Github user MLnick commented on the pull request:
https://github.com/apache/spark/pull/11601#issuecomment-207279502
@hhbyyh @jkbradley I'm thinking that an `Imputer` is a useful feature
transformer to have. However, perhaps we limit this first implementation to
either (a) single numerical column, or (b) multiple numerical columns. i.e. we
don't support vector column at this time.
Supporting vectors is a bit complex as evidenced by the code here, and will
require changes to `colStats` as well as some deeper changes to
`approxQuantile` if we wish to support array/vectors - and even then, are we
able to support sparse vectors, etc? (and I do think we want to use
`approxQuantile` since computing exactly is expensive). Overall it adds a lot
of complexity, and I'm not convinced it adds that much usefulness.
What do you think?
---
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]
