Open Access
Table 1.
Hyperparameters of the XGBoost classifier and corresponding final scores, for the nuclear and broad models.
| Name | Value | |
|---|---|---|
| Nuclear | Broad | |
| Manually set | ||
| max_depth | 3 | 3 |
| Optimized for F2 score | ||
| n_estimators | 200 | 200 |
| subsample | 0.8 | 0.8 |
| reg_lambda | 1 | 1 |
| reg_alpha | 1 | 1 |
| learning_rate | 0.06 | 0.06 |
| colsample_bytree | 1.0 | 1.0 |
| min_child_weight | 5 | 5 |
| Final scores | ||
| F2 score | 0.37 | 0.25 |
| F1 score | 0.20 | 0.13 |
| Precision | 0.12 | 0.07 |
| Recall | 0.76 | 0.74 |
Notes. The hyperparameters were optimized by a stratified fivefold cross-validation routine, while the final scores were derived by the leave-one-out method.
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.