DOC Remove mention of deprecated multi_class in LogisticRegression
#29998
Conversation
|
Since |
doc/modules/linear_model.rst
Outdated
| @@ -1000,6 +1000,8 @@ logistic regression, see also `log-linear model | |||
| | `ElasticNet` | :math:`\frac{1 - \rho}{2}\|W\|_F^2 + \rho \|W\|_{1,1}` | | |||
| +----------------+----------------------------------------------------------------------------------+ | |||
|
|
|||
| .. _Logistic_regression_solvers: | |||
There was a problem hiding this comment.
Maybe drop the capital letter if possible.
There was a problem hiding this comment.
Done, I agree its weird with the capital. I initially thought I should copy what was used as reference header for logistic regression main section:
scikit-learn/doc/modules/linear_model.rst
Line 852 in 8d01acf
| @@ -827,6 +821,11 @@ class LogisticRegression(LinearClassifierMixin, SparseCoefMixin, BaseEstimator): | |||
| the L2 penalty. The Elastic-Net regularization is only supported by the | |||
| 'saga' solver. | |||
|
|
|||
| For :term:`multiclass` problems, only 'newton-cg', 'sag', 'saga' and 'lbfgs' | |||
There was a problem hiding this comment.
@jeremiedbb Since we are in the long summary and we have an alternative to use OvR, I would advocate from now on to use OneVsRestClassifier. It looks legit to me.
|
Thanks @lucyleeow LGTM. |
Reference Issues/PRs
multi_classdeprecated in #28703What does this implement/fix? Explain your changes.
multi_classinLogisticRegression. Also moves the section on multiclass down, as it seems to make sense to introduce the class first.Any other comments?