Identifying cluster members of NGC 869 and NGC 884 using a combined machine learning method

Abstract

Abstract This work describes a method for calculating cluster membership probability from big, high dimensional newly released GAIA-EDR3 observations that combine the Gaussian Mixture Model (GMM) with the Random Forest (RF). This technique has the considerable advantage of rapidly identifying possible cluster members in big data sets using modest training samples. The result indicates that the double clusters NGC 869 has 20270 points ultra-high (p≥ 0.8) probable members, 9294 are high (p≥ 0.6) probable members, and 6181 stars are moderate (0.2 <= p < 0.6) members. NGC 884 has ultra-high (p≥ 0.8) probable members of 21317, high (p≥ 0.6) probable members with 10871 and moderate (0.2 <= p < 0.6) of 5998. Cluster centers of NGC 884 and NGC 869 were also determined to be at α = 2h 22m 17s, δ = 57h 8m 60s and α = 2h 19m 2s, δ = 57h 7m 48s respectively. The double cluster also has a mean proper motion of -0.690 ± 12 mas/yr, or -1.150 ± 16 mas/yr, and a mean parallax of 0.410 ± 04 mas. The old cluster member in NGC 884 was quantified to be ± 621, and the new cluster member to be ± 1129. The new members of NGC 869 are ± 1511, up from the previous 920 members. We also found that the combined GMM and RF are effective in identifying cluster membership using the GAIA-EDR3 observations.