Combining density peaks clustering and gravitational search method to enhance data clustering

Data clustering is a valuable field for extracting effective information and hidden patterns from datasets. In this paper we propose a clustering approach based on density peaks clustering (DPC) and a modified gravitational search algorithm (GSA), called GSA-DPC. To take advantage of the distance measure and nearest neighbor rule among the data points, our method simultaneously combines the distance and density mechanisms. Based on the optimized cluster center set selected by DPC working with density measure, the best clustering distribution is achieved according to the distance criterion of GSA. We compare the performance of GSA-DPC with other well-known clustering approaches, including density-based spatial clustering of applications with noise (DBSCAN), density peaks clustering, K-Means, spectral clustering (SC), grey wolf optimizer for clustering (GWO-C), gravitational search algorithm for clustering (GSA-C) and data clustering algorithm based on GSA and K-Means (GSA-KM). The experimental results indicate that GSA-DPC outperforms these competing approaches.