I've recently been using a package known as 'iai' for interpretable machine learning.
However, after going through their user guides to obtain a plotted AUC curve, I wondered if it was possible to construct a bootstrapped confidence interval around the AUC.
I've since tried the standard commands in the pROC package, such as 'ci(roc)' and ci.auc(roc). But unfortunately, I've had no success.
I wondered therefore if I had my AUC score (single value), where the object 'grid' is my decision tree learner.
iai::score(grid, train_X, train_y, criterion = "auc")
Further, the AUC is plotted using the following argument:
iai::roc_curve(grid, test_X, test_y, positive_label = 1)
I wondered if anyone can think with these limited arguments on how to construct confidence intervals for the AUC?
Would be appreciated.