Abstract We introduce the application of the clustering algorithm to the preliminary compiled list of probable wide pairs of co-moving stars. The main aim of such development is a possibility of unsupervised blind search of coeval loose stellar structures over significant regions in space. Using Gaia EDR3 data, we investigated the application of the method to nearby region hosting recently discovered loose structures – tidal tails of Coma Ber star cluster and a nearby stellar group named Group X. We compare the results of straightforward clustering of stellar data with results of using our method with varying parameters. We then compare the results of our method to the recent results of the two groups of authors who independently discovered the discussed structures. We find parameters allowing the method to find the full scope of distributed stellar groups without preliminary knowledge of their characteristics. It decreases the risk of false positive clustering and improves the ability to discover loose stellar groups, in comparison with the application of clustering algorithm to the individual stars. Further we obtain a refined dataset of probable members of both stellar groups and independently obtain their ages (700 ± 70 Myr and 450 ± 100 Myr) and space velocities ((U,V,W) = (8.63 ± 0.13, 6.63 ± 0.20, 6.65 ± 0.95) km/s for Coma Ber star cluster, and (U,V,W) = (7.70 ± 0.12, 3.27 ± 0.45, 5.69 ± 0.80) km/s for Group X). Our results are in very good agreement with those of previous investigators.