R package of Fast Sparse Group Lasso (beta version)
FSGL can be installed via Github using devtools
# install.packages("devtools")
library(devtools)
devtools::install_github("xhnnnnn/FSGL")
You'll need a working C++11 compiler, which can obtained by installing Xcode on MacOS, and RTools on Windows.
library(FSGL)
n = 100; p = 20; size.groups = 10
index <- ceiling(1:p / size.groups)
X = matrix(rnorm(n * p), ncol = p, nrow = n)
beta = (-2:2)
y = X[,1:5] %*% beta + 0.1*rnorm(n)
data = list(x = X, y = y)
fit = FSGL(data, index)
Ida Y, Fujiwara Y, Kashima H. Fast Sparse Group Lasso[J]. Advances in Neural Information Processing Systems, 2019, 32: 1702-1710.