bayesplain.core.beta_binomial.draws¶
-
bayesplain.core.beta_binomial.draws(successes, n, a=
1.0, b=1.0, size=100000, rng=None)[source]¶ Draw independent samples from the exact Beta posterior.
These are i.i.d. draws from a distribution known in closed form – the same operation as rolling a die, not a Markov chain. There is nothing to check for convergence and no burn-in to discard.
- Parameters:¶
- successes : int¶
Number of successes observed.
- n : int¶
Number of trials.
- a : float, default 1.0¶
Shape parameters of the Beta prior.
- b : float, default 1.0¶
Shape parameters of the Beta prior.
- size : int, default 100_000¶
Number of draws.
- rng : numpy.random.Generator, optional¶
Random generator. Supply one for reproducibility.
- Returns:¶
Array of
sizedraws from the posterior.- Return type:¶
ndarray