Jérôme Dockès

@Vincent-Maladiere why not! that can be a nice "good first issue" for the upcoming sprint

I think the `plot_marker` should return something that you can use to build the colorbar rather than having to recreate the cmap and normalize yourself having the plotting functions returning...

the ndcg score in this case is 0 / 0 so maybe the output should be NaN? and skip nans when computing the overall score. Otherwise if we don't want...

thanks for reporting. dependencies are checked when importing the package (not just when installing it), and the machinery that does that uses `packaging`. This was added in https://github.com/skrub-data/skrub/pull/365 maybe @GaelVaroquaux...

relevant discussion in scikit-learn: https://github.com/scikit-learn/scikit-learn/pull/22722 I would say for now I am +0 on this, I suggest we discuss it in one of the skrub meetings. some potential drawbacks IMHO:...

> At least now, @TheooJ and @jeromedockes are aware that this is existing and it would cost defining a dictionary ;) indeed, thanks :) also, this doesn't have to be...

the failure in the min-deps environment is not related to this pr; the fix is in #1122

I think "ignore" is a bit ambiguous -- if I ignore a column does it mean I don't use it / leave it out of the output? so I would...

I find it a bit weird to have a DropColumnIfNull that does not drop the column and just raises an exception. maybe it should be named something like CheckNulls or...

> Consider a situation where you have a TableVectorizer in a scikit-learn pipeline, trained in a production environment every week. If you built this pipeline assuming you would have access...