hi: you also may want to look at the admit package. it does metropolis
hastings using a weighted mixture of t-distributions so you just need to
write a function for the likelihood you're trying to get the parameters for.
I don't know about it's speed or efficiency for large data sets but
you could
64 Bit R w/JAGS seems to be stalling out as well, I ran a test run of 100
iterations and it's been hanging for 8 hours so that doesn't seem to be the
solution. I'll take a look at PYMC. That CppBUGS package looks pretty
interesting, I'll keep my eye on it.
My C programming book arrives today fro
There are better alternatives for big data than to revert to C.
http://code.google.com/p/pymc/
http://github.com/armstrtw/CppBugs (still alpha)
-Whit
On Mon, Mar 14, 2011 at 11:06 AM, nblarson wrote:
> Has anybody had issues running MCMC (either BUGS or JAGS) on data sets of
> this magnitude (
Has anybody had issues running MCMC (either BUGS or JAGS) on data sets of
this magnitude (ie 30k x 20-30). I've been trying to run a hierarchical
random effects model on expression data but R completely stalls out on jobs
run on 32bit R on our server (doesn't respond...then eventually crashes out
4 matches
Mail list logo
