Table 2. P-values of the significance test testing the difference of the AUC ROC values between each pair of prediction methods.
Method | Biehl et al. | Vilar | Mean/LR-LASSO | EDF-MSE/LR-LASSO | Mean/LDA |
Biehl et al. | – | – | – | – | – |
Vilar | 0.1598 | – | – | – | – |
Mean/LR-LASSO | 0.1537 | 0.6809 | – | – | – |
EDF-MSE/LR-LASSO | 0.2509 | 0.3486 | 0.6509 | – | – |
Mean/LDA | 0.0003 | 0.0004 | 0.0019 | 0.0040 | – |
The null hypothesis is that the AUC ROC values between two methods are the same. Given p-values are provided by the online ROC analysis tool StAR [14].