Score specimens with a frozen trimmed-rCLR signed panel — score_ws_rclr

Applies [ws_rclr_trimmed()] independently to each incoming specimen and sums the transformed values over the training-frozen feature panel after applying the training-frozen signs. Extra features are ignored. Missing panel features fail closed rather than being imputed from a scored batch.

Usage

score_ws_rclr_panel(model, X, meta = NULL)

Arguments

model: A `ws_rclr_panel_model` returned by [fit_ws_rclr_panel()].
X: Numeric specimens-by-features matrix.
meta: Unused; accepted for the canonical roster score contract.

Value

One finite numeric score per specimen.