bayesplain.core.beta_binomial.log_marginal_likelihood

bayesplain.core.beta_binomial.log_marginal_likelihood(successes, n, a=1.0, b=1.0)[source]

Log marginal likelihood of binomial data under a Beta prior.

Integrating the binomial likelihood against a Beta(a, b) prior gives the beta-binomial probability mass

\[p(x) = \binom{n}{x} \frac{B(a + x,\; b + n - x)}{B(a, b)}\]

which is closed form – there is no integral left to evaluate numerically.

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.

Returns:

The natural log of the marginal likelihood, including the binomial coefficient. Computed in log space so that large n does not overflow.

Return type:

float