Figure 2.
Feature weights and contributions to the models: (A) logistic regression with LASSO; (B) decision tree; (C) random forest; and (D) gradient-boosting tree. The beta coefficient in logistic regression with LASSO and the importance of variables scaled by the maximum value of 100 in the decision tree, random forest, and gradient-boosting tree were shown. LASSO, Absolute shrinkage and selection operator; [], reference variable.