Giters

sdv-dev / Copulas

A library to model multivariate data using copulas.

Home Page:https://sdv.dev/Copulas/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

.fit method parameter

helderc opened this issue · comments

commented

Hi all. First, thanks for the great library!

Considering the GaussianMultivariate as an example, does the input of .fit method expect the last column to be the samples' classes?

I'm asking because it is not mentioned in the documentation [1] that only shows the parameter X. But looking at the example [2], we see the concatenation of X and y. If the example is correct, maybe we should update the doc for clarification?

[1] https://sdv.dev/Copulas/api/copulas.multivariate.gaussian.html#copulas.multivariate.gaussian.GaussianMultivariate

[2] https://sdv.dev/Copulas/tutorials/04_Syntehtic_Data_for_Machine_Learning.html

commented

Hi there @helderc 👋 I'd encourage you to give SDV a try, specifically the GaussianCopulaSynthesizer. This synthesizer class is a drop-in replacement for GaussianMultivariate. SDV is a framework that sits one layer above our Copulas library and offers a significantly better user experience.

I'm also curious to know what your use case is for using gaussian copulas?

commented

Closing this issue for now since GaussianCopulaSynthesizer is our recommended path forward! If you have more questions, you're welcome to join our Slack and we can also discuss more there: https://bit.ly/sdv-slack-invite

commented

Thanks for the feedback @srinify , I'll definitely check SDV out.