CFS<T>

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

Correlation-based Feature Selection (CFS) for selecting feature subsets.

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

Implements: IFixedCountFeatureSelector<T>, ISupervisedFeatureSelector<T>

For Beginners

CFS looks for features that are good at predicting the target but aren’t redundant with each other. If two features give the same information, CFS keeps only one. This prevents selecting features that are copies of each other while maximizing predictive power.

How It Works

CFS evaluates feature subsets based on a heuristic: good feature sets contain features highly correlated with the class but uncorrelated with each other. Uses greedy forward selection with the CFS merit function.

Constructors

Constructor Summary
CFS(int nFeaturesToSelect, int[] columnIndices)

Properties

Property Summary
Vector<T> FeatureClassCorrelations { 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)