[ https://issues.apache.org/jira/browse/SPARK-21812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16387068#comment-16387068 ]
Bryan Cutler commented on SPARK-21812: -------------------------------------- Adding SPARK-15009 as an example of how to restructure the model class hierarchy, using CountVectorizer, to own params instead of depending on transfer from Scala. > PySpark ML Models should not depend transfering params from Java > ---------------------------------------------------------------- > > Key: SPARK-21812 > URL: https://issues.apache.org/jira/browse/SPARK-21812 > Project: Spark > Issue Type: Improvement > Components: ML, PySpark > Affects Versions: 2.3.0 > Reporter: holdenk > Priority: Major > > After SPARK-10931 we should fix this so that the Python parameters are > explicitly defined instead of relying on copying them from Java. This can be > done in batches of models as sub issues since the number of params to be > update could be quite large. -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org