Package org.moeaframework.util.statistics

Class KruskalWallisTest

java.lang.Object
org.moeaframework.util.statistics.OrdinalStatisticalTest
org.moeaframework.util.statistics.KruskalWallisTest
All Implemented Interfaces:
StatisticalTest
public class KruskalWallisTest extends OrdinalStatisticalTest
The Kruskal-Wallis One-Way Analysis of Variance by Ranks is a non-parametric statistical test determining if (at least) two out of K >= 2 populations have differing medians.
  • Null Hypothesis: All populations have equal medians.
  • Alternative Hypothesis: Not all populations have equal medians.

Assumptions:

  1. Samples are randomly selected from their corresponding populations
  2. Samples are independent
  3. The dependent variable (value being sampled) is continuous
  4. The underlying distributions of the populations are identical in shape

References:

  1. Kruskal, W.H. and Wallis W.A. "Use of Ranks in One-Criterion Variance Analysis." Journal of the American Statistical Association, 47(260):583-621, 1952.
  2. Sheskin, D.J. "Handbook of Parametric and Nonparametric Statistical Procedures, Third Edition." Chapman & Hall/CRC. 2004.

  • Constructor Details

    • KruskalWallisTest

      public KruskalWallisTest(int numberOfGroups)
      Constructs a Kruskal-Wallis test with the specified number of groups.
      Parameters:
      numberOfGroups - the number of groups being tested

  • Method Details

    • add

      public void add(double value, int group)
      Description copied from class: OrdinalStatisticalTest
      Adds a new observation with the specified value and group.
      Overrides:
      add in class OrdinalStatisticalTest
      Parameters:
      value - the value of the new observation
      group - the group to which the new observation belongs

    • addAll

      public void addAll(double[] values, int group)
      Description copied from class: OrdinalStatisticalTest
      Adds several new observations to the specified group.
      Overrides:
      addAll in class OrdinalStatisticalTest
      Parameters:
      values - the values of the new observations
      group - the group to which the new observations belong

    • test

      public boolean test(double alpha)
      Description copied from interface: StatisticalTest
      Returns true if the null hypothesis is rejected; false otherwise. The meaning of the null hypothesis and alternative hypothesis depends on the specific test.

      The prespecified level of confidence, alpha, can be used for either one-tailed or two-tailed (directional or non-directional) distributions, depending on the specific test. Some tests may only support specific values for alpha.

      Parameters:
      alpha - the prespecified level of confidence
      Returns:
      true if the null hypothesis is rejected; false otherwise