IFeatureSelector<T>

Interfaces · AiDotNet.Interfaces

Defines a fitted transformer that keeps a subset of its input’s columns: a feature selector.

Known implementations: ClusteringBasedFS<T>, CorrelationFilter<T>, CorrelationRemovalSelector<T>, IFixedCountFeatureSelector<T>, IQRThreshold<T>, ISupervisedFeatureSelector<T>, LowCardinalitySelector<T>, MADThreshold<T>, MissingValueRatio<T>, MissingValueSelector<T>, SpectralClusteringFS<T>, VarianceInflationSelector<T>, VarianceThreshold<T>, VarianceThreshold<T>

For Beginners

If your data has 100 columns (features) and only 10 of them help predict the target, a feature selector finds those 10. SelectedIndices tells you which column numbers it kept, and GetSupportMask gives you one true/false flag per original column.

How It Works

A feature selector learns, during Fit, which input columns to keep, and its Transform returns only those columns. This interface exposes that choice so it can be inspected or reused, in the same way as scikit-learn’s SelectorMixin.get_support.

Properties

Property Summary
int[] SelectedIndices { get; } Gets the zero-based indices of the input columns the selector kept, or null before it has been fitted.

Methods

Method Summary
bool[] GetSupportMask() Returns one flag per input column of the data the selector was fitted on: true where the column is kept.