I've known the theorem for years, and been vaguely aware of the principles, but I've never properly studies Baysian methods before. I just started a bit of self-study (actually I'm trying to improve my facility with the R language) and WOW! It's starting to make sense.

Any other fun-loving Baysians out there?

John

I know and practice that of which you speak.

As you may recall, we had a discussion about R in this here forum a few months back. How are you using it to do Bayesian analysis? That's not really what it is specifically designed for, after all. Though you can obviously use it for that purpose.

But if you want to blow your mind with some computing software that is specifically designed to implement Bayesian analyses (using MCMC), you should look into BUGS / WinBUGS and OpenBUGS:

http://www.mrc-bsu.cam.ac.uk/bugs/

http://mathstat.helsinki.fi/openbugs/

And you might also enjoy this article about the Rev. Bayes:

http://www.york.ac.uk/depts/maths/histstat/bayesbiog.pdf

I grabbed the pdf book Statistics using R with Biological Examples by Seefeld and Linder.

The first part of the book gives a bit of basic R and reviews a some elementary statistics in a mathematically nonrigorous way. When I got up to Chapter 9, I followed a couple of basic Baysian examples that just made the ideas clearer to me than they had been before.

At a completely "noob" level, I modeled a coin toss by first assuming that I didn't know what the probability distribution should look like. Then using R to successively modify the Beta distribution as I added new trials, the visual impact was quite compelling.

I've heard about BUGS, but I haven't used it yet. I'll have to check it out in the next while, thanks. Learning more about MCMC is definitely on my todo list.

Now I feel a bit goofy, but I guess what I first posted about was the intensity of the AHA! when I used R to work through some simple ideas.

John

WinBUGS is much easier to use than R, for doing Bayesian problems ... especially advanced ones.

This is getting a bit off topic, but Persi Diaconis (a student of Dai Vernon's) wrote a paper recently about the bias in coin tossing. I have not yet read it myself, but it might interest you:

http://www-stat.stanford.edu/~cgates/PER......ithJ.pdf

http://www.codingthewheel.com/archives/t......position

Also:

http://www.dartmouth.edu/~chance/chance_......ml#item4

Coin spinning is a step up from coin tossing. Because of the way different coins land when spun, a mixture of Beta distributions can be used as as sensible prior in this case. That makes for a more dramatic numerical illustration.

All the time, but only in conjunction with playing bridge or poker.

Well, and evaluating socio-political claims.

.

Thanks for all the links, Balducci. I'll read the Diaconis paper this week. I'm looking at BUGS, but it'll take me a while to decide if I'm ready for it yet.

John

PS

I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.
I will spell "Bayesian" correctly.

Lines don't count when you use copy\paste