Performs Bayesian analyses using some simple commonly-used models. The multivariate generalized ratio-of-uniforms method is used to simulate random samples from the required posterior distribution. The user can either choose hyperparameter values of a default prior distribution or specify their own prior distribution.
Currently three conjugate hierarchical models are available:
beta-binomial, gamma-Poisson and 1-way Analysis of Variance (ANOVA).
The function hef produces random posterior samples from for the
beta-binomial and gamma-Poisson models. The function hanova1
does this for the 1-way Analysis of Variance (ANOVA).
The rust package is used to
produce these samples.
See
vignette("bang-a-vignette", package = "bang") for a brief introduction
to the package and
vignette("bang-b-hef-vignette", package = "bang") and
vignette("bang-c-anova-vignette", package = "bang") for illustrations
of the use of the hef and hanova1 functions.
Northrop, P. J. (2017). rust: Ratio-of-Uniforms Simulation with Transformation. R package version 1.2.3. https://cran.r-project.org/package=rust.
hef for hierarchical exponential family models.
hanova1 for hierarchical one-way analysis of
variance (ANOVA).
set_user_prior to set a user-defined prior.