lasso.cv: Select Predictors via (10-fold) Cross-Validation of the Lasso

In hdi: High-Dimensional Inference

Description

Performs (n-fold) cross-validation of the lasso (via cv.glmnet) and determines the prediction optimal set of parameters.

Usage

lasso.cv(x, y,
         nfolds = 10,
         grouped = nrow(x) > 3*nfolds,
         ...)

Arguments

x	numeric design matrix (without intercept) of dimension n * p.
y	response vector of length n.
nfolds	the number of folds to be used in the cross-validation
grouped	corresponds to the grouped argument to cv.glmnet. This has a smart default such that glmnet does not give a warning about too small sample size.
...	further arguments to be passed to cv.glmnet.

Details

The function basically only calls cv.glmnet, see source code.

Value

Vector of selected predictors.

Author(s)

Lukas Meier

See Also

hdi which uses lasso.cv() by default; cv.glmnet. An alternative for hdi(): lasso.firstq.

Examples

x <- matrix(rnorm(100 * 1000), nrow = 100, ncol = 1000)
y <- x[,1] * 2 + x[,2] * 2.5 + rnorm(100)
sel <- lasso.cv(x, y)
sel

Example output

Loading required package: scalreg
Loading required package: lars
Loaded lars 1.2

[1] 1 2

hdi documentation built on May 27, 2021, 3 p.m.