BooST: Estimate BooST

In gabrielrvsc/BooST: Boosting Smooth Trees (BooST)

Description

Estimates Boosting of Smooth Trees (BooST)

Usage

BooST(
  x,
  y,
  v = 0.2,
  p = 2/3,
  d_max = 4,
  gamma = seq(0.5, 5, 0.01),
  M = 300,
  display = FALSE,
  stochastic = FALSE,
  s_prop = 0.5,
  node_obs = nrow(x)/200,
  random = FALSE
)

Arguments

x	Design matrix with explanatory variables.
y	Response variable.
v	Learning rate (default 0.2).
p	Proportion of variables tested in each node split (default 2/3).
d_max	Number of splits in each tree (default 4).
gamma	Transiction function intensity. Bigger numbers makes the transition less smoth. The default is a sequence of values (0.5:5) to be randomized in each new node. Multiple values may be supplied in a vector to increase the model randomness.
M	Number of trees.
display	If TRUE, displays iteration counter.
stochastic	If TRUE the model will be estimated using Stochasting Gradient Boosting.
s_prop	Used only if stochastic=TRUE. Determines the proportion of data used in each tree.
node_obs	Equivalent to the minimum number of observations in a termina node for a discrete tree.
random	If TRUE trees are grown randomly (default = FALSE)

Value

An object with S3 class "Boost".

Model	A list with all trees.
fitted.values	Final model fitted values.
brmse	Boost rmse in each iteratiob.
Model	A list with all trees.
ybar	Average value of y used in the first iteration.
v	Chosen learning rate.
rho	Vector of gradient estimates for each iteration.
nvar	Numver of variables in x
varnames	colnames of x to be used in other functions.
params	Model parameters.
call	The matched call.

References

blablabla

Examples

## == to be made == ##

gabrielrvsc/BooST documentation built on July 27, 2021, 3:08 p.m.