SynthOpt is a Python package for creating different levels of synthetic data and optimising the privacy/utility trade-off for different requirements. Requires python version 3.8.x, 3.9.x or 3.10.x.
There are three aspects to SythOpt:
Generate different levels of fidelity of synthetic data. You can create synthetic data either from metadata or from datasets (both single-table and multi-table). Different synthesiser models are available to choose from, including differentially private synthesisers.
Evaluate privacy, utility and quality of synthetic data using our metrics, as well as create visualisations and reports.
Optimise the privacy/utility trade-off depending on your requirements - generate highly private synthetic data, highly useful synthetic data, or find the right balance in between.
Here is a link to the documentation for this package: SynthOpt Documentation