Clustering method based on point density analysis

This article discusses a new clustering method for a set of points based on an analysis of their density. Unlike the well-known DBSCAN method, in this method, the search for the optimal value of the radius of the circle, within which neighboring points are considered for each point to assign them to the same cluster, is performed computationally based on the initial data set. This makes it possible to exclude the selection of the radius value experimentally. A software implementation of the method is given. The results of a clustering study for several datasets with different point densities are presented.