Improving the open cluster census

The census of open clusters in the Milky Way is a never-before seen state flux. Recent works have reported hundreds new thanks to incredible astrometric quality Gaia satellite, but other also that many discovered pre era may be associations. We aim conduct comparison clustering algorithms used detect clusters, attempting statistically quantify their strengths and weaknesses by deriving sensitivity, specificity, precision each as well true positive rate against larger sample. selected DBSCAN, HDBSCAN, Gaussian mixture models for further study, owing speed appropriateness use with data. developed preprocessing pipeline data specific application clusters. derived detection rates all 1385 fields our study more detailed performance statistics 100 these DBSCAN was sensitive 50% 62% sample, generally very good specificity precision. HDBSCAN traded higher sensitivity up 82%, especially across different distances scales were slow only 33% which tended objects. Additionally, we report on 41 cluster candidates detected three are closer than 500 pc. When additional post-processing mitigate its false positives, found most effective algorithm recovering Our results suggest already yet