Constructors for prior distributions used by bayesma(), robma(),
meta_reg(), and related fitting functions. Each returns an object of
class bayesma_prior that stores the family and its hyperparameters.
Usage
normal(mean, sd)
half_normal(mean, sd)
half_cauchy(location, scale)
half_student_t(df, location, scale)
exponential(rate)
uniform(lower, upper)
dirichlet(alpha)
beta(alpha, beta)
scaled_inv_chi_sq(df, scale)
lkj(eta)Arguments
- mean
Numeric. Mean of a normal or half-normal prior.
- sd
Numeric. Standard deviation of a normal or half-normal prior.
- location
Numeric. Location of a Cauchy or Student-t prior.
- scale
Numeric. Scale of a Cauchy, Student-t, or scaled inverse chi-squared prior.
- df
Numeric. Degrees of freedom for a Student-t or scaled inverse chi-squared prior.
- rate
Numeric. Rate parameter of an exponential prior.
- lower, upper
Numeric. Endpoints of a uniform prior.
- alpha
Numeric (vector for Dirichlet, scalar for Beta). Shape parameter(s).
- beta
Numeric. Second shape parameter of a Beta prior.
- eta
Numeric. Shape parameter of an LKJ correlation prior.
eta = 1is uniform;eta > 1concentrates toward the identity;eta < 1concentrates toward perfect correlation.