commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mark Devaney (JIRA)" <j...@apache.org>
Subject [jira] Commented: (MATH-392) calculateYVariance in OLS/GLSMultipleLinearRegression uses residuals not Y vars
Date Mon, 26 Jul 2010 16:32:51 GMT

    [ https://issues.apache.org/jira/browse/MATH-392?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12892352#action_12892352
] 

Mark Devaney commented on MATH-392:
-----------------------------------

Can't test a patch as I'm not able to build current repository version:
math/src/test/java/org/apache/commons/math/optimization/univariate/BrentOptimizerTest.java:[28,39]
cannot find symbol
symbol  : class SincFunction

Implementation for both GLS/OLS:

protected double calculateYVariance() {
    return new Variance().evaluate(Y);
}


> calculateYVariance in OLS/GLSMultipleLinearRegression uses residuals not Y vars
> -------------------------------------------------------------------------------
>
>                 Key: MATH-392
>                 URL: https://issues.apache.org/jira/browse/MATH-392
>             Project: Commons Math
>          Issue Type: Bug
>    Affects Versions: 2.1
>            Reporter: Mark Devaney
>            Priority: Minor
>             Fix For: 2.2
>
>
> Implementation of OLS/GLSMultipleLinearRegression is:
> @Override
> 173        protected double calculateYVariance() {
> 174            RealVector residuals = calculateResiduals();
> 175            return residuals.dotProduct(residuals) /
> 176                   (X.getRowDimension() - X.getColumnDimension());
> 177        }
> This gives variance of residuals not variance of the dependent (Y) variable as the documentation
suggests.

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


Mime
View raw message