Pre-trained OSSL prediction — predict_ossl

Applies the OSSL-distributed pre-trained PLSR / Cubist models for a set of soil properties to pre-processed spectra. Pre-trained models are loaded from ossl_models, a named list of property models that each must implement a predict(model, X) interface returning a data.frame with columns value, pi95_low, pi95_high. When ossl_models is NULL, the synthetic predictor is used.

Usage

predict_ossl_pretrained(
  X,
  properties,
  region = "global",
  ossl_models = NULL,
  ...
)

Arguments

X: A pre-processed numeric matrix (rows = horizons, columns = wavelengths).
properties: Character vector of OSSL-supported property names.
region: One of "global", "south_america", "north_america", "europe", "africa".
ossl_models: Optional named list of pre-trained models, keyed by property name.
...: Reserved.

Value

A data.table with columns horizon_idx, property, value, pi95_low, pi95_high, n_neighbors. n_neighbors is NA_integer_ for pre-trained models. The "backend" attribute records which path was taken.