run_sl: Run a Super Learner for the provided subset of features

In bdwilliamson/nova: Perform Inference on Algorithm-Agnostic Variable Importance

Run a Super Learner for the provided subset of features

Description

Run a Super Learner for the provided subset of features

Usage

run_sl(
  Y = NULL,
  X = NULL,
  V = 5,
  SL.library = "SL.glm",
  univariate_SL.library = NULL,
  s = 1,
  cv_folds = NULL,
  sample_splitting = TRUE,
  ss_folds = NULL,
  split = 1,
  verbose = FALSE,
  progress_bar = NULL,
  indx = 1,
  weights = rep(1, nrow(X)),
  cross_fitted_se = TRUE,
  full = NULL,
  vector = TRUE,
  ...
)

Arguments

Y	the outcome
X	the covariates
V	the number of folds
SL.library	the library of candidate learners
univariate_SL.library	the library of candidate learners for single-covariate regressions
s	the subset of interest
cv_folds	the CV folds
sample_splitting	logical; should we use sample-splitting for predictiveness estimation?
ss_folds	the sample-splitting folds; only used if sample_splitting = TRUE
split	the split to use for sample-splitting; only used if sample_splitting = TRUE
verbose	should we print progress? defaults to FALSE
progress_bar	the progress bar to print to (only if verbose = TRUE)
indx	the index to pass to progress bar (only if verbose = TRUE)
weights	weights to pass to estimation procedure
cross_fitted_se	if TRUE, uses a cross-fitted estimator of the standard error; otherwise, uses the entire dataset
full	should this be considered a "full" or "reduced" regression? If NULL (the default), this is determined automatically; a full regression corresponds to s being equal to the full covariate vector. For SPVIMs, can be entered manually.
vector	should we return a vector (TRUE) or a list (FALSE)?
...	other arguments to Super Learner

Value

a list of length V, with the results of predicting on the hold-out data for each v in 1 through V

bdwilliamson/nova documentation built on Feb. 1, 2024, 10:04 p.m.