QuadraticDiscriminantAnalysis.GroupSeparability Property

public Matrix<double> GroupSeparability { get; }

Public ReadOnly Property GroupSeparability As Matrix(Of Double)
	Get

public:
property Matrix<double>^ GroupSeparability {
	Matrix<double>^ get ();
}

member GroupSeparability : Matrix<float> with get

Property Value

The group separability matrix provides a symmetric measure of how well each pair of classes can be distinguished from each other. It combines the Mahalanobis distances with the determinants of the covariance matrices to provide a scale-invariant measure of class separation.

Higher values indicate better separation between classes, making classification more reliable. Lower values suggest that the classes overlap more in feature space, potentially leading to higher misclassification rates.

Reference