culebra.fitness_function.cooperative module

Cooperative feature selection fitness functions.

This sub-module provides fitness functions designed to the cooperative solving of a feature selection problem while the classifier hyperparamters are also being optimized. It provides the following fitness functions:

• KappaNumFeatsC: Tri-objective fitness class for feature selection. Maximizes the Kohen’s Kappa index and minimizes the number of features that a solution has selected and also the C regularazation hyperparameter of the SVM-based classifier.

• KappaFeatsPropC: Tri-objective fitness class for feature selection. Maximizes the Kohen’s Kappa index and minimizes the number of features that a solution has selected and also the C regularazation hyperparameter of the SVM-based classifier.

• AccuracyNumFeatsC: Tri-objective fitness class for feature selection. Maximizes the accuracy and minimizes the number of features that a solution has selected and also the C regularazation hyperparameter of the SVM-based classifier.

• AccuracyFeatsPropC: Tri-objective fitness class for feature selection. Maximizes the accuracy and minimizes the number of features that a solution has selected and also the C regularazation hyperparameter of the SVM-based classifier.