Resolve PP variance components from metrics

Usage

resolve_ppi_variances(
  var_f = NULL,
  var_res = NULL,
  metrics = NULL,
  metric_type = NULL,
  m_labeled = NULL,
  correction = TRUE
)

var_f: Optional numeric scalar supplying $\operatorname{Var}(f)$ directly.
var_res: Optional numeric scalar supplying $\operatorname{Var}(Y-f)$ directly.
metrics: Optional named list of predictive-performance summaries. The required fields depend on metric_type.
metric_type: Character string identifying the metric bundle. Supported values are: "continuous" (regression-style metrics), "prob" (binary probabilistic metrics such as the Brier score), and "classification" (binary classification metrics such as a confusion matrix or precision/recall).
m_labeled: Labeled sample size associated with metrics; defaults to metrics$m_obs when present.
correction: Logical; apply the finite-sample adjustment when deriving moments from metrics (default TRUE).

List with numeric elements var_f and var_res.