Add DistfitFitter distribution fitter wrapping the distfit package

[pyarchana]

Reference Issues/PRs

Towards #1086.

What does this implement/fix? Explain your changes.

Adds DistfitFitter, a skpro distribution fitter that wraps the distfit package's parametric fitting mode: it fits a set of candidate scipy.stats distributions to the data, scores them, and returns the best-scoring one as a fitted skpro scalar distribution. It follows the existing EmpiricalFitter/NormalFitter pattern (_fit/_proba, get_test_params).

Rather than reimplementing pdf/cdf/mean/var per distribution, the returned distribution reuses skpro's existing _ScipyAdapter (skpro/distributions/adapters/scipy/) via a small private wrapper (_DistfitDistribution) that takes the scipy distribution name and shape/loc/scale that distfit produces.

Currently scoped to distfit's method="parametric" mode only. I checked distfit's other three modes (quantile, percentile, discrete) and they return a differently shaped model dict that this adapter doesn't handle yet. Happy to extend if that's wanted before merge, or leave it for a follow-up.

Does your contribution introduce a new dependency? If yes, which one?

distfit, but it's already listed in pyproject.toml's all_extras, so no dependency-list changes were needed.

What should a reviewer concentrate their feedback on?

• Whether scoping to method="parametric" is acceptable for a first version, or whether the other three distfit modes should be wired up before merge.
• The _DistfitDistribution adapter design (reuse of _ScipyAdapter vs. an alternative approach).
• Whether this should wait on [ENH] Add distfitter scitype to base class registry #1093 / [ENH] Add __dynamic_tags__ and __post_init__ to base framework #1094. I verified it does not currently depend on either (both BaseDistFitter and the distfitter registry entry are already on main).

Did you add any tests for the change?

Yes: skpro/distfitter/tests/test_distfitfitter.py, plus it's automatically picked up by the existing generic TestAllDistFitters contract test suite.

Any other comments?

Opened as draft for early feedback on scope. Full local test suite, doctest, and lint (black/isort/flake8/pydocstyle) pass.

Note: docs/source/api_reference/distfitter.rst doesn't exist yet. The whole distfitter module (EmpiricalFitter, NormalFitter, MOMFitter) is currently undocumented there, not just this PR's addition. Flagging it here rather than silently leaving the checklist item unchecked. Happy to add the doc file for the whole module as part of this PR if that's preferred, or leave it for a separate cleanup PR. Let me know which you'd rather I do.

PR checklist

For all contributions

• I've added myself to the list of contributors
• The PR title starts with either [ENH], [MNT], [DOC], or [BUG].

For new estimators

• I've added the estimator to the API reference
• I've added one or more illustrative usage examples to the docstring, in a pydocstyle compliant Examples section.
• If the estimator relies on a soft dependency, I've set the python_dependencies tag and ensured dependency isolation.

[patelchaitany]

i think your github is wrong.

[pyarchana]

i have recently changed the username, i will update it here too!

[patelchaitany]

@pyarchana, add DistfitFitter under Parametric fitters in that RST file

[patelchaitany]

this are the not documentted, Should't we document this?

[fkiraly]

Thanks!

• the distribution returned should be existing distributions in skpro, not a new distribution without any clear features.
• for tests, use run_test_for_class instead of an explicit dependency check

[pyarchana]

Thanks!

• the distribution returned should be existing distributions in skpro, not a new distribution without any clear features.
• for tests, use run_test_for_class instead of an explicit dependency check

thanks!
two ques:

• distfit's "popular" list includes distributions skpro doesn't have like genextreme and dweibull. should i limit what distfit can pick from or raise when it picks one?
• and should _distfit_adapter.py go entirely or kept as a fallback?

[pyarchana]

to be clear on where this is: run_test_for_class is done and conflicts with main are resolved. the distribution mapping is waiting on your answer to the two ques above.