Package elki.math.statistics.distribution.estimator.meta

Class WinsorizingEstimator<D extends Distribution>

  • Type Parameters:
    D - Distribution type
    All Implemented Interfaces:
    DistributionEstimator<D>
    @Reference(authors="C. Hastings, F. Mosteller, J. W. Tukey, C. P. Winsor",
           title="Low moments for small samples: a comparative study of order statistics",
           booktitle="The Annals of Mathematical Statistics, 18(3)",
           url="https://doi.org/10.1214/aoms/1177730388",
           bibkey="doi:10.1214/aoms/1177730388")
public class WinsorizingEstimator<D extends Distribution>
extends java.lang.Object
implements DistributionEstimator<D>
    Winsorizing or Georgization estimator. Similar to trimming, this is supposed to be more robust to outliers. However, instead of removing the extreme values, they are instead replaced with the cutoff value. This keeps the quantity of the data the same, and will have a lower impact on variance and similar measures.

    Reference:

    C. Hastings, F. Mosteller, J. W. Tukey, C. P. Winsor
    Low moments for small samples: a comparative study of order statistics.
    The Annals of Mathematical Statistics, 18(3) *

    Since:
    0.6.0
    Author:
    Erich Schubert

    • Constructor Detail

      • WinsorizingEstimator

        public WinsorizingEstimator​(DistributionEstimator<D> inner,
                            double winsorize)
        Constructor.
        Parameters:
        inner - Inner estimator.
        winsorize - Winsorize parameter.