How do I obtain the DIC for model comparison purposes?
The following provides an example:
library(rjags)
mod1 <- jags.model("model.jags", data=Data, n.chains=4, n.adapt=500)
update(mod1, 500) # burn in
dic.pD <- dic.samples(mod1, 1000, "pD") # Deviance Information Criterion
dic.popt <- dic.samples(mod1, 1000, "popt") # Penalized expected deviance
dic.samples returns an estimate of the DIC for a given modelAnd here’s an example of comparing DIC between two models where mod1 and
mod2 are the jags model objects returned from jags.model() and then a burn-in
update(...) as above.
dic.mod1 <- dic.samples(mod1, 1000, "pD")
dic.mod2 <- dic.samples(mod2, 1000, "pD")
dic.mod1;
dic.mod2;
diffdic(dic.mod1, dic.mod2)
