I have trained a naive Bayes classifier with on a dataset with a dichotomous outcome and multinomial attributes (predictors).
I managed to get a Maximum a posteriori (MAP) estimate which is good enough to determine class membership of a given instance of attributes.
However, to calculate things like the Highest Density Interval (HDI) I would need to get the posterior distribution rather than a point estimate. How can I do that?
And more specifically, how could I sample from that distribution? (via BUGS for instance).
Any help is appreciated