TABLE 1.
Comparison of the performance of machine learning classification models based on two forms of molecular representation on the test set.
| Models | Accuracy | Precision | Recall | F1-score | AUC | |
|---|---|---|---|---|---|---|
| MACCS Fingerprint | RF | 0.913 | 0.930 | 0.898 | 0.913 | 0.975 |
| SVM | 0.908 | 0.929 | 0.887 | 0.907 | 0.973 | |
| ANN | 0.844 | 0.886 | 0.796 | 0.838 | 0.930 | |
| Voting | 0.873 | 0.923 | 0.818 | 0.868 | 0.968 | |
| Stacking | 0.884 | 0.925 | 0.841 | 0.881 | 0.970 | |
| RDKit Descriptor | RF | 0.942 | 0.954 | 0.932 | 0.943 | 0.979 |
| SVM | 0.936 | 0.953 | 0.921 | 0.936 | 0.977 | |
| ANN | 0.908 | 0.939 | 0.875 | 0.906 | 0.971 | |
| Voting | 0.930 | 0.952 | 0.909 | 0.930 | 0.980 | |
| Stacking | 0.931 | 0.942 | 0.921 | 0.931 | 0.982 |