fatf.utils.models.validation.check_model_functionality

fatf.utils.models.validation.check_model_functionality(model_object: Union[object, type], require_probabilities: bool = False, suppress_warning: bool = False) → bool

Checks whether a model object has all the required functionality.

Examines a model_object and ensures that it has all the required methods with the correct number of parameters (excluding self): __init__ (at least 0), fit (at least 1 for unsupervised and at least 2 for supervised models), predict (at least 1) and, if required (require_probabilities=True), predict_proba (at least 1, in which case fit requires at least 2).

Parameters

model_objectUnion[object, type]

A Python object (either instantiated or just an object reference) that represents a predictive model.

Changed in version 0.0.2: Added the possibility of checking functionality of non-initialised objects.

require_probabilitiesboolean, optional (default=False)

A boolean parameter that indicates whether the model object should contain a predict_proba method. Defaults to False.

suppress_warningboolean, optional (default=False)

A boolean parameter that indicates whether the function should suppress its warning message. Defaults to False.

Returns

is_functionalboolean

A boolean variable that indicates whether the model object has all the desired functionality.

Raises

TypeError

The require_probabilities or suppress_warning parameter is not a boolean.

Warns

UserWarning

Warns about the required functionality that the model object lacks.