Cluster Validity Analysis of Alternative Results from Multi-Objective Optimization
Yimin Liu, Tansel Özyer, Reda Alhajj, Ken Barker · 2005
This paper investigates validity analysis of alternative clustering results obtained using the algorithm named Multi-objective K-Means Genetic Algorithm (MOKGA). The reported results are promising. MOKGA gives the optimal number of clusters as a solution set. The achieved clustering results are then analyzed and validated under several cluster validity techniques proposed in the literature. The optimal clusters are ranked for each validity index. The approach is tested by conducting experiments using three well-known data sets. The obtained results for each dataset are compared with those reported in the literature to demonstrate the applicability and effectiveness of the proposed approach.