Stacked Elastic Net Regression — starnet • starnet

Implements stacked elastic net regression.

starnet(
  y,
  X,
  family = "gaussian",
  nalpha = 21,
  alpha = NULL,
  nfolds = 10,
  foldid = NULL,
  type.measure = "deviance",
  alpha.meta = 1,
  penalty.factor = NULL,
  intercept = NULL,
  upper.limit = NULL,
  unit.sum = NULL,
  ...
)

Arguments

y: response: numeric vector of length \(n\)
X: covariates: numeric matrix with \(n\) rows (samples) and \(p\) columns (variables)
family: character "gaussian", "binomial" or "poisson"
nalpha: number of alpha values
alpha: elastic net mixing parameters: vector of length nalpha with entries between \(0\) (ridge) and \(1\) (lasso); or NULL (equidistance)
nfolds: number of folds
foldid: fold identifiers: vector of length \(n\) with entries between \(1\) and nfolds; or NULL (balance)
type.measure: loss function: character "deviance", "class", "mse" or "mae" (see cv.glmnet)
alpha.meta: meta-learner: value between \(0\) (ridge) and \(1\) (lasso) for elastic net regularisation; NA for convex combination
penalty.factor: differential shrinkage: vector of length \(n\) with entries between \(0\) (include) and \(Inf\) (exclude), or NULL (all \(1\))
intercept, upper.limit, unit.sum: settings for meta-learner: logical, or NULL (intercept=!is.na(alpha.meta), upper.limit=TRUE, unit.sum=is.na(alpha.meta))
...: further arguments passed to glmnet

Value

Object of class starnet. The slots base and meta contain cv.glmnet-like objects, for the base and meta learners, respectively.

Details

Post hoc feature selection: consider argument nzero in functions coef and predict.

References

Armin Rauschenberger, Enrico Glaab, and Mark A. van de Wiel (2021). "Predictive and interpretable models via the stacked elastic net". Bioinformatics 37(14):2012-2016. doi:10.1093/bioinformatics/btaa535 . (Click here to access PDF.)

Examples

# \donttest{
set.seed(1)
n <- 50; p <- 100
y <- rnorm(n=n)
X <- matrix(rnorm(n*p),nrow=n,ncol=p)
object <- starnet(y=y,X=X,family="gaussian")# }