On Tuesday, Andy asked about ridge regression:
> Ridge regression, for use in models where there is known but
> unavoidable collinearity, all I can find is something from STB28! Is
> there anything more recent around please or were the STB28 routines the
> last word?
>
Andy probably did
findit ridge
which finds an ado file which does not work in version 7 or later. This
problem came to the author's (Bob Obenchain) attention recently and has
been fixed. Bob is not a Statalister, but here is the response I
received from Bob concerning Andy's question:
<snip>
The new Stata version 7+ material is in a file named RX_STATA.ZIP that
can be downloaded from http://members.iquest.net/~softrx/
As to what's the latest word on how to quantify "success" in a
particular shrinkage regression application, my biased opinion is that
my stuff remains at the head-of-the-pecking-order. For example, my
Stata ADO functions estimate the likelihood of achieving optimal
squared-error-risk-reduction in three different ways: classical
normal-distrbution-theory, empirical bayes, and random coefficient
formulations. And other concepts implemented (such as relative or
"scaled" mean squared error risk estimation, estimation of the "inferior
direction," etc.) come directly from concepts introduced in my landmark
Annals of Statistics paper, 1978.
A wealth of relevant ridge/shrinkage information is available for
viewing/download from my personal web site. ...BobO.
<snip>
yours,
--May
*
* 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/
