friedmanPost: Friedman's post hoc raw p-values

In scmamp: Statistical Comparison of Multiple Algorithms in Multiple Problems

Description

This function computes the raw p-values for the post hoc based on Friedman's test.

Usage

friedmanPost(data, control = NULL, ...)

Arguments

data	Data set (matrix or data.frame) to apply the test. The column names are taken as the groups and the values in the matrix are the samples.
control	Either the number or the name of the column for the control algorithm. If this parameter is not provided, the all vs all comparison is performed.
...	Not used.

Details

The test has been implemented according to the version in Demsar (2006), page 12.

Value

A matrix with all the pairwise raw p-values (all vs. all or all vs. control).

References

J. Demsar (2006) Statistical Comparisons of Classifiers over Multiple Data Sets. Journal of Machine Learning Research, 7, 1-30.

Examples

data(data_gh_2008)
friedmanPost(data.gh.2008)
friedmanPost(data.gh.2008, control=1)

Example output

                   C4.5   k-NN(k=1)   NaiveBayes       Kernel         CN2
C4.5                 NA 0.004848763 8.064959e-01 4.486991e-08 0.012763008
k-NN(k=1)  4.848763e-03          NA 1.011233e-02 7.963489e-03 0.743971478
NaiveBayes 8.064959e-01 0.010112334           NA 1.736118e-07 0.024744672
Kernel     4.486991e-08 0.007963489 1.736118e-07           NA 0.002880485
CN2        1.276301e-02 0.743971478 2.474467e-02 2.880485e-03          NA
     C4.5   k-NN(k=1) NaiveBayes       Kernel        CN2
[1,]   NA 0.004848763  0.8064959 4.486991e-08 0.01276301

scmamp documentation built on May 1, 2019, 10:10 p.m.