[
https://issues.apache.org/jira/browse/SPARK-7130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14995375#comment-14995375
]
Joseph K. Bradley commented on SPARK-7130:
------------------------------------------
I think it reaches a little farther than that. The logic for deciding whether
to sample is in BaggedPoint.scala, though you're correct that the line 88
affects it. This JIRA description was a little out-of-date; I'll update it to
indicate that the implementations are still shared.
> spark.ml RandomForest* should always do bootstrapping
> -----------------------------------------------------
>
> Key: SPARK-7130
> URL: https://issues.apache.org/jira/browse/SPARK-7130
> Project: Spark
> Issue Type: Improvement
> Components: ML
> Affects Versions: 1.4.0
> Reporter: Joseph K. Bradley
> Priority: Minor
>
> Currently, spark.ml RandomForest does not do bootstrapping if numTrees = 1.
> For consistency and a simpler API, it should always do bootstrapping. The
> current behavior is an artifact of the old API, in which RandomForest and
> DecisionTree share the same implementation. This change should happen after
> the implementation is moved to spark.ml (which we need to do so that the
> implementation can be generalized).
--
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
