[R] Problem with predict in systemfit

Leeds, Mark (IED) Mark.Leeds at morganstanley.com
Sat Apr 14 00:03:22 CEST 2007


A friend of mine sent me below so I am posting below. If it is not
enough information, please just
let me know and I will tell him. Thanks.


-----Original Message-----
Sent: Friday, April 13, 2007 3:58 PM
To: Leeds, Mark (IED)
Subject: R question

I am using the "predict" function after I have done a simultaneous
estimation of a system using "systemfit". fitsur is the output of
systemfit and "system" is my system of equations, "SUR" is the Seemingly
Unrelated Regrssions method of estimation.

fitsur<-systemfit("SUR",system)

I am supposed to be able to use the estimated coefficients for
out-of-sample prediction using a different data set than the one
originally used for estimation. That is my XSf data frame. Unfortunately
whether I include that option(XSf) as in p or not as in q I get the same
result from predict in this case. I have confirmed that "predict" works
properly for the output of "lm" 

p<-predict.systemfit(fitsur,XSf)
q<-predict.systemfit(fitsur)
--------------------------------------------------------

This is not an offer (or solicitation of an offer) to buy/se...{{dropped}}



More information about the R-help mailing list