You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, the model specification is that of a normal likelihood. It'd be potentially useful to add options for other likelihoods --- like a lognormal, gamma, or poisson (for unbounded count outcomes). These would give a bit more flexibility in model specification at the user level while mechanically enforcing 0 as a lower bound for the outcome. I imagine that the lognormal would be the most straightforward to implement, but the others would also be useful as roadmap items in the future.