Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: How to retrieve parameter estimates from an ML program, so they can be used in estimation of expected values, variances, etc...


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: How to retrieve parameter estimates from an ML program, so they can be used in estimation of expected values, variances, etc...
Date   Mon, 29 Mar 2010 13:52:48 +0100

I am surprised you need to do this at all, unless your version of
generalised gamma is incompatible with Stata's (which is possible, just
as there are various different generalised exponentials). 

With a token -stset-, even if it is insincere, you can use Stata's
version of this in -streg- without needing to reinvent it. 

Nick 
n.j.cox@durham.ac.uk 

SLG Brilleman

I have recently written a program for ml estimation using the
generalized 
Gamma distribution. The estimation routine seems to run fine and will 
produce estimates of the coefficients for the linear predictor (mu), and

estimates of the two shape parameters (sigma and kappa).

My problem is that I want to estimate the expected value of yi (i.e. the

individual predictions) - an equation involving estimates of all three 
parameters (mu, sigma, kappa). I am reasonably new to STATA and cannot 
figure out the syntax to retrieve and store the estimates of mu, sigma,
and 
kappa, so that I can generate the new variable yi.

I have tried 'ereturn', 'scalar' and 'gen' statements both within the 
program and externally after ml maximise (when outside the program I
have 
the additional problem of distinguishing between the two shape
parameters 
currently both _cons in the output).

Any help would be much appreciated! My code is currently along the lines

of...

/* Full log-likelihood */
program ggm, eclass
	version 10
	args todo b lnf
	
	/* Define parameters of GGM */
	tempvar mu sigma kappa 		
	mleval `mu' = `b', eq(1)	
	mleval `sigma' = `b', eq(2)
	mleval `kappa' = `b', eq(3)
	
	/* Temporary variables to be used in likelihood */
	tempvar n sumlny A B
	quietly gen double `n' = _N
	quietly egen double `sumlny' = total(ln($ML_y1))
	quietly gen double `A' = abs(`kappa')^(-2)
	quietly egen double `B' = 
total(exp((((abs(`kappa'))*(sign(`kappa'))/(`sigma'))*(ln($ML_y1)-`mu'))
))

	/* Log-likelihood */
	mlsum `lnf' = 
(`n'*(`A'-0.5)*ln(`A'))-(`n'*ln(`sigma'))-(`n'*lngamma(`A'))-`sumlny'+
///
	
((sign(`kappa')*(sqrt(`A'))/(`sigma'))*(`sumlny'-`n'*`mu'))-(`A'*`B')

	/* Some of the attempts at extracting sigma */
	scalar __sigma =_b[`sigma':_cons]
	gen __sigma =_b[`sigma':_cons]
	ereturn scalar sigma = _b[sigma:_cons]
end


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   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   |   Site index