Figure 1.

Smoothed ROC curves in test sample derived from multiple splitting of training and test samples. Graphs depict 40 randomly selected ROC curves out of 1000 splits. AUC is the mean area under the ROC curve from 1000 splits (95% confidence interval). FPR is false positive rate (one minus specificity) and TPR is true positive rate (sensitivity).

Baker and Kramer BMC Bioinformatics 2006 7:407   doi:10.1186/1471-2105-7-407
Download authors' original image