Run glmnet with Cross-Validation — run

Fits elastic net regularization using glmnet with cross-validation to select the optimal regularization parameter. Formats output to be compatible with gSuSiE result structure.

Usage

run_glmnet(
  X,
  y,
  family = gaussian(),
  alpha = 1,
  standardize = TRUE,
  nfolds = 10
)

Arguments

X: Design matrix of predictors (n × p).
y: Response vector or survival object.
family: Response family. Can be gaussian(), binomial(), poisson(), Gamma(), or "cox" for Cox regression.
alpha: Elastic net mixing parameter: 0 = ridge, 1 = lasso, 0 < alpha < 1 = elastic net.
standardize: Logical. Should variables be standardized? Default TRUE.
nfolds: Number of cross-validation folds. Default 10.

Value

A list containing:

cs: List with 'sets' component containing singleton credible sets for selected variables
theta: Vector of estimated coefficients (excluding intercept)
pip: Binary vector indicating variable selection (1 = selected, 0 = not selected)
lambda_1se: Lambda value within 1 standard error of minimum CV error
lambda_min: Lambda value that minimizes CV error
selected: Indices of selected variables

Details

This function serves as a wrapper around glmnet::cv.glmnet to provide output compatible with gSuSiE results. Uses lambda.1se for coefficient estimation, which provides more conservative variable selection than lambda.min.

Examples

if (FALSE) { # \dontrun{
# Simulate data
set.seed(123)
n <- 100; p <- 50
X <- matrix(rnorm(n * p), n, p)
y <- X[,1:3] %*% c(1, -1, 0.5) + rnorm(n)

# Run elastic net
result <- run_glmnet(X, y, alpha = 0.5)
print(result$selected)
} # }