[FEA] Import compatible meta-estimators (e.g. pipelines) from sklearn into namespace #3484
Labels
? - Needs Triage
Need team to review and classify
feature request
New feature or request
inactive-30d
Topic for discussion!
Many sklearn meta-estimator features, like Pipeline and GridSearch are fully compatible with cuML models. Many users do not know this. They try to import "cuml.pipeline.Pipeline", find nothing, and assume we do not support these features.
It does not seem wise to simply duplicate these already well-built classes. Alternatively, we could simply import from sklearn into our namespaces to match the existing namespace layout. To accompany this, we would need to add unit test coverage to ensure we never break compatibility.
The text was updated successfully, but these errors were encountered: