Fig. 3From: Construction and validation of a cuproptosis-related diagnostic gene signature for atrial fibrillation based on ensemble learningFeature selection with RF and XGBoost algorithms. (A) Relationship between the error rate and the number of classification trees. The error rate is minimum when ntree = 105. (B) Gini-importance of the 12 cuproptosis-related genes. (C) Relative Gain-importance of the 12 cuproptosis-related genes. (D) Venn plot demonstrating two features shared by RF and XGBoost algorithmsBack to article page