Critical values for the Nemenyi test

The critical distance for the Nemenyi test is calculated as:

$r_{\alpha,K,N} \approx q_{\alpha,K}\sqrt{\frac{K(K+1)}{6N}}$ ,

where $\alpha$ is the confidence level, $K$ is the number of models and $N$ is the number of measurements. To calculate $q_{\alpha,K}$ the Studentised range statistic for infinite degrees of freedom divided by $\sqrt{2}$ is used. The values of $q_{\alpha,K}$ for up to $K=100$ can be download here as csv.

You may also be interested in the implementations of the Nemenyi test in MatLab and R, or a discussion on how to best visualise its results and its connection with the MCB test.

Related Posts

Leave a Reply Cancel reply