spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yanbo Liang (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (SPARK-6080) LogisticRegressionWithLBFGS in PySpark was assigned wrong "regType" parameter
Date Sat, 28 Feb 2015 09:16:04 GMT

    [ https://issues.apache.org/jira/browse/SPARK-6080?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14341426#comment-14341426
] 

Yanbo Liang edited comment on SPARK-6080 at 2/28/15 9:15 AM:
-------------------------------------------------------------

This bug is easy to reproduce. In a PySpark environment, when you call
    model = LogisticRegressionWithSGD.train(data, regType=None) 
it will return a LogisticRegressionModel which was trained with no regularization.

But when you run 
    model = LogisticRegressionWithLBFGS.train(data, regType=None)  
it will throw an exception
    java.lang.IllegalArgumentException: Invalid value for 'regType' parameter. Can only be
initialized using the following string values: ['l1', 'l2', None].

This is due to when invoke callMLlibFunc of LogisticRegressionWithLBFGS at python/pyspark/mllib/classification.py,
the parameter was assigned to "str(regType)" which translate None(Python) to "None"(Java/Scala).
The right way should be translate None(Python) to null(Java/Scala).
We need to do the same thing as LogisticRegressionWithSGD. 


was (Author: yanboliang):
This bug is easy to reproduce. In a PySpark environment, when you call
    model = LogisticRegressionWithSGD.train(data, regType=None) 
it will return a LogisticRegressionModel which was trained with no regularization.

But when you run 
    model = LogisticRegressionWithLBFGS.train(data, regType=None)  
it will throw an exception
    java.lang.IllegalArgumentException: Invalid value for 'regType' parameter. Can only be
initialized using the following string values: ['l1', 'l2', None].

This is due to when invoke callMLlibFunc at python/pyspark/mllib/classification.py, the parameter
was assigned to "str(regType)" which translate None(Python) to "None"(Java/Scala). The right
way should be translate None(Python) to null(Java/Scala).
We need to do the same thing as LogisticRegressionWithSGD. 

> LogisticRegressionWithLBFGS in PySpark was assigned wrong "regType" parameter
> -----------------------------------------------------------------------------
>
>                 Key: SPARK-6080
>                 URL: https://issues.apache.org/jira/browse/SPARK-6080
>             Project: Spark
>          Issue Type: Bug
>          Components: MLlib, PySpark
>            Reporter: Yanbo Liang
>
> Currently LogisticRegressionWithLBFGS in python/pyspark/mllib/classification.py will
invoke callMLlibFunc with a wrong "regType" parameter.



--
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


Mime
View raw message