Also note that the failing check is unrelated with your pull request.

Thank you for reviewing. I modified codes.

Could you review again?

I don't think we should add such a note as part of the docstring because this would be valid for any scikit-learn estimators and we don't want to make all the docstring of all the methods of all scikit-learn estimator that verbose.

Instead, I propose 2.5 things to move forward (should be addressed in distinct PRs):

• for the specific case of spectral_clustering, since SpectralClustering accepts np.matrix I think we should change spectral_clustering to be consistent with it as proposed initially in [MRG] FIX sklearn.cluster.spectral_clustering fails with np.matrix input (#10993) #11070.

• alternatively, for the specific case of spectral_clustering, we could just raise a TypeError directly with a link to https://numpy.org/doc/stable/reference/generated/numpy.matrix.html in the error message as was suggested in sklearn.cluster.spectral_clustering fails with np.matrix input #10993 (comment) .

• for the general case of np.matrix in scikit-learn, I think we should improve check_array to warn the user (using a FutureWarning to state that np.matrix usage is deprecated in scikit-learn 1.0 and will raise a TypeError in the future 1.2 and recommend to explicitly convert to numpy arrays with np.asarray instead with a link to https://numpy.org/doc/stable/reference/generated/numpy.matrix.html). Edit: done in API Deprecates support for np.matrix in check_array #20165.

Sorry @yoh1496 it seems that #10993 should take a different path I didn't understand... @ogrisel do you mind commenting directly in the issue? It makes it no longer Easy neither Good first issue.

Let's close this. Please feel free to open a new PR to address one of the first two items.

Personally I would be in favor of the second solution: just raise TypeError with an informative error message.