Josef Perktold
Josef Perktold
additional measures for multicollinearity and enhanced variance inflation factor relevant issues #2376 treatment of constant for variance inflation factor #1669 vif for ridge regression, base for non-loop calculation #2157 general...
I merged treatment effect #8034 essentially in the version where I stopped 4 months ago Docs still need a lot of work, and we need a notebook (unit test has...
branch to see if my dev setup on the new notebook works I will change PR to have various doc improvements
(I don't find a previous issue, but I mentioned versions of this several times.) Using a specific default method in hypothesis tests and confidence intervals commits us to this method...
(continuation of #7720 roadmap for 0.14) - gmm for combining models, equations - #8216 - look at polynomial expansion for distributions again #8326 no specific tasks, just to pick it...
see #2607 This PR implements an unconditional exact test for the equality of two independent proportions, either based on full max or on Berger/Boos confidence interval test. verified and unit...
**update** this has been generalized to MultiLinkModel, not restricted to mean-variance models anymore. see #7639 getting started code based on miscmodels TLinearModel and `__init__` from BetaModel. class structure not clear...
closes #7778 This will be a continuation of #7778 now starting with distribution dfamilies similar to genmod families but focused on MLE with scipy optimizers I'm planning to add cleaner...
Patsy currently does not provide an factor/categorical encoding that does not drop columns that are already linear dependent available in "earlier" factor columns https://github.com/pydata/patsy/issues/60 https://github.com/pydata/patsy/issues/42 (discussed with Nathaniel at pycon...
I'm planning to merge some PRs that will need follow-up before the 0.14 release. This is mainly because I have to bring different parts together and I want to postpone...