You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by "Lance Norskog (JIRA)" <ji...@apache.org> on 2011/03/19 00:51:29 UTC

[jira] Commented: (MAHOUT-628) Add an option to prune away users with less than a given number of preferences to ItemSimilarityJob and RecommenderJob

    [ https://issues.apache.org/jira/browse/MAHOUT-628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13008683#comment-13008683 ] 

Lance Norskog commented on MAHOUT-628:
--------------------------------------

Could this be done with a DataModel wrapper instead?

> Add an option to prune away users with less than a given number of preferences to ItemSimilarityJob and RecommenderJob
> ----------------------------------------------------------------------------------------------------------------------
>
>                 Key: MAHOUT-628
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-628
>             Project: Mahout
>          Issue Type: New Feature
>          Components: Collaborative Filtering
>    Affects Versions: 0.5
>            Reporter: Sebastian Schelter
>            Assignee: Sebastian Schelter
>         Attachments: MAHOUT-628.patch
>
>
> Some real-world datasets (especially those created from implicit feedback) might include users with only a tiny number of preferences (like one-time-visitors only viewing a single item) that a users of ItemSimilarityJob or RecommenderJob might want to prune away. I added a new parameter "minPrefsPerUser" that makes those jobs throw out users with less than a given number of preferences. It is per default set to 1 so that the input data stays untouched.
> It's just a small patch to make those jobs more usable in real-world scenarios.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira