sugeno-classifier 1.0.0

Implementation of the Sugeno classifier

Sugeno classifier

Implementation of the Sugeno classifier, which was presented in the paper "Machine Learning with the Sugeno Integral: The Case of Binary Classification" [1].

Installation

Dependencies

The Sugeno classifier requires:

• Python (>=3.8)
• NumPy (>=1.19.0)
• scikit-learn (>=0.24.1)
• PuLP (>=2.4)

User installation

Use the package manager pip to install the Sugeno classifier.

pip install sugeno-classifier

Usage

The implementation is compatible to scikit-learn and can be used like other algorithms from this library. In order to use the Sugeno classifier, import the class SugenoClassifier from the module sugeno_classifier from the package classifier. Some examples are shown below:

First example

Use the contructor and the function fit to initialize the Sugeno classifier for a given dataset.

>>> from classifier.sugeno_classifier import SugenoClassifier
>>> X = [[1, 3, 2],
...      [2, 1, 3]]
>>> y = [0, 1]
>>> sc = SugenoClassifier()
>>> sc.fit(X, y)

Use the function predict to classify samples.

>>> Z = [[3, 2, 1],
...      [1, 2, 3]]
>>> sc.predict(Z)
array([0, 1])

Example with hyperparameters

The Sugeno classifier has two hyperparameter, the maxitivity and the margin, which can be set in the constructor. Both can influence the classification performance. See [1] for more information.

>>> from classifier.sugeno_classifier import SugenoClassifier
>>> X = [[1, 3, 2],
...      [2, 1, 3]]
>>> y = [0, 1]
>>> sc = SugenoClassifier(maxitivity=2, margin=0.01)
>>> sc.fit(X, y)

Again, the function predict can be used to classify samples. Note the different output compared to the first example.

>>> Z = [[3, 2, 1],
...      [1, 2, 3]]
>>> sc.predict(Z)
array([1, 1])

Example with different class labels

The classes do not have to be labeled with 0 and 1, they can be any integer numbers or strings. The label, which is smaller in terms of the relation < or lexicographically ordering, is assigned to negative class and the other one to the positive class.

The first example contains the class labels 2 and 4. Label 2 is assigned to the negative class and label 4 is assigned to the positive class because of 2<4.

>>> from classifier.sugeno_classifier import SugenoClassifier
>>> X = [[1, 3, 2],
...      [2, 1, 3]]
>>> y = [2, 4]
>>> sc = SugenoClassifier()
>>> sc.fit(X, y)
>>> Z = [[3, 2, 1],
...      [1, 2, 3]]
>>> sc.predict(Z)
array([2, 4])

The second example contains the class labels 'one' and 'two'. Label 'one' is assigned to the negative class and label 'two' is assigned to the positive class because 'one' comes lexicographically first.

>>> from classifier.sugeno_classifier import SugenoClassifier
>>> X = [[1, 3, 2],
...      [2, 1, 3]]
>>> y = ['one', 'two']
>>> sc = SugenoClassifier()
>>> sc.fit(X, y)
>>> Z = [[3, 2, 1],
...      [1, 2, 3]]
>>> sc.predict(Z)
array(['one', 'two'])

License

MIT

References

[1] Sadegh Abbaszadeh and Eyke Hüllermeier. Machine Learning with the Sugeno Integral: The Case of Binary Classication. 2020.