ccboost: fit a predictive model with robust boosting algorithm

In ccboost: Robust Boosting

Description

Fit a predictive model with robust boosting algorithm. For loss functions in the CC-family (concave-convex), apply composite optimization by conjugation operator (COCO), where optimization is conducted by functional descent boosting algorithm. Models include the generalized linear models.

Usage

ccboost(
  x,
  y,
  weights,
  cfun = "ccave",
  s = 1,
  delta = 0.1,
  dfun = "gaussian",
  iter = 10,
  nrounds = 100,
  del = 1e-10,
  trace = FALSE,
  ...
)

Arguments

x	input matrix, of dimension nobs x nvars; each row is an observation vector. Can accept dgCMatrix
y	response variable. Quantitative for family="gaussian", or family="poisson" (non-negative counts). For family="binomial" should be a factor with two levels
weights	vector of nobs with non-negative weights
cfun	concave component of CC-family, can be "hacve", "acave", "bcave", "ccave", "dcave", "ecave", "gcave", "hcave"
s	tuning parameter of cfun. s > 0 and can be equal to 0 for cfun="tcave". If s is too close to 0 for cfun="acave", "bcave", "ccave", the calculated weights can become 0 for all observations, thus crash the program
delta	a small positive number provided by user only if cfun="gcave" and 0 < s <1
dfun	type of convex component in the CC-family, can be "gaussian", "binomial", "poisson"
iter	number of iteration in the COCO algorithm
nrounds	boosting iterations
del	convergency criteria in the COCO algorithm
trace	if TRUE, fitting progress is reported
...	other arguments passing to xgboost

Value

An object with S3 class xgboost.

weight_update

weight in the last iteration of the COCO algorithm

Author(s)

Zhu Wang
Maintainer: Zhu Wang [e-mail]

References

Wang, Zhu (2021), Unified Robust Boosting, arXiv eprint, https://arxiv.org/abs/2101.07718

Examples

x <- matrix(rnorm(100*2),100,2)
g2 <- sample(c(0,1),100,replace=TRUE)
fit1 <- ccboost(x, g2, cfun="acave",s=0.5, dfun="gaussian", trace=TRUE, 
                verbose=0, max.depth=1, nrounds=50)
fit2 <- ccboost(x, g2, cfun="acave",s=0.5, dfun="binomial", trace=TRUE,  
                verbose=0, max.depth=1, nrounds=50)
fit3 <- ccboost(x, g2, cfun="acave",s=0.5, dfun="poisson", trace=TRUE,      
                verbose=0, max.depth=1, nrounds=50)

ccboost documentation built on Jan. 25, 2021, 5:05 p.m.