If we have different number of samples in each classes, in which way the parameters will be arranged before applying Kruskal Wallis statistical test in matlab.
As far as I know, there is no problem using K–W with groups with different numbers of observations.
The significance of the the test is determined in the same way as other tests following the Null Hypothesis Significance Test framework.
The interpretation of significant K–W tests can be a little tricky. People will often report the result as the medians being different, but this is not a correct interpretation in many cases.
Null hypothesis and significance testing:
http://rcompanion.org/handbook/D_01.html
K–W test assumptions and interpretation:
http://rcompanion.org/handbook/F_08.html
More discussion on K–W test:
http://www.biostathandbook.com/kruskalwallis.html
K-W is a nonparametric test for the non-normal-distribited data. P values are important in interpreting group differences. The commonly used threshold value is 0.05.
- Badges
- Science topic