RMSE vs. standard error
Date: May 29, 2011 10:14PM
RMSE is the square root of the mean squared error.
Standard Error of Estimate (SEE) = square root of sum of squares divided by n-k-1
So does RMSE= SEE?
Date: May 30, 2011 01:25AM
Way to confuse. Throw in a quant question, and stare at the blank faces of candidates.
By the way i'd think the answer to your question is NO.
SEE = std deviation of error terms.
SEE = sqrt(variance of error)
SEE = sqrt(SSE/n-k-1)
where as MSE = SSE/ n-k-1 <-- there is no square root here.
SSE = squared sum of all errors, or residual sum of errors.
SSE/n-k-1 is not equal to SEE.
By the way what is RMSE? seeing it for the first time.
Date: May 30, 2011 01:59AM
they are not the same thing, but closely related. RMSE is for the MEAN, not the total errors. it is the average error.
Date: May 30, 2011 09:03AM
RMSE is sqrt(MSE). Same thing as far as I can tell.
It's a tool used to gauge in-sample and out-fo-sample forecasting accuracy. Low RMSE relative to another model = better forecasting.
Date: May 30, 2011 09:30AM
As is with SEE
Date: May 30, 2011 09:50AM
So it boils down to whether MSE = Sum of squares / n, or MSE = sum of squares / n-k-1. On an Anove table you will find MSS and the associated degrees of freedom is n-k-1.
I think denominator for MSE = n, denominator in the SEE is n-k-1 and that's my story.