rflexBART: Draw samples from the regression tree prior

In flexBART: A More Flexible BART Model

Draw samples from the regression tree prior

Description

Draws several binary regression trees from the flexBART tree prior, resulting in a single prior draw from the sum-of-trees prior.

Usage

rflexBART(train_data,nd,verbose = TRUE,
          print_every = floor(nd/10),...)

Arguments

train_data	an object of class data.frame containing data used to train the model. As usual, rows (resp. columns) correspond to observations (resp. variables)
nd	Number of trees to draw
verbose	Logical, indicating whether a message should be printed to the R console after every print_every trees have been sampled. Default is FALSE.
print_every	If verbose == TRUE, then a message is printed to the R console after every print_every trees have been sampled. Default is floor(nd/10).
...	Additional arguments for setting prior hyperparameters (e.g., number of trees, \mu_{0}, \tau, etc.). See flexBART for details about additional arguments.

Details

This function is useful for drawing samples from the regression tree prior underpinning flexBART. Together, these sampled trees form a single ensemble of trees. The main utility of this function is to study how often certain observations are clustered together in individual trees. This is key to understanding how flexBART “borrows strength” across observations.

Value

A list containing

num_leafs	An integer vector of length nd recording the number of leaf nodes in each tree of the ensemble.
num_singletons	An integer vector of length nd recording the number of leaf nodes in each tree that contain only one observation.
num_empty	An integer vector of length nd recording the number of leaf nodes in each tree that contain no observations.
max_leaf_size	An integer vector of length nd recording the maximum number of observation contained in a leaf in each tree of the ensemble.
min_leaf_size	An integer vector of length nd recording the minimum number of observation contained in a leaf in each tree of the ensemble.
kernel	An nrow(train_data) x nrow(train_data) matrix whose (i,j) entry is the proportion of tree draws in which observations i and j land in the same leaf of the tree.

flexBART documentation built on Oct. 2, 2026, 1:07 a.m.