[
https://issues.apache.org/jira/browse/MAHOUT-1072?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13461460#comment-13461460
]
Jeff Eastman commented on MAHOUT-1072:
--------------------------------------
Seems to me the outlier threshold and emitMostLikely do different things:
- Outlier threshold emits only points with pdf>threshold. This is relevant for
all cluster types.
- EmitMostLikely emits only the point with max(pdf) and is only relevant for
Dirichlet & FuzzyK that emit all points subject to the threshold above.
I'm not convinced removing EmitMostLikely makes sense but am not hung up and
would like to hear more discussion from our user base.
> FuzzyKmeansDriver emitMostLikely has no effect
> ----------------------------------------------
>
> Key: MAHOUT-1072
> URL: https://issues.apache.org/jira/browse/MAHOUT-1072
> Project: Mahout
> Issue Type: Bug
> Affects Versions: 0.8
> Reporter: Dave Byrne
> Assignee: Paritosh Ranjan
> Attachments: fkmeans_emit.patch
>
>
> FuzzyKmeansDriver.clusterData() has emitMostLikely hardcoded to true and
> ignores argument passed to function
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
