[
https://issues.apache.org/jira/browse/SPARK-16365?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15382197#comment-15382197
]
Sean Owen commented on SPARK-16365:
-----------------------------------
My perspective is that this is a concern that is separable from Spark, and,
Spark can't and shouldn't try to do everything. Real-time scoring is a fairly
different creature, and we have clean interfaces between the model building and
scoring worlds already like PMML. So, there's a fairly bright line here.
I'm skeptical that Spark will produce something finished here, given the
experience of implementing export capability here. For whatever reason, I just
don't think this ever gets well thought out and implemented in MLlib, and would
rather not continue trying until it's clear there's an organized effort. That
would begin by finishing and cleaning up the export/import situation.
I'm also skeptical that something appreciably better than existing libraries
would be implemented by Spark.
Who knows; things change. My suggested conclusion here is more along the "do
nothing" lines, or at least, "shore up existing capability first".
> Ideas for moving "mllib-local" forward
> --------------------------------------
>
> Key: SPARK-16365
> URL: https://issues.apache.org/jira/browse/SPARK-16365
> Project: Spark
> Issue Type: Brainstorming
> Components: ML
> Reporter: Nick Pentreath
>
> Since SPARK-13944 is all done, we should all think about what the "next
> steps" might be for {{mllib-local}}. E.g., it could be "improve Spark's
> linear algebra", or "investigate how we will implement local models/pipelines
> in Spark", etc.
> This ticket is for comments, ideas, brainstormings and PoCs. The separation
> of linalg into a standalone project turned out to be significantly more
> complex than originally expected. So I vote we devote sufficient discussion
> and time to planning out the next move :)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
