My implementation of the stacked ensemble Super Learner as described in Mark J. van der Laan et al, (2007). The Super Learner is a heterogeneous stacked ensemble classifier. This is a classification model that uses a set of base classifiers of different types, the outputs of which are then combined in another classifier at the stacked layer.
superlearner.py
contains my implementation of the classifier.analysis.ipynb
is a jupyter notebook demonstrating an example usage on the fashion MNIST dataset.
Navigate to the repository folder and simply run from superlearner import SuperLearnerClassifier
in python.
For example on the Iris dataset:
from superlearner import SuperLearnerClassifier
from sklearn.datasets import load_iris
iris = load_iris()
sl_model = SuperLearnerClassifier(use_stacked_prob=False)
sl_model.fit(pd.DataFrame(iris.data), iris.target)
sl_model.predict(pd.DataFrame(iris.data))