An overview on density peaks clustering.

• Analyze the theory of density peaks clustering (DPC) and its performance. • Summarizes the improvement of DPC and analyzes it with experimental data. • The related application research of improved DPC in different fields is introduced. Density peaks clustering (DPC) algorithm is a new algorithm based on density clustering analysis, which can quickly obtain the cluster centers by drawing the decision diagram by using the calculation of local density and relative distance. Without prior knowledge and iteration, the parameters and structure are simple and easy to implement. Since it was proposed in 2014, it has attracted a large number of researchers to explore experiments and improve applications in recent years. In this paper, we first analyze the theory of DPC and its performance advantages and disadvantages. Secondly, it summarizes the improvement of DPC in recent years, analyzes the improvement effect, and shows it with experimental data. The related application research of DPC in different fields is introduced. Finally, the clustering results of DPC, LCDPC, DCHDPC and NADPC algorithms in different data sets are analyzed. Experiments show that the improved algorithm can divide the clustering more accurately, which provides new ideas for improving DPC algorithm in the future. At the same time, this paper summarizes and prospects the improvement and development of DPC. [ABSTRACT FROM AUTHOR]