RMSE to Error Variance convertion

Hey,

Is there a function that would convert the Root Mean Square Error to an Error Variance? I need to get the error value in terms of the R^2 error.

[Ferd]

or could anyone tell me the convertion factor(s).

Answer 1

[the cyclist]

This page has the conversion formula:

http://en.wikipedia.org/wiki/Coefficient_of_determination#Definitions

The diagram with the red and blue squares is helpful to understand the relationship.

SS_err is proportional to RMSE squared and SS_tot is proportional to "RMSE_0" squared, i.e. the squared error before any model has been applied.

So you can think of it as

R^2 = 1 - (RMSE^2)/(RMSE_0^2)

I can probably whip up a little sample code if that's not clear enough.

[Ferd]

Will check. If I have further questions, will reply here.

If not, a big thankyou.

Appreciate it The Cyclist!