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 2011/03/25 09:56:05 UTC

[jira] [Created] (MAHOUT-635) RMS variant of RunningAverage classes

RMS variant of RunningAverage classes
-------------------------------------

                 Key: MAHOUT-635
                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
             Project: Mahout
          Issue Type: Improvement
          Components: Math
    Affects Versions: 0.5
            Reporter: Lance Norskog
            Priority: Minor


Added as classes because I don't know if this is interesting.

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

[jira] [Commented] (MAHOUT-635) RMS variant of RunningAverage classes

Posted by "Lance Norskog (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MAHOUT-635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011557#comment-13011557 ] 

Lance Norskog commented on MAHOUT-635:
--------------------------------------

This lets the RecommenderEvaluators calculate the RMS via a RunningAverage object in _that other_ patch, thus filling a missing bit of feature.

> RMS variant of RunningAverage classes
> -------------------------------------
>
>                 Key: MAHOUT-635
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.5
>            Reporter: Lance Norskog
>            Assignee: Sean Owen
>            Priority: Minor
>             Fix For: 0.5
>
>         Attachments: RMSRunningAverage.java, RMSRunningAverageAndStdDev.java, RMSRunningAverageTest.java
>
>
> Added as classes because I don't know if this is interesting.

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

[jira] [Resolved] (MAHOUT-635) RMS variant of RunningAverage classes

Posted by "Sean Owen (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MAHOUT-635?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sean Owen resolved MAHOUT-635.
------------------------------

       Resolution: Won't Fix
    Fix Version/s: 0.5
         Assignee: Sean Owen

(This status change is by no means final. I'm just keen to make the JIRA status always reflect the likely current thinking. More discussion and/or a new patch and we'd possibly go for something like this.)

> RMS variant of RunningAverage classes
> -------------------------------------
>
>                 Key: MAHOUT-635
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.5
>            Reporter: Lance Norskog
>            Assignee: Sean Owen
>            Priority: Minor
>             Fix For: 0.5
>
>         Attachments: RMSRunningAverage.java, RMSRunningAverageAndStdDev.java, RMSRunningAverageTest.java
>
>
> Added as classes because I don't know if this is interesting.

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

[jira] [Commented] (MAHOUT-635) RMS variant of RunningAverage classes

Posted by "Sean Owen (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MAHOUT-635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011128#comment-13011128 ] 

Sean Owen commented on MAHOUT-635:
----------------------------------

This isn't implementing RMS though. It's computing the average of a series, and then taking the square root. If that series you're feeding are the squared differences of estimates and actual values, then yes you've got RMS. But then this class adds nothing but a call to Math.sqrt(). I think we're back to where we left that thread, yes.

Are you intending this for commit or just as food for thought?

> RMS variant of RunningAverage classes
> -------------------------------------
>
>                 Key: MAHOUT-635
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.5
>            Reporter: Lance Norskog
>            Priority: Minor
>         Attachments: RMSRunningAverage.java, RMSRunningAverageAndStdDev.java, RMSRunningAverageTest.java
>
>
> Added as classes because I don't know if this is interesting.

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

[jira] [Commented] (MAHOUT-635) RMS variant of RunningAverage classes

Posted by "Lance Norskog (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MAHOUT-635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011125#comment-13011125 ] 

Lance Norskog commented on MAHOUT-635:
--------------------------------------

Should RMS allow changing or removing a datum?

CompactRunningAverage not implemented for reasons discussed separately.

> RMS variant of RunningAverage classes
> -------------------------------------
>
>                 Key: MAHOUT-635
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.5
>            Reporter: Lance Norskog
>            Priority: Minor
>         Attachments: RMSRunningAverage.java, RMSRunningAverageAndStdDev.java, RMSRunningAverageTest.java
>
>
> Added as classes because I don't know if this is interesting.

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

[jira] [Commented] (MAHOUT-635) RMS variant of RunningAverage classes

Posted by "Lance Norskog (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MAHOUT-635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011553#comment-13011553 ] 

Lance Norskog commented on MAHOUT-635:
--------------------------------------

bq. This isn't implementing RMS though. It's computing the average of a series, and then taking the square root.

FullRunningAverage: *datum*
{code}
  public synchronized void addDatum(double datum) {
    if (++count == 1) {
      average = datum;
    } else {
      average = average * (count - 1) / count + datum / count;
    }
  }
{code}

RMSRunningAverage: *(datum * datum)*

{code}
   if (++count == 1) {
      average = datum * datum;
    } else {
      average = average * (count - 1) / count + (datum * datum) / count;
    }
{code}


> RMS variant of RunningAverage classes
> -------------------------------------
>
>                 Key: MAHOUT-635
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.5
>            Reporter: Lance Norskog
>            Assignee: Sean Owen
>            Priority: Minor
>             Fix For: 0.5
>
>         Attachments: RMSRunningAverage.java, RMSRunningAverageAndStdDev.java, RMSRunningAverageTest.java
>
>
> Added as classes because I don't know if this is interesting.

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

[jira] [Updated] (MAHOUT-635) RMS variant of RunningAverage classes

Posted by "Lance Norskog (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MAHOUT-635?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Lance Norskog updated MAHOUT-635:
---------------------------------

    Attachment: RMSRunningAverageTest.java
                RMSRunningAverageAndStdDev.java
                RMSRunningAverage.java

> RMS variant of RunningAverage classes
> -------------------------------------
>
>                 Key: MAHOUT-635
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.5
>            Reporter: Lance Norskog
>            Priority: Minor
>         Attachments: RMSRunningAverage.java, RMSRunningAverageAndStdDev.java, RMSRunningAverageTest.java
>
>
> Added as classes because I don't know if this is interesting.

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

[jira] [Commented] (MAHOUT-635) RMS variant of RunningAverage classes

Posted by "Sean Owen (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MAHOUT-635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011592#comment-13011592 ] 

Sean Owen commented on MAHOUT-635:
----------------------------------

OK, I see that much, there's a root mean and square. The caller is still computing the "error" bit. That's tiny, though turned around, it makes me wonder whether this is worth another class?

> RMS variant of RunningAverage classes
> -------------------------------------
>
>                 Key: MAHOUT-635
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-635
>             Project: Mahout
>          Issue Type: Improvement
>          Components: Math
>    Affects Versions: 0.5
>            Reporter: Lance Norskog
>            Assignee: Sean Owen
>            Priority: Minor
>             Fix For: 0.5
>
>         Attachments: RMSRunningAverage.java, RMSRunningAverageAndStdDev.java, RMSRunningAverageTest.java
>
>
> Added as classes because I don't know if this is interesting.

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