CorrelationFilter<T>
Models & Types ยท AiDotNet.Preprocessing.FeatureSelection.Unsupervised
Correlation Filter for removing highly correlated features.
Inheritance: TransformerBase<T, Matrix<T>, Matrix<T>> โ CorrelationFilter<T>
Implements: IFeatureSelector<T>
For Beginners
If two features are very similar (highly correlated), keeping both is redundant. This filter finds such pairs and keeps only one, reducing data size without losing much information.
How It Works
Identifies pairs of features with correlation above a threshold and removes one from each pair. This reduces multicollinearity and feature redundancy.
Constructors
| Constructor | Summary |
|---|---|
CorrelationFilter(double threshold, KeepStrategy<T> strategy, int[] columnIndices) |
Properties
| Property | Summary |
|---|---|
Matrix<T> CorrelationMatrix { get; } |
|
bool SupportsInverseTransform { get; } |
|
double Threshold { get; } |
|
int[] SelectedIndices { get; } |
Methods
| Method | Summary |
|---|---|
Matrix<T> FitTransform(Matrix<T> data) |
|
bool[] GetSupportMask() |
|
string[] GetFeatureNamesOut(string[] inputFeatureNames) |