An empirical study for density peak clustering

Density Peak Clustering (DPC) is one of the most effective density-based clustering algorithms due to its ability to detect arbitrary clusters while being robust to noise. Since the first introduction in 2014, it has been cited a thousand times. This paper presents a comprehensive analysis of the DPC algorithm's performance on some UCI and Gaussian data sets. These used data sets have different properties such as intersecting clusters, unbalanced data, or different densities, such that not many clustering algorithms can perform well. From the obtained results, we aim to evaluate advantages and disadvantages of the algorithm and propose some open research directions.