[Jprogramming] Bayesian inference?

[Bill Harris]

"Shaw, Ewart" <Ewart.Shaw at warwick.ac.uk> writes:

> [This is a belated follow-up to posts by Bill Harris & Mike Day, Aug 30th & 31st]
>
> I think Mike Day's probably right in recommending following up links
> between J, R & BUGS.  Note also the recently published (August 2007)
> paperback "Bayesian Computation with R" by Jim Albert - but beware of
> misprints etc.!
>
> MCMC is inherently iterative, and can be relatively slow in J.

This is an even more belated follow-up.  I've been doing a bit of
Bayesian analysis in RStan (http://mc-stan.org/).  While realizing I
could use the J - R - RStan connection, it might be easier and faster to
call Stan directly from J.

I don't have the time for the foreseeable future to implement something
like that, but I could see Stan giving J some high-end analysis
capability it lacks with a relatively rapid sampler.  I could foresee
utility in a verb to make calling Stan easy, a verb to replace RStan's
extract function to get the data out of a Stan object, and a lab to show
both how to use Stan from J and a few simple (or not-so-simple) examples
of analysis using Stan -- perhaps as easy as working through the
examples in the Stan manual.  I gather that the Stan community would
welcome Stan support in other languages; there's already a PyStan out
there.

Just an idea.

Bill
-- 
Bill Harris                  
Facilitated Systems
http://makingsense.facilitatedsystems.com/