Enhancement of Kernel Clustering Based on Pigeon Optimization Algorithm

Clustering is one of the essential branches of data mining, which has numerous practical uses in real-time applications.The Kernel K-means method (KK-means) is an extended operative clustering algorithm. However, this algorithm entirely dependent on the kernel function's hyper-parameter. Techniques that adequately explore the search spaces are needed for real optimization problems and to get optimal answers. This paper proposes an enhanced kernel K-means clustering by employing a pigeon optimization algorithm in clustering. The suggested algorithm finds the best solution by tuning the kernel function's hyper-parameter and alters the number of clusters simultaneously. Based on five biological and chemical datasets, the results acquired the potential result from the suggested algorithm that is compared to other approaches based on intra-cluster distances and the Rand index. Moreover, findings confirm that the suggested KK-means algorithm achieves the best computation time. The proposed algorithm achieves the necessary support for data clustering.