[jira] Updated: (MAHOUT-444) Need on-line distribution summary statistics ... mean, median, min, max q25, q75

["Ted Dunning (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/MAHOUT-444?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ted Dunning updated MAHOUT-444:
-------------------------------

    Status: Patch Available  (was: Open)

Here is a patch with an object that estimates the five quartiles, mean and standard deviation.  The associated tests indicate that it is pretty much as accurate as if all of the samples were kept and the empirical rank statistics were computed directly.

> Need on-line distribution summary statistics ... mean, median, min, max q25, q75
> --------------------------------------------------------------------------------
>
>                 Key: MAHOUT-444
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-444
>             Project: Mahout
>          Issue Type: Bug
>            Reporter: Ted Dunning
>
> For the on-line learning algorithms it is very helpful to have robust on-line estimate of various summary statistics.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.