[R--gR] FW: mixed graphical models

Steffen Lauritzen steffen at stats.ox.ac.uk
Wed Sep 14 18:17:34 CEST 2005


Can anyone help with this?
Steffen

-----Original Message-----
From: jw2256 at columbia.edu [mailto:jw2256 at columbia.edu] 
Sent: 14 September 2005 10:48
To: Steffen Lauritzen
Subject: mixed graphical models

Dear Prof. Steffen Lauritzen,
Recently, I tested mixed graphical model in gR
and find something unclear for me when
using the hierarichical interaction model and
hope you may give me some advice.

I try to fit a model ABCG/BCGV,ABCU,ACGZ,ABCGY,ABCGX, ABCGW/UWXYZ, VWXYZ

by
using hierarchical interaction model but the
algorithm seems fitting below models
VWXYZ,UWXYZ,CGVWXYZ,BCGVWXYZ,ACUWXYZ,ABCUWXY,ACGWXYZ,ABCGWXY
during the each cycle..

Do you know why the models calculated by fitting algorithm
are different from the fitting model ?
How to generate those models from the fitting model
ABCG/BCGV,ABCU,ACGZ,ABCGY,ABCGX, ABCGW/UWXYZ,VWXYZ
before the calculation?

I really appreciated for your kindly reply!

yours sincerely


junbai wang

Biological Science Department
Columbia university
New York
USa




More information about the R-sig-gR mailing list