Table 2 The hyperparameter values of LightGBM optimized by MSFWOA

learning_rate

Boosting learning rate

0.2

(0.01, 0.3)

num_leaves

Maximum tree leaves for base learners

14

(3, 60)

max_depth

Maximum tree depth for base learners

15

(3, 25)

subsample

Subsample ratio of the training instance

0.9

(0.5, 1)

colsample_bytree

Subsample ratio for columns of the tree

0.7

(0.5, 1)

reg_alpha

L1 regularization term on weights

0.5

(0, 100)

reg_lambda

L2 regularization term on weights

3.4

(0, 100)