[
https://issues.apache.org/jira/browse/SPARK-4894?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14277702#comment-14277702
]
Joseph K. Bradley commented on SPARK-4894:
------------------------------------------
[~rnowling] Thanks for looking into this issue! I was thinking about 2
possibilities for generalizing NaiveBayes:
* Specify the model type with simple strings, and keep current API
** This is simpler and maintains API stability.
* Generalize the model to allow other feature and label types
** This is what we really should do long-term since NaiveBayes should not be
limited to discrete labels.
** This would require more work, including defining a Factor concept (using the
terminology from Probabilistic Graphical Models) and updating the NaiveBayes
API to use factors. Different factors would handle discrete and/or continuous
variables, and would encode different types of distributions. This setup is
common in graphical model libraries like [Factorie|
http://factorie.cs.umass.edu/index.html].
** Alternative: We could separate NaiveBayes into 2 classes based on discrete
and continuous labels, but that might be even more work in the long term (to
maintain 2 copies of the API).
What do you think?
> Add Bernoulli-variant of Naive Bayes
> ------------------------------------
>
> Key: SPARK-4894
> URL: https://issues.apache.org/jira/browse/SPARK-4894
> Project: Spark
> Issue Type: New Feature
> Components: MLlib
> Affects Versions: 1.2.0
> Reporter: RJ Nowling
> Assignee: RJ Nowling
>
> MLlib only supports the multinomial-variant of Naive Bayes. The Bernoulli
> version of Naive Bayes is more useful for situations where the features are
> binary values.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
