Figure 5

True positive rate versus false positive rate for different estimated posterior odds ratios. The true positive rate (TPR) versus false positive rate (FPR) is plotted for different values of the posterior odds ratio estimated for the dataset by five-fold cross-validation. As the posterior odds ratio increases, the false positive rate decreases and the ratio of the true positive rate divided by the false positive ratio increases. Thus, higher quality datasets can be generated by requiring higher posterior odds ratios. The TPR is calculated as the number of true positives predicted divided by the total number of positives in the test set. The FPR is calculated as the number of false positives predicted divided by the total number of negatives in the test set.