Models from distributions

Open williambdean opened this issue 1 year ago • 1 comments

Could there be an API to only use the distributions? How would that make sense in the distributions that have various conjugate priors or models with fixed parameters (normal, etc)?

Some might be straight forward

from conjugate.distributions import Beta, Binomial, BetaBinomial

prior = Beta(1, 1)
posterior = Binomial...
posterior_predictive: BetaBinomial = Binomial(n=10, p=posterior)

Sep 26 '24 14:09 williambdean

These are now closer linked via the add_associated_likelihood function:

https://github.com/williambdean/conjugate/blob/49adf097bca219c442172317aa2ce76a676145b5/conjugate/models.py?plain=1#L84-L90

Jun 06 '25 19:06 williambdean