Yay! I think this change is more of an "infrastructure" change, it doesn't enable "cool new features" for users. IMHO this is the way to do it. It does make me wonder: do we want to spend effort on user facing features that this change enables? Two immediate ideas I have is supporting more dataframe types as input and doing computation "in the dataframe" instead of in numpy.
T On Thu, 7 May 2026 at 10:42, Adrin via scikit-learn <[email protected]> wrote: > Hi, > > We've added narwhals as a dependency to handle dataframes in scikit-learn > in #31127 <https://github.com/scikit-learn/scikit-learn/pull/31127>. It's > now merged for us to test edge cases before the next release. > > If there are any major issues or objections, we can revert this before the > release. > > This is thanks to everyone who contributed, especially Christian who > authored the PR, as well as narwhals folks holding our hands through the > process. > > Thanks, > Adrin > _______________________________________________ > scikit-learn mailing list -- [email protected] > To unsubscribe send an email to [email protected] > https://mail.python.org/mailman3//lists/scikit-learn.python.org > Member address: [email protected] >
_______________________________________________ scikit-learn mailing list -- [email protected] To unsubscribe send an email to [email protected] https://mail.python.org/mailman3//lists/scikit-learn.python.org Member address: [email protected]
