Compute exact Shapley values for an ahead::dynrmf model

dynrmf_shap(
  y,
  xreg_fit,
  xreg_predict,
  fit_func = ahead::ridge,
  verbose = TRUE,
  tol = 1e-10
)

Arguments

y: Training time series (passed to ahead::dynrmf as y)
xreg_fit: Regressor matrix used during training
xreg_predict: Regressor matrix for the forecast horizon (h x p); h and feature names are derived from this matrix.
fit_func: Fitting function for ahead::dynrmf (default: ahead::ridge)
verbose: Print progress per horizon and efficiency check (default: TRUE)
tol: Tolerance for the efficiency-axiom check (default: 1e-10)

Value

A list of class "dynrmf_shap" with: $S : h x p matrix of Shapley values (per variable, per horizon) $X : test regressor matrix $baseline : scalar baseline (mean of empty-coalition forecasts) $baselines : per-horizon baseline forecasts $predictions : per-horizon full-coalition forecasts $feature_names $h