Skip to main content

Table 2 The value process of MGA

From: Prediction of postoperative infectious complications in elderly patients with colorectal cancer: a study based on improved machine learning

Parameter adjustment order

Parameter name

Parameter adjustment standard

max_depth, min_child_weight

Choose the best two groups

gamma

Choose the best two groups

subsample, colsample_bytree

Choose the best two groups

reg_alpha、reg_lambda

Choose the best one groups

learning_rate

Choose the best one groups

num_boost_round

The selection of num _ boost _ round is to make training _ AUC > 0.9995