[
https://issues.apache.org/jira/browse/SPARK-8849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14615988#comment-14615988
]
DB Tsai commented on SPARK-8849:
--------------------------------
I hope that we can port MLOR to ML pipeline framework in 1.5 so we don't have
to keep adding new feature into old codebase.
> expose multiclass scores in LogisticRegression
> ----------------------------------------------
>
> Key: SPARK-8849
> URL: https://issues.apache.org/jira/browse/SPARK-8849
> Project: Spark
> Issue Type: New Feature
> Components: MLlib
> Reporter: Asher Krim
> Priority: Minor
> Labels: LogisticRegression, mllib, multiclass
>
> In [SPARK-2309], [~dbtsai] implemented multinomial logistic regression for
> mllib. It would be great if users could get the score for each class
> directly, since they are already computed. It doesn't look like this would be
> a terribly complicated thing to do and I've sketched out what it could look
> like: https://gist.github.com/Krimit/f61b5e3be2b1e380d1ca
> We could even move the score methods further up into GeneralizedLinearModel,
> since in the binary case we could still just return a 2-element vector of the
> scores (or margin in SVM's) for each of the 2 class, so algorithms lacking
> multiclass support at the moment can still support this.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
