[ https://issues.apache.org/jira/browse/SPARK-11834?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15241511#comment-15241511 ]
Jeremy edited comment on SPARK-11834 at 4/14/16 5:16 PM: --------------------------------------------------------- Looking into this JIRA, checkThresholdConsistency() should be called in getThreshold() in ml.classification.LogisticRegression.scala, but I can set inconsistent values with setThreshold() and setThresholds() and get predictions that run cleanly and that are consistent with the value for setThreshold(). Checking the documentation, it has been updated to reflect that only binary classes are supported. Pictures here: https://goo.gl/wCpJzx was (Author: jeremynixon): Looking into this JIRA, checkThresholdConsistency() and validateParams() are not called in ml.classification.LogisticRegression.scala, and so I can set inconsistent values with setThreshold() and setThresholds() and get predictions that run cleanly and that are consistent with the value for setThreshold(). Checking the documentation, it has been updated to reflect that only binary classes are supported. Pictures here: https://goo.gl/wCpJzx > Ignore thresholds in LogisticRegression and update documentation > ---------------------------------------------------------------- > > Key: SPARK-11834 > URL: https://issues.apache.org/jira/browse/SPARK-11834 > Project: Spark > Issue Type: Improvement > Components: Documentation, ML > Affects Versions: 1.6.0 > Reporter: Xiangrui Meng > Assignee: Xiangrui Meng > Priority: Minor > > ml.LogisticRegression does not support multiclass yet. So we should ignore > `thresholds` and update the documentation. In the next release, we can do > SPARK-11543. -- 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