SDV

SDV, the Synthetic Data Vault, is an open-source Python library for generating synthetic tabular, relational, and sequential data. It learns statistical patterns from real data and produces artificial datasets that preserve those patterns while containing no real records, which helps teams work around privacy rules, data scarcity, and legal restrictions.

It ships several synthesizers, from the classical GaussianCopula to deep learning models such as CTGAN, a conditional GAN for tabular data, TVAE, a tabular variational autoencoder, and PAR, an RNN for time series. The project is developed by DataCebo and is widely used in research and in ML pipelines that need realistic test data without touching production databases.

Install with pip install sdv, load a dataframe, fit a synthesizer, and sample new rows.