Fully Bayesian Computing(stat.columbia.edu)
stat.columbia.edu
Fully Bayesian Computing
http://www.stat.columbia.edu/~gelman/research/unpublished/fullybayesiancomputing-nonblinded.pdf
5 comments
The "Probablistic Functional Programming" paper was really fun, and it builds on some decades-old observations in mathematics.
You can easily extend the PFP framework to handle Bayes rule. I experimented with this a few years back, and was pleasantly surprised how natural it was:
http://www.randomhacks.net/articles/2007/02/22/bayes-rule-an...
(Or, for those who love details, a PDF: http://www.randomhacks.net/darcs/probability-monads/probabil... )
You can easily extend the PFP framework to handle Bayes rule. I experimented with this a few years back, and was pleasantly surprised how natural it was:
http://www.randomhacks.net/articles/2007/02/22/bayes-rule-an...
(Or, for those who love details, a PDF: http://www.randomhacks.net/darcs/probability-monads/probabil... )
I would welcome a single-paragraph summarization of the findings.
Is it wrong to read a journal paper and keep giggling at "posterior simulations"?
This is not Reddit. Kindly keep comments like this to a minimum.
It is wrong on HN. On digg.com it might be ok. Why is your comment grey?
No. Be pleased that it is not "posterior stimulations".
Butt otherwise, remember that this is not Reddit. Kindly keep comments like this to a minimum.
Butt otherwise, remember that this is not Reddit. Kindly keep comments like this to a minimum.
A Haskell embedding seems easier than adding new syntax to R, though you don't get all the nice algorithms that are already implemented in R.