Suggestions on how to contribute towards work on Games-Howell,

[Geoffrey Oxberry]

Statsmodels,

I'm interested in helping the existing work on Games-Howell get to the point where it could be merged into the main branch of the statsmodels repo.

From what I can tell based on the pull requests and issues in the GitHub repo, it seems like there's a pull request that may contribute that functionality in https://github.com/statsmodels/statsmodels/pull/7332, but it seems to have stalled, with no updates in over a year. Would there be any objections in trying to build upon that PR? Given the time elapsed, would it be better to build off that PR or create a new PR off main and try to reproduce the results in that branch?

Thanks for your help, and for building this package,

Geoff

[josef...@gmail.com]

Hi,

Short answer, Ask me again in 2 or 3 weeks. I need to finish up my part of the next, 0.14, release.

longer answer based on what I remember and can see right now

The PR mainly stalled because I need time to think about how to integrate it into statsmodels.

Essentially, games-howell is just a pairwise welch t-test with ranks, and should reuse most of the existing code.

Also, I didn't have an overview of rank based multiple comparison tests.

My plan is to get multiple comparisons out of the sandbox for 0.15. I didn't have time for it for 0.14.

In preparation, I was reading up on multiple comparison procedures using ranks and pseudo-ranks, but mainly in the direction of brunner-munzel and kruskal.

https://github.com/statsmodels/statsmodels/issues/6994

https://github.com/statsmodels/statsmodels/issues/8396

specific to PR:

I don't like extra pep-8/black changes in a PR because it hides the relevant substantive changes

grouping letters should use (my) numpy function instead of networkx

Josef

--
You received this message because you are subscribed to the Google Groups "pystatsmodels" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pystatsmodel...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pystatsmodels/eb046bf4-da33-482e-b1e6-b87b56166106n%40googlegroups.com.