[
https://issues.apache.org/jira/browse/SPARK-19054?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15793987#comment-15793987
]
Apache Spark commented on SPARK-19054:
--------------------------------------
User 'zhengruifeng' has created a pull request for this issue:
https://github.com/apache/spark/pull/16453
> Eliminate extra pass in NB
> --------------------------
>
> Key: SPARK-19054
> URL: https://issues.apache.org/jira/browse/SPARK-19054
> Project: Spark
> Issue Type: Improvement
> Components: ML
> Reporter: zhengruifeng
> Priority: Minor
>
> {{val numClasses = getNumClasses(dataset)}} in {{NaiveBayes}} are used to
> check the length of thresholds only when {{positiveLabel &&
> isDefined(thresholds)}} is true.
> In the case that {{positiveLabel}} is true and {{isDefined(thresholds)}} is
> false, we don't need to calculate {{getNumClasses}} which may cause a pass on
> the whole data.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
