We would like to see whether adaptive shrinkage prior such as Horseshoe has better performance in variable selection under the negative binomial model. Recall that the hierarchical model with the horseshoe prior on is
where is adaptively shrunken by whose distribution is an independent identical half-Cauchy distribution (iid) and controls the global level of shrinkage. We can rewrite the Horseshoe prior using the following parameter expansion:where we note that , and follows a truncated Gamma distribution. The posterior distribution for all parameters of interests is as follows where we again eliminate the additive effect of the intercept by integrating it out of the posterior distribution
where , and is the normalizing kernel and is normalized version of .
The posterior of follows a -dimensional multivariate distribution
given .The posterior of follows a Gamma distribution
The posterior of follows a truncated Gamma distribution The condition distributions for overdispersion parameter , and for each are the same as the paper.