ROC curves for the predictive performance of different cost sensitive classifiers.

<p>ROC plot depicts a relative trade-off between true positive rate and false positive rate of the predictions. The diagonal value represents a completely random guess. The corresponding scalar values of area under curve are given as auROC in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0097446#pone-0097446-t004" target="_blank">table 4</a>.