Github user srowen commented on the pull request:
https://github.com/apache/spark/pull/3062#issuecomment-72565380
@mengxr I like keeping the export together in one package. Since I imagine
there will likely only ever be 1 or 2 model export formats (really, I imagine
1) I think using a trait is OK. That is, I don't feel too bad about treating
PMML fairly specially.
The `PMML` abstraction from JPMML is pretty solid and could be the return
type of such a `toPMML` API method, although I hesitate to bake in a
third-party library class to the public API. I suppose the return type could
just be `String`, containing an XML document, or at best a generic object
representing the XML DOM?
Yes, definitely mark things private / developer API until this stabilizes.
Writing PMML serializations is a moderate PITA, but using the JPMML library
internally makes it not so bad. I have some serializations already written I
could chuck into the pot once the framework is in place (decision forest,
k-means clustering model)
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
