[
https://issues.apache.org/jira/browse/SPARK-15741?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15313376#comment-15313376
]
Bryan Cutler commented on SPARK-15741:
--------------------------------------
>From what I gathered, explicitly setting a seed to {{None}} in python will
>translate to Scala with a value of {{0L}}, and the default value set in the
>{{HasSeed}} shared param for Python and Scala is a non-random hash value based
>off the class name. If you want a random seed, you need to explicitly set it.
> PySpark Cleanup of _setDefault with seed=None
> ---------------------------------------------
>
> Key: SPARK-15741
> URL: https://issues.apache.org/jira/browse/SPARK-15741
> Project: Spark
> Issue Type: Improvement
> Components: ML, PySpark
> Reporter: Bryan Cutler
> Priority: Minor
>
> Calling Params._setDefault with a param equal to {{None}} will be ignored
> internally silently. There are several cases where this is done with the
> {{seed}} param, making it seem like it might do something. These cases
> should be removed for the sake of consistency.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
