Detection of Open Clusters using Data Mining techinques

Abstract

With the arrival of the third Data Release from the Gaia mission, we receive new information that can be really helpful to detect new clusters in the Galactic disc. So far, cluster hunting methods used five parameters, which are position, parallax and proper motions of the stars, in order to identify clusters. The new release arrives with the mean radial velocity (RV ) measured for 33 million stars, which can be added as the sixth dimension in order to improve the efficiency of these methods. In this work, we implement a six-parameter detection method based on the work developed for previous releases. The method searches for clusters in a region of the sky with two input hyperparameters, the size of the box and the number of neighbour stars considered to form a cluster (L, minPts). We have run the algorithm for 81 different pairs, to determine which one performs better through several regions of the sky. The most efficient pair has been (L, minPts) = (13◦, 11), followed closely by (16◦, 12). Both had near 60% of the clusters found and 70% of correctly clustered stars while having a low number of field stars clustered, which means the results had lower noise.