Message-ID: <Pine.A41.4.44.0202210834530.72740-100000@homer37.u.washington.edu>
Date: 2002-02-21T16:43:10Z
From: Thomas Lumley
Subject: Factor analysis of categorical or mixed categorical/continuousdata in [R]
In-Reply-To: <x2wux6hkat.fsf@blueberry.kubism.ku.dk>
On 21 Feb 2002, Peter Dalgaard BSA wrote:
>
> On the face of it (which is as far as I am able to see), it would seem
> fairly easy to set up an MLE procedure if you treat all discrete
> variables as obtained by setting cutpoints on continuous latent
> variables. I suspect this is what MPlus is doing. The requisite normal
> integrals should be available through library(mvtnorm).
>
My understanding is that the optimisation problem is difficult (even
worse than factor analysis).
OTOH it seems that L-BFGS-B in optim() can work miracles -- I've been
working on another problem related to factor analysis and optim() fits a
model with 4000 constrained parameters in a couple of minutes and gets the
right answer (in stark contrast to my prior attempts to write an optimiser
more tuned to the specific problem).
-thomas
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._