Figure 1
Imbalanced data in two dimensions. The decision boundary of a typical classifier shows a bias towards the majority class.
Figure 2
Visual representation of the SMOTE data generation mechanism.
Figure 3
Comparison between the data generation mechanisms of SMOTE and G-SMOTE. SMOTE uses linear interpolation, while G-SMOTE defines a circle as the permissible data generation area.
Figure 4
UML class diagrams and callgraphs of main classes and methods.
Figure 5
UML BaseDistributor and DensityDistributor class diagrams and callgraphs of main classes and methods.
Figure 6
UML BaseOverSampler and BaseClusterOversampler class diagrams and callgraphs of main classes and methods.