In the experiments, each non-deterministic method was executed a 100 times and statistics such as minimum, mean, and standard deviation were collected for the effectiveness criteria. In each run, the number of clusters (K ) was set equal to the number of classes (K′), as commonly seen in the related literature