1). PopSynWin (developed by University of Illinois at Chicago in 2008), which is based on Iterative Proportional Fitting (IPF) algorithm, and PopGen (Population Generator, developed by Arizona State University in 2009), which is based on Iterative Proportional Update (IPU) algorithm, were used as tools for this purpose, generating two different synthetic populations. Microdata within these synthetic populations were aggregated to validate against actual aggregate census data to evaluate its representativeness to original population. Finally, both the generated populations were compared in various ways to use the more accurate of them for further work.