priorcontrol: Prior information for the DDPstar model

In DDPstar: Density Regression via Dirichlet Process Mixtures of Normal Structured Additive Regression Models

Prior information for the DDPstar model

Description

This function is used to set various parameters controlling the prior information to be used in the DDPstar function.

Usage

priorcontrol(m0 = NA, S0 = NA, nu = NA, Psi = NA, 
  atau = 1, btau = 0.005, a = 2, b = NA, alpha.fixed = FALSE, 
  alpha = 1, aalpha = 2, balpha = 2, L = 10)

Arguments

m0	A numeric vector. Hyperparameter; mean vector of the (multivariate) normal prior distribution for the parametric coefficients. NA signals autoinitialization, with defaults: a vector, of length Q (where Q is the number of parametric coefficients), of zeros, if the data are standardised and the least squares estimates of the regression coefficients if the data are not standardised.
S0	A numeric matrix. Hyperparameter; covariance matrix of the (multivariate) normal prior distribution for the parametric coefficients. NA signals autoinitialization, with defaults: 10I_{Q\times Q} if the data are standardised (where Q is the number of parametric coefficients) and \mathbf{\hat{\Sigma}} if the data are not standardised, where \mathbf{\hat{\Sigma}} is the estimated covariance matrix of the parametric coefficients obtained by fitting a linear model to the data.
nu	A numeric value. Hyperparameter; degrees of freedom of the Wishart prior distribution for the precision matrix associated with the parametric coefficients. NA signals autoinitialization, with default Q+2 (where Q is the number of parametric coefficients)
Psi	A numeric matrix. Hyperparameter; scale matrix of the Wishart prior distribution for the precision matrix associated with the parametric coefficients. NA signals autoinitialization, with defaults: I_{Q\times Q} if the data are standardised (where Q is the number of parametric coefficients) and to 30\mathbf{\hat{\Sigma}} if the data are not standardised, where \mathbf{\hat{\Sigma}} is the estimated covariance matrix of the parametric coefficients obtained by fitting a linear model to the data.
atau	A numeric value. Hyperparameter; shape parameter of the gamma prior distribution for the precisions (inverse variances) of the smooth/nonlinear/random terms. The default is 1.
btau	A numeric value. Hyperparameter; rate parameter of the gamma prior distribution for the precisions (inverse variances) of the smooth/nonlinear/random terms. The default is 0.005.
a	A numeric value. Hyperparameter; shape parameter of the gamma prior distribution for the precisions (inverse variances) of each component. The default is 2.
b	A numeric value. Hyperparameter; shape parameter of the gamma prior distribution for the precisions (inverse variances) of each component. NA signals autoinitialization, with defaults: 0.5 if the data are standardised and \frac{\hat{\sigma}^2}{2} if the data are not standardised
alpha.fixed	A logical value. If TRUE, the precision parameter of the Dirichlet Process is considered fixed. If FALSE, a Gamma prior distribution is placed on it.
alpha	A numeric value. Applicable when alpha.fixed = TRUE. Precision parameter of the Dirichlet Process. The default is 1.
aalpha	A numeric value. Applicable when alpha.fixed = FALSE. Hyperparameter; shape parameter for the Gamma prior distribution of the precision parameter of the Dirichlet process prior. The default is 2.
balpha	A numeric value. Applicable when alpha.fixed = FALSE. Hyperparameter; rate parameter for the Gamma prior distribution of the precision parameter of the Dirichlet process prior. The default is 2.
L	A numeric value. Upper bound on the number of mixture components. Setting L = 1 corresponds to a normal model. The default is 10.

Value

A list with components for each of the possible arguments.

See Also

DDPstar

Examples

library(DDPstar)
data(dde)
dde$GAD <- dde$GAD/7 # GAD in weeks

prior <- priorcontrol(a = 2, b = 0.5, aalpha = 2, balpha = 2, L = 20)

set.seed(10) # For reproducibility
fit_dde <- DDPstar(formula = GAD ~ f(DDE, bdeg = 3, nseg = 20, pord = 2, atau = 1, btau = 0.005), 
  data = dde, mcmc = list(nburn = 20000, nsave = 15000, nskip = 1), prior = prior, 
  standardise = TRUE, compute.lpml = TRUE, compute.WAIC = TRUE, compute.DIC = TRUE)
summary(fit_dde)

DDPstar documentation built on April 3, 2025, 8:46 p.m.