Login  Register

Re: Fit.rSquared

Posted by Michael Schmid on Nov 02, 2012; 9:40am
URL: http://imagej.273.s1.nabble.com/Fit-rSquared-tp5000629p5000632.html

Hi Danju,

it is a division, R^2 = 1 - SSE (divided by) SSD,
where:  SSE = sum of the squared errors (residuals) and
        SSD = sum of the squared deviations about the mean (related to the variance of data)

> Why "1 minus"?

If the errors are zero, SSE is zero, and R^2 becomes 1, this is what one calls perfect correlation.

See also
  http://en.wikipedia.org/wiki/Coefficient_of_determination
(we use the last equation in the paragraph 'Definitions')

Michael
________________________________________________________________
On Nov 2, 2012, at 01:10, Danju Men wrote:

> Hi all,
>
>
>
> Does anyone know what exactly does "Fit.rSquared" return? Is it:
>
> 1 - SSE (divided by) SSD;
>
> Or:
>
> 1 - SSE (or) SSD?
>
> Why "1 minus"?
>
>
>
> Thanks,
>
> Danju
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html