The error that you get in the CIs indicated to you that we don't make deprecation in this way. You should follow the following: https://scikit-learn.org/dev/developers/contributing.html#deprecation

Since we already introduce n_features_in_, we don't have to create a private attribute _n_features to raise the deprecation warning.

Indeed, you only need to add the following property:

# TODO(1.4): remove in 1.4
# mypy error: Decorated property not supported
@deprecated( # type: ignore
    "Attribute `n_features_` was deprecated in version 1.2 and will be "
    "removed in 1.4. Use `n_features_in_` instead."
)
@property
def n_features_(self):
    return self.n_features_in_

In addition, you need to add a test to check that we raise properly the warning:

def test_pca_n_features_deprecation():
    X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [2, 1], [3, 2]])
    pca = PCA(n_components=2).fit(X)
    with pytest.warns(FutureWarning, match="`n_features_` was deprecated"):
        pca.n_features_

We can remove the setting for n_features_in_ since it should be done in the BaseEstimator class already.

Thanks @Kshitij68 LGTM.
@jeremiedbb do you want to have a quick look at it?

LGTM. Thanks @Kshitij68