A Graph-based Hierarchical Clustering Algorithm for population studies of astrophysical objects

We present here a data exploration tool designed to enhance the study of astrophysical objects by integrating traditional hierarchical clustering with graph-based community detection algorithms. This new tool allows in-depth analysis of the distributions of observables across astrophysical catalogs. The method is first validated on the Iris benchmark dataset, where it accurately reproduces the known taxonomic classification and outperforms default implementations of several widely used clustering algorithms. We then showcase applications to the SWIFT catalog of gamma-ray bursts. Finally, we discuss the most representative features that characterise the identified subpopulations to allow for qualitative analysis and model development.