Statalist


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

st: setting initial values for -stcox-


From   Sean.Chu@frb.gov
To   statalist@hsphsun2.harvard.edu
Subject   st: setting initial values for -stcox-
Date   Mon, 31 Mar 2008 09:43:16 -0400

Apologies if this message appears twice.  (I think there was an error on 
my first attempt.)

I'm estimating a Cox proportional hazards model, and am running into 
problems with convergence.  The loglikelihood maximization spits out the 
following:

Iteration 279: log pseudolikelihood = -201462.74  (backed up)
Iteration 280: log pseudolikelihood =  -201462.9  (backed up)
Iteration 281: log pseudolikelihood = -201464.99  (backed up)
flat region resulting in a missing likelihood
r(430);

end of do-file
r(430);

I'd like to set a starting value different from the default value, to see 
whether that helps.  Apparently, -stcox- doesn't allow you to use the 
-from()- option.  I'm currently trying something like

matrix b0 = <initial value>
stset time, id(groupid)  failure(patientdies);
stcox rhsvar1 rhsvar2 rhsvar3, cluster(groupid) from(b0,copy);

Does anyone know of a better solution?  Thanks in advance.
*
*   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