On Wed, 20 Aug 2008, Samuel Field wrote:
List,
Is it possible to decompose the variance of an outcome into trend,
signal and noise components using a SAR model - analogous to what one
would get with an OLS model? This doesn't seem to be straight forward.
With OLS, we decompose Y into two non
List,
Is it possible to decompose the variance of an outcome into trend, signal and
noise components using a SAR model - analogous to what one would get with an
OLS model? This doesn't seem to be straight forward. With OLS, we decompose Y
into two non-overlapping components
Y = fit + error
R-