neighbayes.models.priors.NegBinPriors

class neighbayes.models.priors.NegBinPriors(beta_mu=None, beta_sigma=None, sigma2_alpha=2.0, sigma2_beta=None, nu=4.0, alpha_sigma=2.5, alpha_nu=3.0, alpha_fixed=None)[source]

Priors for neighbayes.models.NegBin.

Adds overdispersion parameters for the NB2 likelihood: a Half-Student-t(alpha_nu, alpha_sigma) prior on α, or alpha_fixed to hold α at a value. A large fixed α (say 10-20 times the typical mean count) makes the NB a Poisson stand-in that keeps the exact Pólya–Gamma sampler; very large values (≳ 1000) mix poorly.

__init__(beta_mu=None, beta_sigma=None, sigma2_alpha=2.0, sigma2_beta=None, nu=4.0, alpha_sigma=2.5, alpha_nu=3.0, alpha_fixed=None)[source]

Methods

__init__([beta_mu, beta_sigma, ...])

Attributes

alpha_fixed

alpha_nu

alpha_sigma

beta_mu

beta_sigma

nu

sigma2_alpha

sigma2_beta

alpha_fixed = None[source]
alpha_nu = 3.0[source]
alpha_sigma = 2.5[source]
beta_mu = None[source]
beta_sigma = None[source]
nu = 4.0[source]
sigma2_alpha = 2.0[source]
sigma2_beta = None[source]