[ https://issues.apache.org/jira/browse/SPARK-16718?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15392662#comment-15392662 ]
Joseph K. Bradley commented on SPARK-16718: ------------------------------------------- Also, it'd be nice to compare with an existing implementation. E.g., if we can compare with R gbm, we can add a unit test doing that, following a few other unit tests in spark.ml. Note: [~vlad.feinberg] is working on this now. > gbm-style treeboost > ------------------- > > Key: SPARK-16718 > URL: https://issues.apache.org/jira/browse/SPARK-16718 > Project: Spark > Issue Type: Sub-task > Components: MLlib > Reporter: Vladimir Feinberg > > As an initial minimal change, we should provide TreeBoost as implemented in > GBM for both L1 and L2 losses: by introducing a new "loss-based" impurity, > tree leafs in GBTs can have loss-optimal predictions for their partition of > the data. > Commit should have evidence of accuracy improvment -- 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