ReliefF<T>

Models & Types · AiDotNet.Preprocessing.FeatureSelection.Filter

ReliefF algorithm for feature selection based on nearest neighbor differences.

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

Implements: IFixedCountFeatureSelector<T>, ISupervisedFeatureSelector<T>

For Beginners

ReliefF picks random samples and looks at their nearest neighbors. If a feature differs between a sample and its nearby “enemies” (different class) but stays similar to nearby “friends” (same class), it’s useful. The algorithm rewards features that help separate classes and penalizes those that don’t.

How It Works

ReliefF estimates feature relevance by sampling instances and comparing them to their nearest neighbors of the same class (hits) and different classes (misses). Features that differentiate between classes get higher weights.

Constructors

Constructor Summary
ReliefF(int nFeaturesToSelect, int nNeighbors, int nSamples, int? randomState, int[] columnIndices)

Properties

Property Summary
Vector<T> FeatureWeights { get; }
bool SupportsInverseTransform { get; }
int NFeaturesToSelect { get; }
int NNeighbors { 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)