I am attempting to do some GAM modelling of time series of water quality
data. The problem is that the residuals are autocorrelated, and this
needs to be fixed by fitting a lag-1 autoregressive model to the errors.
Is there an easy way of doing this within the GAM procedure (similar to
using "corr=corAR1(x)" in GLS? Or is there some other work-around that
has been developed?
I am using S-Plus 6.2 for Windows.
Any help would be much appreciated.
Angus Webb
--
Dr J. Angus Webb
Research Fellow
Water Studies Centre
CRC for Freshwater Ecology & Department of Chemistry
Monash University
Clayton Campus, 3800
Vic, Australia
ph: +61 3 9905 4198
fax: +61 3 9905 4196
email: angus.webb@sci.monash.edu.au
|