sdv-dev/Copulas
A library to model multivariate data using copulas.
What it solves
It allows users to create synthetic numerical data that maintains the same statistical properties as a real dataset. This is useful for generating new data samples that mimic the behavior and correlations of original multivariate distributions.
How it works
The library uses copula functions to model multivariate distributions. It fits a model to a table of numerical data using various univariate distributions and copula types—such as Gaussian, Archimedian, and Vine Copulas—and then samples from that learned distribution to generate new data points.
Who it’s for
Data scientists and researchers who need to generate high-fidelity synthetic numerical datasets for testing, privacy, or data augmentation.
Highlights
- Multiple Copula Types: Supports Gaussian, Archimedian, and Vine Copulas for flexible modeling.
- Visual Comparison: Includes built-in tools to compare real and synthetic data using 1D histograms, 2D scatterplots, and 3D scatterplots.
- Parameter Control: Provides full access to the model's internal parameters for manual tuning and adjustment.