Compute Stacking Weights for Base Learners

Computes the stacking weights for an ensemble of base learners using cross-validated out-of-sample predictions.

Usage

ensemble_weights(
  y,
  X,
  type = "average",
  learners = NULL,
  cv_folds = 5,
  cv_subsamples = NULL,
  cv_results = NULL,
  custom_weights = NULL,
  silent = FALSE
)

Arguments

y

The outcome variable.

X

The feature matrix.

type

A character string or vector indicating the type(s) of ensemble weights to compute. Default is "average".

learners

Optional list of base learners. Required when cv_results is not supplied (learners are needed to run cross-validation). When cv_results is supplied, learners may be omitted; the number of learners is inferred from the cross-validation residuals. See ddml-intro for the full specification.

cv_folds

Number of cross-validation folds.

cv_subsamples

Optional list of subsamples for cross-validation.

cv_results

Optional pre-computed cross-validation results.

custom_weights

Optional custom weights matrix.

silent

A boolean indicating whether to suppress progress messages.

Value

A list containing:

weights

A matrix of computed ensemble weights.

cv_results

Cross-validation results used for computing weights.

See also

Other utilities: crosspred(), crossval(), ddml(), diagnostics(), ensemble(), shortstacking()

Examples

# \donttest{
y = AE98[, "worked"]
X = AE98[, c("age","agefst","black","hisp","othrace")]

# Compute stacking weights via NNLS
ew = ensemble_weights(y, X,
                      type = "nnls",
                      learners = list(list(what = ols),
                                     list(what = mdl_glmnet)),
                      cv_folds = 5,
                      silent = TRUE)
ew$weights
#>          nnls
#> [1,] 0.000000
#> [2,] 0.999721
# }