For classical univariate ROC analysis,
The AUC in the table below is 0.837, but when you create a curve with “ROC curve View”, the AUC becomes 0.845.
The same goes for ROC curve based model evaluation (tester).
Why is this and what should I trust?