Table 1.
Summary of literature review on K-means hybridization with metaheuristic algorithms.
Fig 1.
Proposed hybrid SOSK-means clustering algorithm.
Table 2.
Initial parameter setting for classical SOS, classical K-means and proposed hybrid SOSK-means.
Table 3.
Initial parameter setting for the compared algorithms.
Table 4.
Dataset characteristics.
Table 5.
SOSK-means results in over forty independent runs with DB and CS validity indices as the fitness function.
Fig 2.
The mean run time achieved by SOSK-means on DB and CS measures over forty independent runs for the twelve datasets.
Table 6.
SOSK-means compared with SOS and K-means for forty replications.
Table 7.
SOSK-means results compared with results from existing algorithms in the literature.
Table 8.
The Friedman means rank test results for the SOS, K-means and hybrid SOSK-means algorithms.
Table 9.
Wilcoxon rank-sum test for equal medians showing corresponding p-values.
Fig 3.
Clustering illustration of hybrid SOSK-means for the listed datasets using DB-Index.
Fig 4.
Clustering illustration of hybrid SOSK-means for the listed datasets using CS-Index.