estimator_checks

Utilities for unit testing and sanity checking estimators.

check_disappearing_features(model, dataset)

The model should work fine when features disappear.

check_emerging_features(model, dataset)

The model should work fine when new features appear.

check_estimator(model)

Check if a model adheres to river's conventions. This will run a series of unit tests. The nature of the unit tests depends on the type of model.

PARAMETER	DESCRIPTION
model

check_init_default_params_are_not_mutable(model)

Mutable parameters in signatures are discouraged, as explained in https://docs.python-guide.org/writing/gotchas/#mutable-default-arguments We enforce immutable parameters by only allowing a certain list of basic types.

check_learn_one(model, dataset)

learn_one should return the calling model and be pure.

check_predict_proba_one(classifier, dataset)

predict_proba_one should return a valid probability distribution and be pure.

check_predict_proba_one_binary(classifier, dataset)

predict_proba_one should return a dict with True and False keys.

check_shuffle_features_no_impact(model, dataset)

Changing the order of the features between calls should have no effect on a model.

check_tags(model)

Checks that the _tags property works.

seed_params(params, seed)

Looks for "seed" keys and sets the value.

wrapped_partial(func, *args, **kwargs)

Taken from http://louistiao.me/posts/adding-name-and- doc-attributes-to-functoolspartial-objects/

yield_checks(model)

Generates unit tests for a given model. Parameters: model (base.Estimator)