Comparative analysis of synthetic data vault (SDV) and privacy-preserving graphical models with latent variables (PRIVLAVA) in synthetic data generation for relational databases: an evaluation study

Synthetic data generation has become a pivotal technique in overcoming the limitations posed by data privacy concerns and data scarcity in developing and testing data-driven applications. This study aims to compare two leading synthetic data generation tools, “Synthetic Data Vault” (SDV) and “Privacy-Preserving graphical models with latent variables” (PrivLava), focusing on relational databases. By evaluating their performance across various metrics, we seek to understand their strengths, weaknesses and best application contexts.