Estimating the cluster tree of a density by analyzing the minimal spanning tree of a sample

We present runt pruning, a new clustering method that attempts to find modes of a density by analyzing the minimal spanning tree of a sample. The method exploits the connection between the minimal spanning tree and nearest neighbor density estimation. It does not rely on assumptions about the specific form of the data density (e.g., normal mixture) or about the geometric shapes of the clusters, and is computationally feasible for large data sets.