commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Phil Steitz (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (MATH-1133) NaNStrategy for UnivariateStatistic and friends
Date Fri, 27 Jun 2014 14:15:29 GMT

    [ https://issues.apache.org/jira/browse/MATH-1133?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14045983#comment-14045983
] 

Phil Steitz commented on MATH-1133:
-----------------------------------

We should discuss on the mailing list.  My inclination is to leave things as they are.   NaNStrategy
exists because for rank transformations we *have* to do something special for NaNs.  In floating
point computations, we have traditionally just allowed NaNs to propagate. 

> NaNStrategy for UnivariateStatistic and friends
> -----------------------------------------------
>
>                 Key: MATH-1133
>                 URL: https://issues.apache.org/jira/browse/MATH-1133
>             Project: Commons Math
>          Issue Type: Improvement
>            Reporter: Sergei Lebedev
>            Priority: Minor
>
> In the current implementation univariate statistics don't allow to specify NaN behaviour.
For instance, the following code simply yields NaN as the result.
> {code:java}
> Mean mean = new Mean();
> mean.increment(42);
> mean.increment(Double.NaN);
> System.out.println(mean.getResult());  // prints 'NaN'.
> {code}
> Is it possible to parametrize all available statistics by NaNStrategy with the default
behaviour being a no-op?



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message