alpbench.util.ensemble_constructor

Classes

Ensemble(estimator, num_estimators[, ...])

This class is used to create an ensemble of estimators.

class alpbench.util.ensemble_constructor.Ensemble(estimator, num_estimators, max_neighbors=None)

Bases: object

This class is used to create an ensemble of estimators. The ensemble can be used to predict the probabilities of the ensemble members and the classes of the ensemble members.

Parameters:

• estimator – object

• num_estimators – int

• max_neighbors – int (for k nearest neighbors) else None

estimator

object (the estimator to construct the ensemble of)

num_estimators

int (the number of estimators in the ensemble)

max_neighbors

int (for k nearest neighbors)

random_states

list (random states for the ensemble members)

estimators_

list (list containing the ensemble members)

learner_fqn

str (fully qualified name of the estimator)

fit(X, y)

Fits the ensemble and sets the attributes of the class.

Parameters:

• X (array-like of shape (n_samples, n_features)) – The training input samples.

• y (array-like of shape (n_samples,)) –

• integers. (The target values (class labels) as) –

Return type:

None

init()

Initializes the ensemble members.

Returns:

None

predict(X, alpha=None)

Predicts the classes of the ensemble members.

Parameters:

X (array-like of shape (n_samples, n_features)) – The training input samples. alpha : float, optional (default=None) The threshold for the normalized likelihoods of the ensemble members.

Returns:

preds

Return type:

predicted classes, array-like of shape (n_samples, n_estimators)

predict_proba(X, alpha=None)

Predicts the probabilities of the ensemble members.

Parameters:

X (array-like of shape (n_samples, n_features)) – The training input samples. alpha : float, optional (default=None) The threshold for the normalized likelihoods of the ensemble members.

Returns:

preds

Return type:

predicted probabilities, array-like of shape (n_samples, n_classes, n_estimators)