Prefers cross-validation class metrics when available; otherwise derives
metrics from the provided truth and prediction vectors.
Usage
class_metrics_from_fit(fit, test_truth = NULL, test_pred = NULL)
Arguments
- fit
List returned by [train_custom_rf()].
- test_truth
Optional truth vector (used only if CV metrics absent).
- test_pred
Optional prediction vector (used only if CV metrics absent).
Value
Data frame with columns `class`, `tp`, `fp`, `fn`, `support`,
`precision`, `recall`, `f1`, and optionally `fold` when sourced from CV.
Examples
truth <- c("Bcell", "Bcell", "Tcell", "Tcell")
pred <- c("Bcell", "Tcell", "Tcell", "Tcell")
class_metrics_from_fit(list(), truth, pred)
#> class tp fp fn support precision recall f1
#> Bcell Bcell 1 0 1 2 1.0000000 0.5 0.6666667
#> Tcell Tcell 2 1 0 2 0.6666667 1.0 0.8000000
