If you can write down the likelihood, you don't really need the EM
algorithm. Just code this with -ml-. (I've done this for univariate
mixtures years ago... for Stata 6, if not Stata 5). EM is a nice trick
that gives you faster convergence, but without the standard errors.
-ml- is going to take somewhat longer, but it is so well written and
tested that it is worth a little waiting, compared to writing your own
EM algorithm from scratch.
On 7/28/06, marcel spijkerman <[email protected]> wrote:
Hello,
I have written a ML procedure for a mixture in which the parameters of two
regimes are estimated together with the probabilities. Now I want to iterate
(in a loop) the ML procedure by first setting the probabilities , estimate
the other parameters (M-step), compute new probabilities given the estimated
parameters (E-step), re-estimate the other parameters and so on, up to the
moment the model converges.
1) How can I pass the computed probabilities into the ML procedure in
automated way?
2) Does ML store the scores of the maximised denistiy function (the are
needed to compute the new probabilities)?
Kind regards,
Marcel
*
* 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/
--
Stas Kolenikov
http://stas.kolenikov.name
*
* 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/
