BUG/ENH: transparams in loglike for non-newton methods
Nobody has claimed this yet.
Assessment
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Newbie friendliness
- 35/100
Research direction
Start by tracing transparams in discrete.NegativeBinomial and GeneralizedPoisson, then compare the new generic scipy.optimize path with the existing newton and ncg whitelist. Done means optimizer selection no longer applies an incompatible transformation for Hessian-based methods, while the GeneralizedPoisson underdispersion case can be handled explicitly.
Written by the indexing model from the issue text.
Description
discrete.NegativeBinomial and I think several other models including GeneralizedPoisson in #3727 use internal transformation to impose non-negativity.
Because hessian is not define for the transformed dispersion parameter, the transform is only used for non-Newton methods, i.e. optimizers that don't use Hessian.
The problem is that newton methods use a white-list newton, ncg which does not include the new access through generic scipy.optimize which allows the use of newer newton methods like dogleg.
We need to expand or come up with an alternative to the white-list.
Special case, in GeneralizedPoisson we can add an option to turn it off because the distribution is also defined for underdispersion (negative dispersion parameter, that is not too negative)
- Dominant language
- Python
- Stars
- 11.6k
- Forks
- 3.6k
- Avg merge
- 7h 37m
- Merged PRs (30d)
- 96
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from statsmodels/statsmodels
-
Difficulty 1/5 Under an hour Newbie friendliness 90/100
statsmodels/statsmodels#10271 ·
-
type-bug
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
statsmodels/statsmodels#10269 ·
-
Documentation
Difficulty 2/5 1-3 hours Newbie friendliness 92/100
statsmodels/statsmodels#10266 ·
-
Difficulty 1/5 Under an hour Newbie friendliness 78/100
statsmodels/statsmodels#9627 · 1 comment ·
-
Difficulty 2/5 1-3 hours Newbie friendliness 62/100
statsmodels/statsmodels#9293 · 1 comment ·
All issues in statsmodels/statsmodels
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
bancolombia/sentinel#23 ·
-
test md OpenCI
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
-
integration:quickjs org:external priority:backlog topic:code-interpreter topic:middleware type:feature
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
langchain-ai/deepagents#6450 ·
-
bug client
Difficulty 2/5 1-3 hours Newbie friendliness 88/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 74/100