xpectra.protocols

Experimental orchestrators shared by the driver scripts.

Functions

get_estimator(model_name)	One fresh unfitted estimator from the xpectrass zoo.
run_calibration_eval(model_name, route, X, ...)	CV metrics per calibration method + a holdout reliability table.
run_generalization(model_name, route, X, y, ...)	Random 80/20 vs leave-one-study-out macro-F1 for one model.

xpectra.protocols.get_estimator(model_name)

One fresh unfitted estimator from the xpectrass zoo.

Parameters:

model_name (str)

xpectra.protocols.run_calibration_eval(model_name, route, X, y, n_classes, trust)

CV metrics per calibration method + a holdout reliability table.

Parameters:

• model_name (str)

• route (str)

• X (ndarray)

• y (ndarray)

• n_classes (int)

• trust (TrustConfig)

Return type:

tuple[list[dict[str, Any]], dict[str, Any]]

xpectra.protocols.run_generalization(model_name, route, X, y, study, trust)

Random 80/20 vs leave-one-study-out macro-F1 for one model.

Parameters:

• model_name (str)

• route (str)

• X (ndarray)

• y (ndarray)

• study (ndarray)

• trust (TrustConfig)

Return type:

dict[str, Any]