Table 5
Benchmark of different clustering algorithms separated by stellar group.
| D | N | H | C | V | S | ⋆ | |||||
|
|
|||||||||||
| All | 31 clusters (177 stars) | K-Means++ | 5 | – | 0.78 | 0.68 | 0.72 | 0.29 | 100% | ||
| Affinity prop. | 5 | 29 | 0.77 | 0.69 | 0.72 | 0.30 | 100% | ||||
| DBSCAN | 5 | 30 | 0.48 | 0.70 | 0.57 | 0.03 | 100% | ||||
| Mitschang (0.68/0) | 17* | 42 | 0.85 | 0.66 | 0.74 | 0.11 | 100% | ||||
| Mitschang (0.68/2) | 17* | 32 | 0.84 | 0.66 | 0.74 | 0.12 | 89% | ||||
| Mitschang (0.90/0) | 17* | 42 | 0.86 | 0.66 | 0.75 | 0.11 | 97% | ||||
| Mitschang (0.90/2) | 17* | 32 | 0.84 | 0.66 | 0.74 | 0.13 | 86% | ||||
|
|
|||||||||||
| Dwarfs | 11 clusters (60 stars) | K-Means++ | 5 | – | 0.83 | 0.63 | 0.72 | 0.32 | 100% | ||
| Affinity prop. | 5 | 13 | 0.85 | 0.61 | 0.71 | 0.31 | 100% | ||||
| DBSCAN | 5 | 13 | 0.59 | 0.69 | 0.64 | 0.01 | 100% | ||||
| Mitschang (0.68/0) | 17* | 11 | 0.82 | 0.59 | 0.69 | 0.07 | 98% | ||||
| Mitschang (0.68/2) | 17* | 10 | 0.81 | 0.59 | 0.68 | 0.11 | 95% | ||||
| Mitschang (0.90/0) | 17* | 11 | 0.84 | 0.59 | 0.69 | 0.09 | 90% | ||||
| Mitschang (0.90/2) | 17* | 10 | 0.83 | 0.58 | 0.68 | 0.13 | 87% | ||||
|
|
|||||||||||
| Giants | 26 clusters (117 stars) | K-Means++ | 5 | 31 | 0.81 | 0.74 | 0.78 | 0.28 | 100% | ||
| Affinity prop. | 5 | 24 | 0.79 | 0.74 | 0.76 | 0.28 | 100% | ||||
| DBSCAN | 5 | 28 | 0.66 | 0.83 | 0.74 | 0.06 | 100% | ||||
| Mitschang (0.68/0) | 17* | 31 | 0.84 | 0.70 | 0.77 | 0.12 | 100% | ||||
| Mitschang (0.68/2) | 17* | 21 | 0.82 | 0.71 | 0.76 | 0.14 | 83% | ||||
| Mitschang (0.90/0) | 17* | 31 | 0.84 | 0.70 | 0.77 | 0.12 | 100% | ||||
| Mitschang (0.90/2) | 17* | 21 | 0.82 | 0.71 | 0.76 | 0.14 | 83% | ||||
Notes.The number of dimension (D) refers to the number of principal components used, except those marked with The number of predicted clusters (N) is not reported for the K-Means++ algorithms because it is one of the parameters of the method and the real number was used. The V-Measure is calculated from the homogeneity (H) and completeness (C) values. The silhouette coefficient (S) and the number of classified stars (⋆) are reported in the last two columns.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.