This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ed in (14), namely
p(θ|y ) ∼ Beta (mH + α, m − mH + β ) .
we run a chain with r = 0.1 until we have collected 25,000 accepted draws. We
then discard the initial 200 samples (burn-in) and keep one out of every 100
samples from what remains (thinning). The next ﬁgure shows a normalized
32 histogram of the resulting draws, along with the analytical posterior. The
running time to generate the MCMC draws was less than a second, and they
are a reasonable approximation to the posterior. 7 MCMC with OpenBUGS There is a great deal of “art” to MCMC simulation, and a large body of
research on the “right” way to do the simulations. OpenBUGS is a nice
software package that has built-in years of research in MCMC and can draw
posterior samples in a fairly automated way. It is great for doing Bayesian
analysis without having to get your hands too dirty with MCMC details.
Here we demonstrate OpenBUGS using two examples.
OpenBUGS is old and is infrequently updated, but is still functional and
powerful. The graphical interface version of OpenBugs is available only as
a Windows executable...
View Full Document
This note was uploaded on 03/24/2014 for the course MIT 15.097 taught by Professor Cynthiarudin during the Spring '12 term at MIT.
- Spring '12