[Datasets] Allow MultiHotEncoder
to encode arrays
#31365
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Signed-off-by: Balaji Veeramani [email protected]
Why are these changes needed?
tl;dr:
MultiHotEncoder
doesn't work with Arrow datasets. This PR updates theMultiHotEncoder
implementation to fix the issue.MultiHotEncoder
implements the_transform_pandas
method. So, whenMultiHotEncoder
transforms a dataset, it converts Arrow blocks to pandas DataFrames. During the conversion, lists are converted to ndarrays. This causesMultiHotEncoder
to error, becauseMultiHotEncoder
doesn't support ndarrays (only built-in lists).Related issue number
Fixes #31353
Checks
git commit -s
) in this PR.scripts/format.sh
to lint the changes in this PR.