RAEN: Random Ensemble Variable Selection for High Dimensional Data

In RAEN: Random Approximate Elastic Net (RAEN) Variable Selection Method

Description

Perform variable selection for high dimensional data

Usage

RAEN(
  x,
  y,
  B,
  ngrp = floor(15 * ncol(x)/nrow(x)),
  parallel = TRUE,
  family = "competing",
  ncore = 2
)

## S3 method for class 'RAEN'
predict(object, newdata, ...)

Arguments

x	the predictor matrix
y	the time and status object for survival
B	times of bootstrap
ngrp	the number of blocks to separate variables into. Default is 15*p/N, where p is the number of predictors and N is the sample size.
parallel	Logical TRUE or FALSE. Whether to use multithread computing, which can save consideratable amount of time for high dimensional data. Default is TRUE.
family	what family of data types. Default is 'competing'. Quantile regression for competing risks will be available through the developmental version on github
ncore	Number of cores used for parallel computing, if parallel=TRUE
object	the RAEN object containing the variable selection results
newdata	the predictor matrix for prediction
...	other parameters to pass

Value

a dataframe with the variable names and the regression coefficients

the linear predictor of the outcome risk

Examples

library(RAEN)
data(toydata)
x=toydata[,-c(1:2)]
y=toydata[,1:2]
fgrp<-deCorr(x, ngrp=20)

RAEN documentation built on Feb. 22, 2021, 9:07 a.m.