FIX potential AttributeError when passing list to ValidationCurveDisplay - #27311
Conversation
|
I prefer to validate So we can have param_range=np.array(param_range, copy=False)We will need a non-regression test as well that call the display by passing a list as |
ogrisel
left a comment
There was a problem hiding this comment.
Thanks for the PR. Was this already a bug in scikit-learn 1.3 or was this a bug in a new option first introduced for scikit-learn 1.4?
If the former, we need a dedicated changelog entry.
Co-authored-by: Olivier Grisel <[email protected]>
…nto fix_AttributeError
glemaitre
left a comment
There was a problem hiding this comment.
This should solve the CIs issues.
Co-authored-by: Guillaume Lemaitre <[email protected]>
|
We have the unrelated |
|
This is a related failure :) assert_array_equal(display.param_range, param_range) |
…nto fix_AttributeError
…arn into fix_AttributeError
|
Merging since the previous build already passed. |
…lay (scikit-learn#27311) Co-authored-by: ArturoAmorQ <[email protected]> Co-authored-by: Olivier Grisel <[email protected]> Co-authored-by: Guillaume Lemaitre <[email protected]>
…lay (#27311) Co-authored-by: ArturoAmorQ <[email protected]> Co-authored-by: Olivier Grisel <[email protected]> Co-authored-by: Guillaume Lemaitre <[email protected]>
…lay (scikit-learn#27311) Co-authored-by: ArturoAmorQ <[email protected]> Co-authored-by: Olivier Grisel <[email protected]> Co-authored-by: Guillaume Lemaitre <[email protected]>
Reference Issues/PRs
What does this implement/fix? Explain your changes.
Passing a list to the
param_rangeparameter inValidationCurveDisplaymay rise a hard to debugif the spacing in the grid looks appropriate to use a "symlog" according to the current logic in the private function
_interval_max_min_ratio.Hopefully this PR fixes it.
Any other comments?