Hi,
I want to transpose a model written in R into a JAGS model, however, in one part of this model I use the function copula to obtain a correlation between several binomial distributions.
Is there a way to reproduce that in JAGS ?
Thanks,
Arnaud
Arnaud
2013-10-23
Hi,
I want to transpose a model written in R into a JAGS model, however, in one part of this model I use the function copula to obtain a correlation between several binomial distributions.
Is there a way to reproduce that in JAGS ?
Thanks,
Arnaud
Martyn Plummer
2013-10-24
Well, there are lots of different copulas. Which one are you using?
Arnaud
2013-11-14
Hi Martyn,
Sorry for the delay for my answer.
Currently, I use normal copula to sample into correlated binomial distributions using a code similar to the following example.
require(copula) probAndSize <- matrix(c(0.8,0.6,0.3,0.2,50,20,30,10), byrow=F, ncol=2) correl <- 0.8 norm.cop <- normalCopula(correl, dim=4) rcop <- rCopula(1, norm.cop) out <- sapply(seq(1,4), FUN = function(x){qbinom(rcop[,x], prob = probAndSize[x,1], size=probAndSize[x,2])/probAndSize[x,2]})
So, if I transpose that into a Bayesian model, I need a way to define correlations between several binomial distributions.
Arnaud
Arnaud
2013-11-14