Github user mengxr commented on the pull request:
https://github.com/apache/spark/pull/3099#issuecomment-62684488
@shivaram I found that this is not the only place we need
`Estimator`/`Transformer`. In `CrossValidator`, we need to set the underlying
estimator. If we replace it by `PipelineStage` there, I'm certain that it is
going to change soon. If you suggest removing `CrossValidator`, as I mentioned,
I've been thinking about this for the entire day and I felt that providing a
pipeline without tools for tuning doesn't make sense. Focusing on user API is
indeed our goal here. On the developer API side, it is easy to tell
`HasMaxIter`/`UnaryTransformer` are developer APIs, while it is actually hard
to tell whether `Estimator`, `Transformer`, and `Evaluator` are developer APIs.
Let me try to hide APIs as much as I can and see how it goes.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
