gpbart: GP-BART: Gaussian Processes Bayesian Additive Regression...

In MateusMaiaDS/gpbart: GP-BART: Gaussian Processes Bayesian Additive Regression Trees

GP-BART: Gaussian Processes Bayesian Additive Regression Trees

Description

GP-BART is an extension to to the Bayesian Additive Regression Trees (BART)

Usage

gpbart(
  x_train,
  y,
  x_test,
  n_tree = 20,
  node_min_size = 2,
  n_mcmc = 3500,
  n_burn = 1500,
  alpha = 0.95,
  beta = 2,
  df = 3,
  sigquant = 0.9,
  kappa = 2,
  tau = 100,
  scale_bool = TRUE,
  nu = 1,
  rand_tau_init = TRUE,
  verbose = TRUE
)

Arguments

x_train	Set of explanatory variables of the training data. It must be a data.frame()
y	Response variable for the training data
x_test	Set of explanatory variables of the test data. It must be a data.frame()
n_tree	Number of Trees used in the GP-BART model.
node_min_size	Node minimum size of observations within a terminal node
n_mcmc	The number of MCMC iterations
n_burn	The number of MCMC iterations to be trated as burn in
alpha	Base parameter for the tree prior
beta	Power parameter for the tree prior
df	Degrees of freedom for the residual precision prior
sigquant	The quantile of the residual precision prior;
kappa	The number of prior standard deviations away from the from the range of the response.
tau	Initial value for the residual precision
scale_bool	A Boolean to choose if will be scaled or not.
nu	Value for the GP precision. The default value is \nu = 4\kappa^{2}T
rand_tau_init	A Boolean to let the initial value be initialised or not. The default is TRUE.
verbose	Verbosity flag for printing progress. The default is TRUE

MateusMaiaDS/gpbart documentation built on Jan. 26, 2024, 7:07 a.m.