junyangq/glmnetplus

glmnet package with initialization

glmnetPlus

glmnet package with user-provided initialization

Usage:

library(glmnetPlus)

X <- matrix(rnorm(500*100), 500, 100)
y <- rnorm(500)

fit0 <- glmnetPlus::glmnet(X, y, thresh = 1e-20)
# initialize with the solution at lambda 50, and find solutions for lambda 51 to lambda 60
fitPlus <- glmnetPlus::glmnet(X, y, lambda = fit$lambda[51:60], beta0 = fit$beta[, 50], type.gaussian = "naive", thresh = 1e-20)
max(abs(fitPlus$beta - fit0$beta[, 51:60]))

glmnetplus's Issues

Does glmnetplus contain any screening rules?

Hi,

I wanted to know whether any safe/strong screening rules have already been implemented in glmnetplus for family = "mgaussian"? I am trying to run an elastic net penalty for "mgaussian" (alpha = 0.99) and want to get a sense of what coding i would have to do on top of glmnetplus

Thanks
Prateek Sasan