JMI<T>

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

Joint Mutual Information (JMI) for feature selection.

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

Implements: IFixedCountFeatureSelector<T>, ISupervisedFeatureSelector<T>

For Beginners

JMI doesn’t just look at each feature individually; it considers how pairs of features work together. When selecting a new feature, it asks: “How much new information does this feature provide about the target when combined with each already-selected feature?” This helps find features that complement each other.

How It Works

JMI selects features by maximizing the joint mutual information between candidate features and already-selected features with respect to the target. It considers second-order feature interactions for improved selection.

Constructors

Constructor Summary
JMI(int nFeaturesToSelect, int nBins, int[] columnIndices)

Properties

Property Summary
Vector<T> JmiScores { get; }
Vector<T> RelevanceScores { 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)