Table 4 The ranges of hyperparameters used as Grid search
From: Establishment of prediction model for mortality risk of pancreatic cancer: a retrospective study
Algorithm | Ranges of hyperparameters |
|---|---|
ANN | Learning rate [0.3-1], maximum epoch [100–1000], number of hidden layers [6–30] |
Bagging | Base classifier [J-48, Rep-tree, Random-tree], number of iterations [10–50], calculate out of bag [false, true] |
DT | Confidence factor [0.15–0.3], binary splitting [false, true], minimum number of objects [1–3] |
RF | Maximum depth [6–15], number of estimators [5–20], maximum number of features [5–10], maximum leaf nodes [1–4] |
SVM | RBF gamma [0.3-1], Control parameter (C) [5–30], regression precision [0.1–0.5] |
XG-Boost | Maximum depth [8–20], eta [0.1–0.5] |