SubBoost: SubBoost

In chstaerk/SubBoost: Subspace Boosting With Multivariable Base-learners

Description

This function implements the subspace boosting algorithm (SubBoost).

Usage

SubBoost(
  data,
  Iter,
  size.fixed = NULL,
  tau = 0.01,
  const = 0,
  savings = 1,
  family = "normal",
  s_max = 20,
  automatic.stopping = TRUE,
  plotting = FALSE
)

Arguments

data	should be a list with data$x as design matrix and data$y as response
Iter	iterations
size.fixed	default is set to NULL
tau	parameter tau - default is set to 0.01
const	parameter const - default is set to 0
savings	default is set to 1
family	default is set to "normal"
s_max	default is set to 20
automatic.stopping	default is set to TRUE
plotting	default is set to FALSE

Examples

require(TH.data)

# input data format:
# list with design matrix in data$x and response vector in data$y
data <- list()
data$x <- as.matrix(bodyfat[,-2])
data$y <- as.vector(bodyfat$DEXfat)

# constant (gamma) in EBIC
const <- 0 # classical BIC

# learning rate
tau <- 0.01

# (maximum) number of iterations
Iter <- 1000

# SubBoost (only applicable for low-dimensional settings, e.g. p<=20)
outputSub <- SubBoost(data = data, Iter = Iter, const = const, tau = tau)
outputSub$selected # selected variables by SubBoost
outputSub$coef # estimated coefficient vector by SubBoost

chstaerk/SubBoost documentation built on Dec. 19, 2021, 4:06 p.m.