Table 4
Comparison of DBNets performance with other mass inference methods through the two metrics lmse and r2-score.
| Method | lmse | r2-score |
|---|---|---|
| Random guess | 12.3 | −22 |
| Mean guess | 0.53 | 0 |
| Empirical formula (entire dataset) | 0.13 | 0.77 |
| Empirical formula (without outliers) | 0.08 | 0.85 |
| DBNets | 0.016 | 0.97 |
Notes. Empirical formulae refer to the empirical relation, Eq. (21), between the gap width and the disc and planet parameters. In the other two methods, the planet mass is predicted with a random value in the dataset’s planet mass range (random guess) and with the mean of the target masses (mean guess). These methods serve as references to help interpret the metrics’ values.
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.