Genetic Algorithms in Matrix Representation and Its Application in Synthetic Data

This paper is the implementation of an earlier position paper (Chen, Elliot & Sakshaug, 2016) and explains how to use a new form of genetic algorithms (matrix GAs) to generate synthetic data and provides a proof of concept using a small individual-level microdata set. The new method is able to iteratively optimise synthetic data based on a set of utility parameters until its difference from the original data achieves a desired level. The paper describes the advantages of this method and its potential in synthetic data production. It covers theoretical and computerised model design and specifies further development of this study.