[19388] in s-news-athena
[S] Error in predict.gam()
daemon@ATHENA.MIT.EDU (Cande V. Ananth)
Mon Oct 4 10:00:22 1999
Date: Mon, 4 Oct 1999 09:49:25 -0400 (EDT)
From: "Cande V. Ananth" <ananthcv@epi.umdnj.edu>
Message-Id: <199910041349.JAA07935@epi.umdnj.edu>
To: s-news@wubios.wustl.edu
Splus'ers:
I'm trying to predict an interaction-effect from the gam() model
gam.int <- gam(y ~ lo(x1,x2,span=0.4,degree=2), binomial)
but get the following error:
>> Error in .Fortran("lowese",: extrapolation not allowed with blending
>> Dumped
>> Error was while calling subroutine "lowese"
I don't think that I'm forcing the predict() function to extrapolate
the data. The gam() model was fit with approximately 18,000 obs, and
the data is not sparce for combinations of x1 and x2. I'd greatly
appreciate help in rectifying the problem.
I'm sunning S-Plus 3.4 on the UNIX. Thanks.
Cande Ananth
Dept of OB/GYN, UMDNJ
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
