Skip to main content
[Preprint]. 2023 Jun 13:rs.3.rs-3012879. [Version 1] doi: 10.21203/rs.3.rs-3012879/v1

Table 4:

Parameters for fitting the gradient-boosted trees

Parameter(s) Tested values Best value

Learning rate 0.0125, 0.05, 0.2 0.0125
Maximum tree depth (max_depth) 3, 4, 5 3
Data subsampling ratio (subsample) 0.6, 0.8, 1 0.8
Minimum weight of a leaf node (min_child_weight) 1, 2, 4 1
L1 regularization (alpha) 0, 1, 2 2
L2 regularization (lambda) 1, 2, 4 1
Number of trees to fit per iteration (n_estimators) 1, 2, 4 4