[
https://issues.apache.org/jira/browse/SPARK-14076?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15210226#comment-15210226
]
Yanbo Liang edited comment on SPARK-14076 at 3/24/16 1:41 PM:
--------------------------------------------------------------
[~mengxr] Because NaiveBayesModel extends from
ProbabilisticClassificationModel, I think we should modify the "transform"
function of ProbabilisticClassificationModel to add the label column's
attributes into the output prediction column. Other ClassificationModels who
also extends from ProbabilisticClassificationModel can benefit from this
improvement.
If I understand it correctly, I can work on this.
was (Author: yanboliang):
[~mengxr] Because NaiveBayesModel extends from ProbabilisticClassificationModel
and the output prediction column was produced there, so I think we should
modify the transform function of ProbabilisticClassificationModel to add the
label column's attributes in the output prediction column. Other
ClassificationModels who also extends from ProbabilisticClassificationModel can
benefit from this improvement.
If I understand it correctly, I can work on this.
> Naive Bayes should output attributes in predictions
> ---------------------------------------------------
>
> Key: SPARK-14076
> URL: https://issues.apache.org/jira/browse/SPARK-14076
> Project: Spark
> Issue Type: Improvement
> Components: ML
> Affects Versions: 2.0.0
> Reporter: Xiangrui Meng
> Labels: naive-bayes
>
> In naive Bayes, we outputs predicted labels without feature attributes. So
> from the metadata, we can only tell that it is a nominal feature but the
> values are missing. We should copy the attributes from input label column to
> output prediction column.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
