rmse vs see

[mikecocos]

are the root mean squared error and the standard error of the etimate both =
(SSE/n-k-1)^(1/2) ?

 

[cpk123]

in a sense - yes they are
but the major difference - RMSE pertains to sample that is OUT OF PERIOD. It is based on Actual vs. Predicted for a future period sample outside of original regression.
SEE - pertains to in period sample.

 

[mikecocos]

thanks.

 

[AudreyMwala]

SEE = 1/ square root of n (number of observations)

 

[cpk123]

audrey
that is not correct. This is the std error of the autocorrelations on Time series only.
SEE = sqrt(SSE/(n-k-1))
on multiple and single variable regressions..

 

[AudreyMwala]

ok, thanks. i thought it the same

 

[coshair]

CP
for SEE = 1/ square root of n (number of observations)
it’s also the t-stat for the correclation coefficient for time series, right?