sktime/sktime

[ENH] employ AI plus human oversight to complete remaining testing exceptions in `tests._config`

Ouverte

#10 652 ouverte le 26 juil. 2026

 (8 commentaires) (0 réaction) (0 personne assignée)Python (1 192 forks)batch import
enhancementgood first issuemodule:base-framework

Métriques du dépôt

Stars
 (7 162 étoiles)
Métriques de merge PR
 (Merge moyen 26j 10h) (86 PRs mergées en 30 j)

Description

There are three tasks in this issue which can be worked on separately.

1. fixing all remaining missing parameter sets as in issue #3429

The issue #3429 has been open for long as an issue for new contributors to learn about contributing to sktime.

A few factors are making this issue less helpful to new contributors, so we should use AI to go through all remaining cases and finalize it.

The contributing factors why I think this can now be wrapped up:

  • AI assisted coding is now so prevalent that the learning effect has been reduced
  • most estimators have been handled already, leaving only a short list left over
  • the checklist in the issue is no longer updated regularly

This issue is about wrapping up #3429 in a single go - this can be done easily in an AI assisted way. However, a human MUST go through the estimators one-by-one to check whether the changes are sensible.

2. migrate test skip lists to estimator tags

In addition - possibly a separate PR - the EXCLUDED_TESTS, EXCLUDE_ESTIMATORS should be migrated to the estimator tags as well.

3. remove test skip lists and respective hooks.

In a third, final PR - or the same PR - the EXCLUDED_TESTS_BY_TEST, EXCLUDED_TESTS, EXCLUDE_ESTIMATORS and EXCLUDED_TESTS lists should be removed, and the hooks injecting them to the test framework should be removed as well.

Someone working on this issue should say which of the three possible PR scopes they are working on.

Guide contributeur