Open Access

Table 1

Features used for the machine learning classifier.

Name Description Importance
pm Estimated proper motion 0.01
e_pm Statistical error of estimated proper motion 0.17
N Number of epochs with detections 0.10
corr_30 Pearson correlation coefficient of the coordinates vs time for all points within 3″ from expected position at different epochs 0.23
corr Pearson correlation coefficient of the coordinates vs time for all points within 3σ brightness dependent coordinate uncertainty from expected position at different epochs 0.16
magstd Actual RMS of individual magnitude measurements in W2 band 0.14
magerr Mean error of individual magnitude measurements in W2 band 0.06
mag Mean magnitude in W2 band 0.05
magchi2 Reduced χ2 of individual magnitude measurements in W2 band 0.02
b Galactic latitude 0.01
NNNN1 Number of detections with 1 or more catalog points not belonging to the track within its 3σ positional uncertainty 0.02
NNNN3 the same for 3 or more points 0.01
NNNN2 the same for 2 or more points 0.01
NNNN4 the same for 4 or more points 0.001
NNNN5 the same for 5 or more points 0.001

Notes. The classifier using these features is described in Section 2.1. The last column shows the relative importance of the features for the classification at the final iteration.

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.