KruskalWallisTest
javanpst.tests.equality.kruskalWallisTest

Class KruskalWallisTest



  • public class KruskalWallisTestextends StatisticalTest
    The Kruskal-Wallis test. This test is the natural extension of the two-sample Wilcoxon test for location to the case of k independent samples. Assuming the location model, the null hypothesis is that the k medians of the populations are equal, whereas the alternative is that at least two location parameter differ. Ties are broken by the midrank method (e.g. a tie in ranks 6 and 7 is assumed as a 6.5 rank for both values). A multiple-comparisons procedure (Bonferroni-Dunn), applied at 0.10 and 0.05 levels of significance is also provided.
    • Constructor Detail

      • KruskalWallisTest

        public KruskalWallisTest()
        Default builder
      • KruskalWallisTest

        public KruskalWallisTest(DataTable newData)
        Builder. Load data to test.
        Parameters:
        newData - data to test.
    • Method Detail

      • setData

        public void setData(DataTable newData)
        Sets data to test
        Parameters:
        newData - data to test.
      • getH

        public double getH()
        Get H statistic
        Returns:
        H statistic
      • getPValue

        public double getPValue()
        Get p-value
        Returns:
        p-value computed
      • getAvgRanks

        public double getAvgRanks(int pop)
        Get average rank of the selected population
        Parameters:
        pop - population index (from 1 to k)
        Returns:
        average rank of the population
      • printMultipleComparisonsProcedureReport

        public String printMultipleComparisonsProcedureReport()
        Prints a report with the results of the multiple comparisons procedure
        Returns:
        Output report

SCaVis 1.7 © jWork.org