Table 7. Selected hyperparameters for the high recall classifier.
The hyperparameters are selected based on the performance of the “Top 13” features described earlier.
Hyperparameter | Value |
---|---|
learning_rate | 0.025 |
max_depth | 9 |
min_child_weight | 6.441 |
gamma | 2.444 |
colsample_bytree | 0.873 |
num_boost_round | 307 |
reg_lambda | 5.897 |
scale_pos_weight | 8.198 |
subsample | 0.174 |