Table 2.
Statistics of classification results for each class (source counts, classification and F1 scores, completeness and contamination rates) and their distribution with respect to the corresponding SOS modules (sources in ‘common’ between the ones classified as a given class and the corresponding SOS module, ‘extra’ sources in classification, sources classified as ‘other’ classes or ‘missed’ by classification).
| Class | NCLS | NSOS | Classification versus SOS module |
Classification score |
Compl. | Cont. | F1 | Max F1 | Min score | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Common | Extra | Other | Missed | Min | Avg | Max | (%) | (%) | (%) | (%) | (at max F1) | |||
| ACV|CP|...|SXARI (1) | 10 779 | < 54 476 | 0 | < 10 779 | 54 476 | 0 | 0.00 | 0.54 | 1.00 | 13.0 | 83.4 | 14.5 | 14.6 | 0.04 |
| ACYG (1,*) | 329 | < 54 476 | 5 | < 324 | 54 471 | 0 | 0.07 | 0.54 | 0.99 | < 89.1 | > 44.7 | < 68.3 | < 81.2 | 0.63 |
| AGN | 1 035 207 | 872 228 | 872 181 | 163 026 | 5 | 42 | 0.00 | 0.54 | 1.00 | 50.6 | > 0.1 | < 67.1 | < 67.1 | 0.00 |
| BCEP (1,*) | 1475 | < 54 476 | 174 | 1301 | 54 302 | 0 | 0.02 | 0.47 | 0.99 | < 78.8 | > 44.9 | < 64.9 | < 71.8 | 0.51 |
| BE|GCAS|SDOR|WR (1) | 8560 | < 54 476 | 223 | 8337 | 54 253 | 0 | 0.01 | 0.72 | 1.00 | 3.9 | 65.3 | 7.0 | 7.0 | 0.01 |
| CEP | 16 141 | 15 021 | 14 987 | 1154 | 34 | 0 | 0.00 | 0.48 | 1.00 | 75.2 | 4.5 | 84.1 | 84.1 | 0.00 |
| CV | 7306 | – | – | – | – | – | 0.01 | 0.59 | 1.00 | 15.0 | 47.3 | 23.4 | 23.7 | 0.20 |
| DSCT|GDOR|SXPHE (1) | 748 058 | < 54 476 | 53 640 | 694 418 | 836 | 0 | 0.00 | 0.49 | 1.00 | 44.6 | 43.2 | 50.0 | 51.0 | 0.15 |
| ECL | 2 184 356 | 2 184 477 | 2 184 337 | 19 | 60 | 80 | 0.00 | 0.54 | 1.00 | 47.8 | 4.7 | 63.7 | 63.7 | 0.00 |
| ELL (2) | 65 300 | 6306 | 6303 | 58 997 | 3 | 0 | 0.00 | 0.48 | 1.00 | 8.9 | 56.5 | 14.7 | 15.2 | 0.11 |
| EP (3) | 214 | 214 | 214 | 0 | 0 | 0 | 1.00 | 1.00 | 1.00 | (10.2) | (4.7) | (18.4) | (18.4) | 0.00 |
| LPV | 2 325 775 | 1 720 588 | 1 720 066 | 605 709 | 521 | 1 | 0.00 | 0.58 | 1.00 | 54.3 | 4.3 | 69.3 | 69.3 | 0.11 |
| MICROLENSING (4,*) | 254 | 363 | 187 | 67 | 11 | 165 | 0.00 | 0.43 | 0.96 | < 56.6 | > 7.2 | < 70.3 | < 71.1 | 0.22 |
| RCB (*) | 153 | – | – | – | – | – | 0.03 | 0.48 | 0.97 | < 68.6 | > 52.9 | < 55.8 | < 63.2 | 0.30 |
| RR | 297 778 | 271 779 | 271 576 | 26 202 | 0 | 203 | 0.00 | 0.52 | 1.00 | 70.0 | 2.6 | 81.4 | 81.4 | 0.00 |
| RS | 742 263 | – | – | – | – | – | 0.00 | 0.41 | 1.00 | 30.0 | 73.3 | 28.2 | 28.2 | 0.00 |
| S (4) | 512 005 | 471 679 | 499 | 511 506 | 285 249 | 185 931 | 0.00 | 0.59 | 1.00 | (30.5) | (95.2) | (8.3) | (17.8) | (0.87) |
| SDB (*) | 893 | – | – | – | – | – | 0.17 | 0.73 | 0.99 | < 76.1 | > 28.2 | < 73.9 | < 73.9 | 0.00 |
| SN | 3029 | – | – | – | – | – | 0.00 | 0.51 | 1.00 | 46.0 | 6.0 | 61.8 | 61.8 | 0.00 |
| SOLAR_LIKE (4,5) | 1 934 844 | 474 026 | 102 573 | 1 832 271 | 21 575 | 349 878 | 0.00 | 0.39 | 1.00 | 16.7 | 6.3 | 28.4 | 28.4 | 0.00 |
| SPB (1,*) | 1228 | < 54 476 | 434 | 794 | 54 042 | 0 | 0.02 | 0.49 | 1.00 | < 56.4 | > 76.1 | < 33.6 | < 41.8 | 0.50 |
| SYST (*) | 649 | – | – | – | – | – | 0.00 | 0.50 | 1.00 | < 81.7 | > 51.3 | < 61.0 | < 61.8 | 0.08 |
| WD (*) | 910 | – | – | – | – | – | 0.25 | 0.67 | 1.00 | < 15.6 | > 13.6 | < 26.4 | < 26.5 | 0.30 |
| YSO | 79 375 | – | – | – | – | – | 0.00 | 0.52 | 1.00 | 9.1 | < 79.8 | 12.5 | 12.5 | 0.01 |
| GALAXY (6) | 2 451 364 | – | – | – | – | – | 0.00 | 0.74 | 1.00 | 55.5 | 0.2 | 71.4 | 71.4 | 0.00 |
| Total | 12 428 245 | |||||||||||||
Notes. The class group ACV|CP|MCP|ROAM|ROAP|SXARI was abbreviated as ACV|CP|...|SXARI. Dashes indicate cases when columns are not applicable (absence of SOS modules). In total, 9 976 881 sources from 24 variability types (or type groups) and 2 451 364 galaxies are published in the vari_classifier_result and galaxy_candidates tables, respectively. Estimates of completeness and contamination rates generally exclude the training sources of the corresponding class, with the exception of the class labels followed by an asterisk. For the explanation of the values within parentheses, see the sub-sections on the relevant classes in Sect. 4. See Table 3 for further details on completeness and contamination. The following notes refer to specific classes as indicated next to their label. (1) ACV|CP|MCP|ROAM|ROAP|SXARI, ACYG, BCEP, BE|GCAS|SDOR|WR, DSCT|GDOR|SXPHE, and SPB classifications were considered as input for the upper main-sequence oscillator SOS module, but the latter did not split its 54 476 candidates into separate classes. (2) ELL classifications were used as input for the compact companion SOS module, but the latter did not target typical members of this class. (3) EP classifications were used as input for the planetary transit SOS module, but only SOS candidates were retained in the classification results. (4) MICROLENSING, S, and SOLAR_LIKE classifications were not used as input for the respective SOS modules. (5) SOLAR_LIKE classifications included more types (flaring stars) than the rotational modulation SOS module. (6) GALAXY classifications were made available exclusively in the galaxy_candidates table. (*) Less than 12 sources classified as a given class were matched with the literature (Gavras et al. 2023) after the removal of training set objects (aimed at reducing biases in the estimates of completeness and contamination rates). For these classes, training sources were included in the assessment of completeness and contamination (for upper or lower limits, respectively).
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.