This directory contains wrappers for the synthesis methods usable in
the pipeline. Each directory contains a run script used by the
pipeline: see Adding another synthesis
method for the
requirements on this.
- CTGAN (Conditional GAN for Tabular data)
- SGF (Synthetic Data Generation Framework)
- synthpop (multiple imputations library in R)
In addition, there are placeholders and worked examples for other methods:
- Base: a synthesizer Python base class
- mice
- SDV (Synthetic Data Vault, parametric model synthesis library in Python, using the multivariate version of the Gaussian Copula)
- simPop (micro-simulation using IPF, Simulated Annealing and model-based synthesis, designed for datasets with household structure). Currently only on embedded Austrian census data set)