[
https://issues.apache.org/jira/browse/MAHOUT-1493?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14064027#comment-14064027
]
Andrew Palumbo edited comment on MAHOUT-1493 at 7/16/14 8:37 PM:
-----------------------------------------------------------------
[~cviebig], The patch looks good. I've made some edits (against your
develop/issue/1493/4 branch) and will attach a M-1493a.patch shortly. I put
the trainComplementary parameter back in as this is needed to make the
distinction between Standard and Complementary Models in the NaiveBayesModel
constructor.
As well, I've added a thetaNormalizer var which can remain null when passed to
the NaiveBayesModel constructor unless training a Complementary NB model. see:
https://github.com/apache/mahout/blob/master/mrlegacy/src/main/java/org/apache/mahout/classifier/naivebayes/NaiveBayesModel.java
I'm not sure if creating a null var as I've done here is best practice in
scala, but i wanted to give you an idea of the NaiveBayesModel design.
As you've noted, there has been a lot of refactoring going on. As far as
moving the code, I think that for now it might be a good idea to keep this in
`spark` module, and move the
`org.apache.mahout.sparkbindings.drm.classification` package out of
`org.apache.mahout.sparkbindings.drm` and into a new
`org.apache.mahout.classification` package. I believe that for now this would
be a good place for it. There shouldn't be any need to move any of the java
code from mrlegacy.
was (Author: andrew_palumbo):
[~cviebig], The patch looks good. I've made some edits (against your
develop/issue/1493/4 branch) and will attach a MAHOUT-1493a.patch shortly. I
put the trainComplementary parameter back in as this is needed to make the
distinction between Standard and Complementary Models in the NaiveBayesModel
constructor.
As well, I've added a thetaNormalizer var which can remain null when passed to
the NaiveBayesModel constructor unless training a Complementary NB model. see:
https://github.com/apache/mahout/blob/master/mrlegacy/src/main/java/org/apache/mahout/classifier/naivebayes/NaiveBayesModel.java
I'm not sure if creating a null var as I've done here is best practice in
scala, but i wanted to give you an idea of the NaiveBayesModel design.
As you've noted, there has been a lot of refactoring going on. As far as
moving the code, I think that for now it might be a good idea to keep this in
`spark` module, and move the
`org.apache.mahout.sparkbindings.drm.classification` package out of
`org.apache.mahout.sparkbindings.drm` and into a new
`org.apache.mahout.classification` package. I believe that for now this would
be a good place for it. There shouldn't be any need to move any of the java
code from mrlegacy.
> Port Naive Bayes to the Spark DSL
> ---------------------------------
>
> Key: MAHOUT-1493
> URL: https://issues.apache.org/jira/browse/MAHOUT-1493
> Project: Mahout
> Issue Type: Bug
> Components: Classification
> Reporter: Sebastian Schelter
> Assignee: Sebastian Schelter
> Fix For: 1.0
>
> Attachments: MAHOUT-1493.patch, MAHOUT-1493.patch, MAHOUT-1493.patch,
> MAHOUT-1493.patch, MAHOUT-1493a.patch
>
>
> Port our Naive Bayes implementation to the new spark dsl. Shouldn't require
> more than a few lines of code.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
