Comparative Study of Differentially Private Synthetic Data Algorithms from the NIST PSCR Differential Privacy Synthetic Data Challenge

Differentially private synthetic data generation offers a recent solution to release analytically useful while preserving the privacy of individuals in data. In order utilize these algorithms for public policy decisions, policymakers need an accurate understanding algorithms' comparative performance. Correspondingly, practitioners also require standard metrics evaluating analytic qualities this paper, we present in-depth evaluation several differentially using actual sets created by contestants National Institute Standards and Technology Public Safety Communications Research (NIST PSCR) Division's ``"Differential Privacy Synthetic Data Challenge." We offer analyses based on both accuracy they create their usability potential providers. frame methods used NIST PSCR challenge within broader literature. implement additional utility metrics, including two our own, compare mechanism three categories. Our assessment synthesis quality shows relative usefulness, general strengths weaknesses, preferred choices metrics. Finally describe implications seeking future products.