On Mon, 25 Jun 2001, Wei Wang wrote:
> Dear friends,
>
> I am trying to find out the best mixed-effects model for about 200 groups of
> data, respectively . I ended up with using a loop to do the modeling
> automatically. It will produce the the fitted summary of the candidate
> models, as well as some graphs, which can help me to pick the "best" one.
> However, sometimes, though not often, the script stops because of error
> (mainly convergence problems). This makes the automation less meaningful
> So, my question is:
>
> Are we able to force Splus program not to be stopped by the error, in case
> the error will not affect the rest of the computation?
Look back in the recent archives for almost exactly this question,
asked by Jean Adams. Assuming you have a fairly recent version of
S-PLUS, the answer is the function try(), which only has a help page in
6.0, but exists in 2000 and 5.x and (I think) 4.x.
(If you don't, there is an simple version of try in V&R's `S Programming'.)
--
Brian D. Ripley, ripley@stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272860 (secr)
Oxford OX1 3TG, UK Fax: +44 1865 272595
|