Practical Synthetic Data Generation: Balancing Privacy and the Broad Availability of Data

Building and testing machine learning models requires access to large  and diverse data. But where can you find usable datasets without running  into privacy issues? This practical book introduces techniques for  generating synthetic data—fake data generated from real data—so you can  perform secondary analysis to do research, understand customer  behaviors, develop new products, or generate new revenue. Data  scientists will learn how synthetic data generation provides a way to  make such data broadly available for secondary purposes while addressing  many privacy concerns. Analysts will learn the principles and steps for  generating synthetic data from real datasets. And business leaders will  see how synthetic data can help accelerate time to a product or  solution. This book describes: Steps for generating synthetic data using  multivariate normal distributions Methods for distribution fitting  covering different goodness-of-fit metrics How to replicate the simple  structure of original data An approach for modeling data structure to  consider complex relationships Multiple approaches and metrics you can  use to assess data utility How analysis performed on real data can be  replicated with synthetic data Privacy implications of synthetic data  and methods to assess identity disclosure.