[
https://issues.apache.org/jira/browse/SPARK-7423?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14539311#comment-14539311
]
George Dittmar commented on SPARK-7423:
---------------------------------------
Will have a pr for this soon. Just made the changes to the linked JIRA in
another branch and finishing up tests now.
> spark.ml Classifier predict should not convert vectors to dense format
> ----------------------------------------------------------------------
>
> Key: SPARK-7423
> URL: https://issues.apache.org/jira/browse/SPARK-7423
> Project: Spark
> Issue Type: Improvement
> Components: ML
> Reporter: Joseph K. Bradley
> Priority: Minor
> Labels: starter
>
> spark.ml.classification.ClassificationModel and
> ProbabilisticClassificationModel both use DenseVector.argmax to implement
> prediction (computing the prediction from the rawPrediction or probability
> Vectors). It would be best to implement argmax for Vector and SparseVector
> and use Vector.argmax, rather than converting Vectors to dense format.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
