scipy.stats.mstats.

kruskal#

scipy.stats.mstats.kruskal(*samples)[source]#

Compute the Kruskal-Wallis H-test for independent samples

Deprecated since version 2.0.0: scipy.stats.mstats.kruskal is deprecated as of SciPy 2.0.0 and will be removed, along with the scipy.stats.mstats namespace, in SciPy 2.4.0. For similar functionality, use scipy.stats.kruskal with MArray(s) instead of NumPy masked array(s).

Parameters:

*samplesarray_like: Two or more arrays with the sample measurements can be given as arguments.

Returns:

statisticfloat: The Kruskal-Wallis H statistic, corrected for ties
pvaluefloat: The p-value for the test using the assumption that H has a chi square distribution

Notes

For more details on kruskal, see scipy.stats.kruskal.

Examples

>>> from scipy.stats.mstats import kruskal

Random samples from three different brands of batteries were tested to see how long the charge lasted. Results were as follows:

>>> a = [6.3, 5.4, 5.7, 5.2, 5.0]
>>> b = [6.9, 7.0, 6.1, 7.9]
>>> c = [7.2, 6.9, 6.1, 6.5]

Test the hypothesis that the distribution functions for all of the brands’ durations are identical. Use 5% level of significance.

>>> kruskal(a, b, c)
KruskalResult(statistic=7.113812154696133, pvalue=0.028526948491942164)

The null hypothesis is rejected at the 5% level of significance because the returned p-value is less than the critical value of 5%.