I would like to estimate the alpha/beta of a beta-binomial.
The “n” of the binomial is observed, that is, my data is such that the number of trials changes for each experiment:
Throw N1 times - get K1 heads
Throw N2 times - get K2 heads
…
My model has the probability of the coin following a beta distribution.
Do you have an example of how I can configure n to be observed in pymc3?
What you are looking for is a
negative binomial distribution https://docs.pymc.io/api/distributions/discrete.html#pymc3.distributions.discrete.NegativeBinomial