Analysis results of miRNAs based on XGBoost algorithm.

(a) The trend of error rate, logistic loss, AUC and RMSE in the training and testing of miRNA expression data. Iteration steps (x-axis) as well as error rate, logistic loss, AUC and RMSE (y-axis in each of the four panels). Each panel has two lines representing the training set (blue) and test set (red). (b) Ranking of important miRNA candidates. miRNA candidates (x-axis) and the importance score of each miRNA candidate (y-axis), as determined through XGBoost algorithm, are shown.