Copulas

Why we included this project

Copulas handles a narrow but common problem: you have a table of numerical data and want to learn its joint distribution well enough to sample new rows that look statistically like the original. It does this with copula functions, offering Gaussian, Archimedean, and vine copulas paired with a range of univariate marginals. For teams generating synthetic tabular data for testing, sharing, or augmenting datasets without exposing real records, this is a practical building block. The library also exposes the learned parameters directly, so you can inspect and tune the model rather than treating it as a black box, and it includes simple visualizations to compare real and synthetic columns. Since it is part of the Synthetic Data Vault project from DataCebo.