DOC Clarify 'ovr' as the default strategy in the SVM documentation #29651
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…multi-class classification
ArturoAmorQ
reviewed
Aug 13, 2024
There was a problem hiding this comment.
Thanks for the PR @Shashank1202. I would like to emphasize that this is a duplicate of #29363 and in general it is not a good practice to create duplicates without the OP consent.
I still made some suggestions because I think the general rewording is a better solution.
doc/modules/svm.rst
Outdated
| @@ -119,12 +119,14 @@ Multi-class classification | |||
| -------------------------- | |||
|
|
|||
| :class:`SVC` and :class:`NuSVC` implement the "one-versus-one" | |||
There was a problem hiding this comment.
Suggested change
| :class:`SVC` and :class:`NuSVC` implement the "one-versus-one" | |
| :class:`SVC` and :class:`NuSVC` implement the "one-versus-one" ('ovo') |
doc/modules/svm.rst
Outdated
Comment on lines
124
to
130
| classifiers, with each one training data from two classes. | ||
| By default, however, the `decision_function_shape` parameter is set to `'ovr'` | ||
| (one-vs-rest), providing a consistent interface with other classifiers. | ||
|
|
||
| The ``decision_function_shape`` option allows for the results of the "one-versus-one" | ||
| classifiers to be monotonically transformed into a "one-vs-rest" decision | ||
| function of shape ``(n_samples, n_classes)``. |
There was a problem hiding this comment.
Suggested change
| classifiers, with each one training data from two classes. | |
| By default, however, the `decision_function_shape` parameter is set to `'ovr'` | |
| (one-vs-rest), providing a consistent interface with other classifiers. | |
| The ``decision_function_shape`` option allows for the results of the "one-versus-one" | |
| classifiers to be monotonically transformed into a "one-vs-rest" decision | |
| function of shape ``(n_samples, n_classes)``. | |
| classifiers, each trained on data from two classes. Internally, the solver | |
| always uses this ovo strategy to train the models. However, by default, the | |
| `decision_function_shape` parameter is set to `'ovr'` (one-vs-rest), to have | |
| a consistent interface with other classifiers by transforming the ovo results | |
| into a "one-vs-rest" decision matrix of shape ``(n_samples, n_classes)``. |
|
I've merged the other PR. This PR can now improve upon that. |
jeremiedbb
approved these changes
Jul 25, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fix documentation to specify default decision_function_shape as 'ovr'
This update corrects the scikit-learn documentation for
SVCandNuSVCto clearly state that the default setting for thedecision_function_shapeparameter is'ovr'(one-vs-rest), not'ovo'(one-versus-one). This improves clarity and resolves the discrepancy reported in issue #29340.