Creating Synthetic Data Sets for Microsimulation models

Abstract

Several approaches have recently been proposed by different researchers in attempting to bridge the gap between available data and high data needs. The approaches generally rely on the different forms of the available data to synthetically create further data. Population synthesis frequently involves generation of large amounts of data through procedures that are generally computationally intensive. In the past, limitations - both computational and methodological, have limited the use and development of more sophisticated approaches. Recent developments however, have broadened the possibilities available to researchers. As a consequence, new methodologies to generate synthetic data are being developed and proposed. Most of these new methods have not yet been understood by the research community as they are in their early development stages and policy makers are still reluctant to adopt these methods. In this thesis, we have attempted to enhance further understanding of these methods through applying the methods to different research problems of focus that required creation or availing new data. A great part of this thesis has thus been devoted to reviewing and assessing different methods under different scenarios. This research has further served to demonstrate, review and/or propose modifications to existing approaches geared towards providing useful and important information regarding their application. More to this, an integrated model has been proposed for generating synthetic populations for Flanders, which was the main goal of this study.