Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: RE: Bayesian model averaging


From   Ian Watson <adslftis@tpg.com.au>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: Bayesian model averaging
Date   Fri, 02 Jun 2006 12:09:02 -0700

Dear John

Thanks for the link. It's very useful.

I note from the bic help that "bic estimates models based on all possible"
combinations of the independent variables - it is computationally
intensive."

I think the R and S Plus versions of BMA make use of the leaps algorithm to
reduce the number of possible models at the outset, thereby avoiding
some of this computationally intensive work.

I will try bic and see how it goes. I could always leave it running
overnight.

(An interesting note on naming: Scott Long also wrote a routine (for
Stata 5) which
was called bic.ado.)

Ian


Moran, John (NWAHS) wrote:
Dear Ian

As a start, see Paul Millar's site :
http://www.ucalgary.ca/~pemillar/stata.htm




John moran
E-mail: john.moran@nwahs.sa.gov.au


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Ian Watson
Sent: Friday, 2 June 2006 9:13 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: Bayesian model averaging

I have recently been reading some of the Bayesian social science
literature, particularly work by Adrian Raftery, Jennifer Hoeting and
Chris Volinsky that deals with Bayesian model averaging. It seems a very
interesting avenue to pursue, but the BMA routines developed by these
authors are mostly available in S Plus (with translations into R by Ian
Painter). Does anyone know if there are any Stata routines which
implement the BMA approach?


If this sounds a bit cryptic, the description for one of the routines
(bicreg) from the R manual is as follows:

     Bayesian Model Averaging accounts for the model uncertainty
inherent in the variable selection
     problem by averaging over the best models in the model class
according to approximate posterior
     model probability.


If nothing exists in Stata, and I need to write a Stata routine myself,
do people know if there is any useful "R to Stata" documentation around
on the Net?  Apparently, the leaps algorithm used by BMA is originally
based on Fortran code, so presumably Mata can help out there.

Any advice would be most appreciated.

Ian Watson

Sydney University
Australia


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/




*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index