[ https://issues.apache.org/jira/browse/SPARK-11861?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15020829#comment-15020829 ]
Seth Hendrickson commented on SPARK-11861: ------------------------------------------ It so happens that the feature importance implementation for random forests involves computing feature importances for individual decision trees. So, yes, in this case the Jira will only involve exposing an already implemented result. I can' think of a reason to have feature importance available for random forests, but not decision trees, especially since the work has already been done. Entirely possible I'm overlooking something though. Additionally, [SPARK-11730|https://issues.apache.org/jira/browse/SPARK-11730] requests feature importance to be available for GBTs, which again will require it to be available for individual trees. That is actually the reason I created the Jira in the first place. I'm open to feedback. Thanks! > Feature importances for decision trees > -------------------------------------- > > Key: SPARK-11861 > URL: https://issues.apache.org/jira/browse/SPARK-11861 > Project: Spark > Issue Type: New Feature > Components: ML > Reporter: Seth Hendrickson > Priority: Minor > > Feature importances should be added to decision trees leveraging the feature > importance implementation for Random Forests. -- 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