You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@commons.apache.org by "Phil Steitz (JIRA)" <ji...@apache.org> on 2015/07/06 02:46:04 UTC

[jira] [Created] (MATH-1247) Kolmogorov-Smirnov exactP gives incorrect p-values for some D-statistics

Phil Steitz created MATH-1247:
---------------------------------

             Summary: Kolmogorov-Smirnov exactP gives incorrect p-values for some D-statistics
                 Key: MATH-1247
                 URL: https://issues.apache.org/jira/browse/MATH-1247
             Project: Commons Math
          Issue Type: Bug
    Affects Versions: 3.4.1
            Reporter: Phil Steitz
            Priority: Minor
             Fix For: 4.0, 3.5


The exactP method in KolmogorovSmirnovTest, which is used by default for small samples in 2-sample tests, can give slightly incorrect p-values in some cases.  The reason for this is that p(D > d) is computed by examining all m-n partitions and counting the number of partitions that give D values larger than the observed value.  D values are not rounded, so some values that are mathematically identical to the observed value compare less than or greater than it.  This results in small errors in the reported p-values.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)