Metalearner

Library of implemented Metalearners in AUCMEDI.

A Metalearner can be passed to an ensemble like Stacking and merges multiple class predictions into a single prediction.

Metalearner are similar to aggregate() functions, however Metalearners are models which require fitting before usage.

Assembled predictions encoded in a NumPy matrix with shape (N_models, N_classes).
Example: [[0.5, 0.4, 0.1],
          [0.4, 0.3, 0.3],
          [0.5, 0.2, 0.3]]
-> shape (3, 3)

Merged prediction encoded in a NumPy matrix with shape (1, N_classes).
Example: [[0.4, 0.3, 0.3]]
-> shape (1, 3)

Warning

Not all Metalearners support multi-label classification!

If a Metalearner has multi-label support can be found in its docs entry.

Metalearners are based on the abstract base class Metalearner_Base, which allows simple integration of custom Metalearners for Ensemble.

metalearner_dict = {'logistic_regression': LogisticRegression, 'naive_bayes': NaiveBayes, 'support_vector_machine': SupportVectorMachine, 'gaussian_process': GaussianProcess, 'decision_tree': DecisionTree, 'best_model': BestModel, 'weighted_mean': AveragingWeightedMean, 'random_forest': RandomForest, 'k_neighbors': KNearestNeighbors, 'mlp': MLP} module-attribute

Dictionary of implemented Metalearners.