[jira] [Commented] (SPARK-12367) NoSuchElementException during prediction with Random Forest Regressor

["Joseph K. Bradley (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-12367?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15060720#comment-15060720 ] 

Joseph K. Bradley commented on SPARK-12367:
-------------------------------------------

This is likely caused by a feature value 1.0 which did not appear in the training data.  That prevents VectorIndexer from knowing about that value, so it does not have a corresponding index when trying to transform the test data.  It will be handled by [SPARK-12375].

> NoSuchElementException during prediction with Random Forest Regressor
> ---------------------------------------------------------------------
>
>                 Key: SPARK-12367
>                 URL: https://issues.apache.org/jira/browse/SPARK-12367
>             Project: Spark
>          Issue Type: Bug
>          Components: ML
>    Affects Versions: 1.5.2
>            Reporter: Eugene Morozov
>         Attachments: CodeThatGivesANoSuchElementException.java, complete-stack-trace.log, input.gz
>
>
> I'm consistently getting "java.util.NoSuchElementException: key not found: 1.0" while trying to do a prediction on a trained model.
> I use ml package - Pipeline API. The model is successfully trained, I see some stats in the output: total, findSplitsBins, findBestSplits, chooseSplits. I can even serialize it into a file and use afterwards, but the prediction is broken somehow.
> Code, input data and stack trace attached.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org