Twinify
Privacy-preserving synthetic data generation with differential privacy.
About
Twinify is an open-source Python package for generating synthetic twins of sensitive tabular datasets using differential privacy techniques. Developed by the DPBayes group, it implements approaches like NAPSU-MQ and differentially private variational inference (DPVI), letting teams share realistic synthetic data without exposing individual records.
The package integrates with NumPyro for probabilistic modeling and supports automatic modeling for users without deep statistics expertise. It is aimed at researchers and engineers in healthcare, finance, and government who need to work with sensitive data under strict privacy constraints.
What it can do
- Differentially private synthetic data generation
- NAPSU-MQ and DPVI modeling approaches
- NumPyro integration for custom models
- Open-source with PyPI distribution
Setup
Install via pip with pip install twinify, then follow the example notebooks for NAPSU-MQ or DPVI modeling.
Prerequisites
Python 3, a tabular dataset, and basic familiarity with probabilistic modeling.
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 131; Trending score 68/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.