Table 4 Lists the final random forest hyperparameters selected.
From: Efficient feature ranked hybrid framework for android Iot malware detection
Hyperparameter | Selected Value | Description |
|---|---|---|
n_estimators | 100 | Selected for balance between stability and speed |
max_depth | 20 | Prevents overfitting while keeping model expressive |
min_samples_split | 2 | Best-performing value during tuning |
min_samples_leaf | 1 | Ensures deeper tree splits when needed |
max_features | sqrt | Standard RF practice; best CV performance |
Bootstrap | gini | Outperformed entropy in cross-validation |
Criterion | true | Ensures variance reduction |
class_weight | balanced” (if used) | Improves performance on slightly imbalanced datasets |
