Using any univariate distribution from scipy

[npatki]

Problem Description

The copulas univariate package contains 8 different distribution types to choose from. However, the underlying data science library scipy contains a larger variety of univariate distributions.

Is there a way we can seamlessly integrate any available scipy distribution with the copulas package? This would give me more flexibility when modeling my data.

For example when using the multivariate Gaussian Copula, I may want to try using a Weibull distribution instead of any of the selected ones.

Related Issues

There are a number of issues already where users have been requesting different distributions be added:

• Adding Weibull for univariate  #217 (Weibull)
• Hyper Laplace Distribution #193 (Hyper Laplace)
• (From Slack) Requests for Poisson and Negative Binomial distributions

[npatki]

Related (duplicate) issue: #204

[npatki]

This functionality can now be accessed via the XGCSynthesizer.