Free Access
Table A.3.
Feature importances for the classification of 4FGL-DR2 sources for the RF and BDT algorithms.
| Feature | RF: 50, 6 | BDT: 100, 2 |
|---|---|---|
| ln(LP_SigCurv) | 0.297 | 0.465 |
| LP_beta | 0.151 | 0.109 |
| ln(Variability_Index) | 0.085 | 0.253 |
| ln(Unc_Energy_Flux100) | 0.081 | 0.059 |
| ln(Energy_Flux100) | 0.076 | 0.008 |
| HR56 | 0.071 | 0.015 |
| Unc_LP_Index | 0.067 | 0.009 |
| HR34 | 0.035 | 0.005 |
| ln(Pivot_Energy) | 0.031 | 0.006 |
| HR23 | 0.025 | 0.005 |
| LP_Index | 0.015 | 0.016 |
| HR67 | 0.015 | 0.010 |
| HR45 | 0.015 | 0.006 |
| GLON | 0.013 | 0.017 |
| HR12 | 0.009 | 0.004 |
| GLAT | 0.007 | 0.003 |
Notes. RF algorithm: 50 trees with maximal depth 6; BDT algorithm: 100 trees with maximal depth 2. The features are ordered by decreasing importance in the case of the RF algorithm.
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.