Table 3 Parameter selection and optimization
From: Exploratory study on classification of diabetes mellitus through a combined Random Forest Classifier
Classification algorithm | Parameters | Parameter selection range | Final value |
|---|---|---|---|
LR* | L | L1, L2 | L2 |
C | 1 to 10, step size 0.01 | 3.95 | |
SVM | Gamma | 0 to 1, step size 0.01 | 0.06 |
C | 1 to 10, step size 1 | 1.00 | |
Kernel | linear, rbf, sigmoid | rbf | |
GBDT | Learning-rate | 0 to 1, step size 0.01 | 0.06 |
n_estimators | 10 to500, step size 10 | 60 | |
RF* | n_estimators | 10 to 500, step size 10 | 130 |
max_depth | 1 to15, step size 1 | 2 | |
max_features | 1 to 15, step size 1 | 2 |