Package elki.datasource.filter.normalization.instancewise

Class HellingerHistogramNormalization<V extends NumberVector>

  • Type Parameters:
    V - vector type
    All Implemented Interfaces:
    BundleStreamSource, Normalization<V>, ObjectFilter, StreamFilter
    public class HellingerHistogramNormalization<V extends NumberVector>
extends AbstractVectorStreamConversionFilter<V,​V>
implements Normalization<V>
    Normalize histograms by scaling them to unit absolute sum, then taking the square root of the absolute value in each attribute, times the normalization constant \(1/\sqrt{2}\).

    \[ H(x_i)=\tfrac{\sqrt{|x_i|/\Sigma}}{\sqrt{2}} \quad\text{ with } \Sigma=\sum\nolimits_i |x_i| \]

    Using Euclidean distance (linear kernel) and this transformation is the same as using Hellinger distance: HellingerDistance

    Since:
    0.7.0
    Author:
    Erich Schubert