Thumbnail Image

Simulation a Close-to-Reality Synthetic Population of the Greater Accra Region

Frazier, Tyler J.; Alfons, Andreas

VSP Working Paper

The purpose of this research is to simulate a synthetic population of the Greater Accra Metropolitan Region (GAMA) from the 2005 Ghana Living Standards Survey (GLSS5) for use in the Greater Accra Urban Simulation System (GAUSS). A primary goal in simulating the synthetic population of GAMA is to employ a method which generates close-to-reality population data rather than repeatedly drawing samples. In order to generate close-to-reality synthetic data, combinations which were not represented in the original household survey but are likely to occur in the true population must occur in the synthetically generated data. The author estimates the conditional distributions with multinomial logistic regression models in order to simulate categorical and continuous variables. The simulation of random zeros as opposed to structural zeros, are also reflected in the synthetically generated Greater Accra population. One of the main purposes for avoiding pure replication of units from the underlying sample is because this generally leads to small variability of units within smaller subgroups, which results in an increase in unrealistic model behavior when population data is used as input for agent-based simulations of urban dynamics.