[
https://issues.apache.org/jira/browse/SPARK-12183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15046094#comment-15046094
]
Seth Hendrickson commented on SPARK-12183:
------------------------------------------
Do you mean these are lower priority than other fixes within ml/mllib or
specifically within tree based packages? I understand the former, but I think
this is an important part of a larger effort to port tree based models to
spark.ml. There are at least a couple of items related to GBT that more or less
require GBT to move to spark.ml. Removing the duplicate code of Random Forest
will make it easier to port tree-based helper classes (ImpurityStats, Impurity,
Strategy, etc...), which will in turn make it easier to move the GBT
implementation.
I don't have as good of a grasp on the overall big picture, of course, so I'm
interested to hear your thoughts on this. Thanks for creating the Jira!
> Remove spark.mllib tree, forest implementations and use spark.ml
> ----------------------------------------------------------------
>
> Key: SPARK-12183
> URL: https://issues.apache.org/jira/browse/SPARK-12183
> Project: Spark
> Issue Type: Improvement
> Components: ML, MLlib
> Reporter: Joseph K. Bradley
> Priority: Minor
>
> This JIRA is for replacing the spark.mllib decision tree and random forest
> implementations with the one from spark.ml. The spark.ml one should be used
> as a wrapper. This should involve moving the implementation, but should
> probably not require changing the tests (much).
> This blocks on 1 improvement to spark.mllib which needs to be ported to
> spark.ml: [SPARK-10064]
--
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
