optim.lambda: select optimal tuning parameter lambda for lasso based on BIC...

In lcw68/G3proj: BIOS 735 Project

Description

This function will fit the glmlasso with a sequence value of tuning parameter lambdas and return a plot of BIC vs lambda and the optimal lambda with the smallest BIC.

Usage

optim.lambda(Xz, yz, lambda.min, lambda.max, len, plot = FALSE)

Arguments

Xz	the design matrix for glmlasso
yz	a vector of binary outcomes
lambda.min	the smallest lambda
lambda.max	the largest lambda
len	the number of lambdas for fitting
plot	logic variable, indicating if the lambda versus BIC plot should be returned

Details

This function will need glmlasso function.

Value

a plot of BIC vs lambda and the optimal lambda with the smallest BIC

Examples

set.seed(8675309)
Xz = scale(matrix(rnorm(5000), ncol=10))
bz = c(.5, -.5, .25, -.25, .125, -.125, rep(0, 4))
yz = rbinom(Nz,1,exp(Xz %*% bz)/(1+exp(Xz %*% bz)))
optim.lambda(Xz,yz,lambda.min = 0,lambda.max = 0.1,len = 200)

lcw68/G3proj documentation built on Dec. 21, 2021, 9:46 a.m.