RReliefF<T>

Models & Types · AiDotNet.Preprocessing.FeatureSelection.Filter.Relief

RReliefF algorithm for regression problems.

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

Implements: IFixedCountFeatureSelector<T>, ISupervisedFeatureSelector<T>

For Beginners

Standard Relief uses class labels (hit/miss). RReliefF works with numeric targets by weighting neighbors based on how similar their target values are to the sample’s target.

How It Works

RReliefF extends ReliefF to handle continuous targets (regression), following Robnik-Sikonja and Kononenko (1997): over the k nearest neighbours of each sampled instance it accumulates the probability of a different target (N_dC), of a different feature value (N_dA) and of both (N_dC&dA), every neighbour having the same influence 1/k, and weights each feature by W = N_dC&dA / N_dC - (N_dA - N_dC&dA) / (m - N_dC).

Constructors

Constructor Summary
RReliefF(int nFeaturesToSelect, int nNeighbors, int nIterations, 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)