Hello All, This is my third time attempting to post this message. I don't see it in the archive, so I'm guessing it is not getting through. If I am wrong, my apologies. I am trying to do a GLS regression, (X'V^-1X)^-1X'V^-1y, using the gls() function from the nlme package. I have the covariance matrix V. I have been searching for a way to specify the correlation structure using V, and have had little luck. I have found a few other messages from people who have had the same problem. Is there a way to do this? If it requires building my own corStruct, where can I look for information to do this? Thanks, John. -- ========================John Janmaat, Assistant Professor of Economics IK Barber School of Arts and Sciences (Unit 6) University of British Columbia Okanagan 3333 University Way, Kelowna, BC, V1V 1V7 Canada
Janmaat, John wrote:> > Hello All, > > This is my third time attempting to post this message. I don't see it > in the archive, so I'm guessing it is not getting through. If I am > wrong, my apologies. > > I am trying to do a GLS regression, (X'V^-1X)^-1X'V^-1y, using the gls() > function from the nlme package. I have the covariance matrix V. I have > been searching for a way to specify the correlation structure using V, > and have had little luck. I have found a few other messages from people > who have had the same problem. Is there a way to do this? If it > requires building my own corStruct, where can I look for information to > do this? > > Thanks, > > John. > > -- >Don't know if it came through or not before, and not sure of the answer, but just to give you _an_ answer -- if the answer isn't in Pinheiro and Bates (2000), then you'll probably have to dig through the code (which is scary). Ben Bolker -- View this message in context: http://www.nabble.com/GLS-with-nlme-tf4737987.html#a13558299 Sent from the R help mailing list archive at Nabble.com.