Table 6 XGBoost classification algorithm hyperparameters and hyperparameter ranges used in grid-search tuning
From: Predicting chemotherapy response using a variational autoencoder approach
Hyperparameter name | Hyperparameter description | Hyperparameter range |
|---|---|---|
n_estimators | Number of trees to fit | (1, 2, 3, \(\ldots\), 40) |
max_depth | Maximum tree depth | (1, 2, 3, \(\ldots\), 10) |
learning_rate | Boosting learning rate | (0.05, 0.1, 0.2, 0.4, 0.6, 0.8) |
min_child_weight | Minimum sum of instance weight needed in a child | (1, 2, 3, \(\ldots\), 10) |
subsample | Sub-sample ratio of the training instance | (0.1, 0.2, 0.3, \(\ldots\), 1.0) |
colsample_bytree | Sub-sample ratio of columns when constructing each tree | (0.1, 0.2, 0.3, \(\ldots\), 1.0) |
reg_alpha | Coefficient of L1 regularization for the node weights | (0, 1, 2, 3) |
reg_lambda | Coefficient of L2 regularization for the node weights | (1, 2, \(\ldots\), 100) |