tensorflow / tensorflow/probability

Copulas and discrete random variables

Open
#1,460 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Jupyter Notebook
Stars
4.4k
Forks
1.1k
PR merge metrics
No merged PRs in 30d

Description

I've been following Copulas Primer which constructs a bivariate distribution using Kumaraswamy and Gumbel marginals, with the copula inducing correlation between the two.

This seems possible since there is tfb.KumaraswamyCDF and tfb.GumbelCDF bijectors which transform samples from [0, 1]^2 -> R^2. There aren't however any such CDF bijectors for discrete distributions, e.g. tfd.Poisson, which makes sense since the CDF is a step function.

My question is - is there any simple way to code up bivariate distributions using copulas with discrete random variables in TFP? I know there might be some issues with copulas and discrete outcomes, but it'd be great to have a play around if possible.

Thanks!

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the linked Copulas Primer and compare its KumaraswamyCDF and GumbelCDF bijectors with the absence of a CDF bijector for tfd.Poisson. Determine whether TFP can support copula-based bivariate distributions with discrete random variables, and document or prototype the resulting approach if the scope is defined.

Written by the indexing model from the issue text.

Assessment

Tech stack
tensorflow
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.