Table 5 Parameter setting of the second layer model.
From: Emerging infectious disease surveillance using a hierarchical diagnosis model and the Knox algorithm
Model | Parameters | Explanations | Range | Step length |
|---|---|---|---|---|
LightGBM | n_estimators = 150 | The maximum number of iterations | 100–200 | 20 |
max depth = 8 | Maximum depth of tree | 5–15 | 1 | |
num leaves = 50 | Number of leaves of the tree | 10–100 | 10 | |
learning rate = 0.1 | Learning rate | [0.01,0.1,1] | ||
XGBoost | n_estimators = 180, | The maximum number of iterations | 100–200 | 20 |
max depth = 6 | Maximum depth of tree | 5–15 | 1 | |
colsample bytree = 0.6 | Proportion of features used by each tree to all features | 0.5–0.8 | 0.1 | |
learning rate = 0.1 | Learning rate | [0.01,0.1,1] | ||
Random Forest | n_estimators = 120 | The maximum number of iterations | 100–200 | 20 |
max depth = 13, | Maximum depth of tree | 5–20 | 1 | |
min_samples_leaf = 5 | Minimum number of samples required at leaf node | 1–15 | 2 | |
max features = 0.7 | Number of features in the selected feature subset | 0.5–0.8 | 0.1 |
