Fig 1.
The distribution of DCG in the dataset.
Table 1.
Feature name and description.
Table 2.
Binary confusion matrix.
Table 3.
Hyperparameters of machine learning model obtained by Bayesian optimization.
Fig 2.
Comparison of four classification models on test sets.
(a) Evaluation indicators, (b) ROC.
Fig 3.
Correlation heat map between features.
Fig 4.
Feature selection process (a) recursive elimination (b) exhaustion method.
Fig 5.
Characteristic summary of SHAP values.
(a) Average SHAP values (b) SHAP values for each sample.
Fig 6.
The SHAP dependence plot for (a) RDpr (b)SA (c) RDeapoinr (d) Lev (e) Top1 (f) (g) MshareBalance1 (h) ATO.
Table 4.
Comparison of raw data with adjusted data.