4 ms·
It seems statistically dubious to me that the point k = infinity is in the sample space and that the model is well defined. Does this model require a proper pr
by clircle 5y ago
It seems statistically dubious to me that the point k = infinity is in the sample space and that the model is well defined.
Does this model require a proper prior on p?
Cool blog post.
- volkale 5y ago- Using the point k=infinity is how I would define it mathematically. In the Stan code I mapped that point to k=-1. - I don't think a proper prior is required. - Thank you :) (Disclaimer: I am the author of the blog post.)
- clircle 5y agoOn more reflection... the support of p is finite, so the prior ought to be proper.