Closed glemaitre closed 2 years ago
+1 That will be a great addition to sklearn !
+1 as well
+1
+1
+1
Just to clarify: my approval for this PR stands for a +1 for the vote.
+1 with minor adjustment to verbose_feature_names_out
in ColumnTransformer
: https://github.com/scikit-learn/enhancement_proposals/pull/60
The SLEP says nothing of the types of the feature names (str vs other) or the collection of them (list vs array). Should it? Otherwise, I'm happy with this.
(I have wondered, however, whether we give ourselves headaches by trying to ascribe names, rather than objects with structure and behaviour, to the representation of features. But certainly, we'd be giving ourselves more headaches if we wanted to attach such structured objects to DataFrames!)
The SLEP says nothing of the types of the feature names (str vs other) or the collection of them (list vs array).
This is probably something that we should clarify by amending the current proposal as done there: https://github.com/scikit-learn/enhancement_proposals/pull/59#issuecomment-956441998
Opened #61 to add more details regarding the container, dtype, and strings.
The vote has been open for a month and a consensus has been reached. By our decision-making process, I changed the status of this SLEP to "Accepted".
We can provide post-history changes to give more details and improve the current proposal.
ping @adrinjalali @thomasjpfan did I miss anything in the process? I let you review and merge this PR if everything is OK.
I agree, and I think we've gone forward with the implementation and we seem to have a good consensus now.
In this PR, I propose a vote for SLEP007.
As per our governance, the vote will last 1 month and will be announced shortly on the internal mailing list. We will seek a consensus during this month. If no consensus is found, the decision will be escalated to the Technical Committee (TC).