SymmetricalUncertainty<T>

Models & Types · AiDotNet.Preprocessing.FeatureSelection.Filter

Symmetrical Uncertainty for feature selection based on normalized mutual information.

Inheritance: TransformerBase<T, Matrix<T>, Matrix<T>> → SymmetricalUncertainty<T>

Implements: IFixedCountFeatureSelector<T>, ISupervisedFeatureSelector<T>

For Beginners

Mutual information tells you how much one variable reveals about another, but the raw value depends on the variables’ complexity. Symmetrical Uncertainty fixes this by scaling the value between 0 and 1, where 0 means no relationship and 1 means perfect predictability in both directions.

How It Works

Symmetrical Uncertainty normalizes mutual information to the range [0, 1] by considering the entropy of both variables. It measures how much knowing one variable reduces uncertainty about the other, symmetrically.

Constructors

Constructor Summary
SymmetricalUncertainty(int nFeaturesToSelect, int nBins, double minSU, int[] columnIndices)

Properties

Property Summary
Vector<T> SUValues { get; }
bool SupportsInverseTransform { get; }
int NFeaturesToSelect { get; }
int[] SelectedIndices { get; }

Methods

Method Summary
Matrix<T> FitTransform(Matrix<T> data, Vector<T> target)
bool[] GetSupportMask()
string[] GetFeatureNamesOut(string[] inputFeatureNames)
void Fit(Matrix<T> data, Vector<T> target)