Table 1.
The result of AUC (Area under Curve) and test accuracy for all models based on 71 nucleoside derivatives
| Models | Features | Test Accuracy | AUC | ||
|---|---|---|---|---|---|
| Mean | SEM | Mean | SEM | ||
| DT* | Descriptor_4175 | 0.65 | 0.01 | 0.65 | 0.02 |
| LR | Descriptor_4175 | 0.65 | 0.02 | 0.67 | 0.02 |
| RF | Descriptor_4175 | 0.63 | 0.01 | 0.72 | 0.02 |
| XGBoost | Descriptor_4175 | 0.63 | 0.01 | 0.69 | 0.02 |
| DT | Descriptor_144 | 0.64 | 0.01 | 0.64 | 0.01 |
| LR | Descriptor_144 | 0.68 | 0.02 | 0.80 | 0.02 |
| RF | Descriptor_144 | 0.67 | 0.01 | 0.75 | 0.02 |
| XGBoost | Descriptor_144 | 0.64 | 0.02 | 0.72 | 0.02 |
| DT | Descriptor_40 | 0.66 | 0.02 | 0.69 | 0.02 |
| LR | Descriptor_40 | 0.70 | 0.01 | 0.81 | 0.02 |
| RF | Descriptor_40 | 0.67 | 0.01 | 0.74 | 0.02 |
| XGBoost | Descriptor_40 | 0.65 | 0.01 | 0.75 | 0.02 |
| DT | Descriptor_ REF # | 0.59 | 0.02 | 0.63 | 0.02 |
| LR | Descriptor_ REF # | 0.71 | 0.01 | 0.84 | 0.02 |
| RF | Descriptor_ REF # | 0.67 | 0.01 | 0.75 | 0.02 |
| XGBoost | Descriptor_ REF # | 0.70 | 0.02 | 0.79 | 0.02 |
*LR Logistic regression, DT Decision tree, RF Random forest, XGBoost Extreme gradient boosting, SEM Standard error of the mean.
#Descriptors-REF: Recursive feature elimination (REF) has different optimal descriptors for different Algorithms: LR, n = 24; XGBoost, n = 16; DT, n = 30; RF, n = 37.