Supervised Learning

IngeoML.supervised_learning API

class ConvexClassifier

ConvexClassifier combines the probabilities predicted by base classifiers in a convex manner. It is designed to work alongside with StackingClassifier.

>>> from sklearn.datasets import load_iris
>>> from sklearn.svm import LinearSVC
>>> from sklearn.ensemble import RandomForestClassifier
>>> from sklearn.ensemble import StackingClassifier
>>> from IngeoML import ConvexClassifier
>>> X, y = load_iris(return_X_y=True)
>>> svc = LinearSVC()
>>> forest = RandomForestClassifier()
>>> convex = ConvexClassifier()
>>> stack = StackingClassifier([('SVC', svc),
                                 ('Forest', forest)],
                                final_estimator=convex).fit(X, y)
>>> stack.final_estimator_.mixer
array([0.01783184, 0.98216816])

property mixer

Convex combination

__new__(**kwargs)

__init__()

fit(X: ndarray, y: ndarray)

Estimate the parameters given the dataset (X and y)

predict_proba(X)

Predict class probabilities

predict(X)

Predict the classes