Since the focus in this paper is to predict and prevent fraudulent online transactions, it is not
adequate to evaluate algorithms only by the classification accuracy rate, training time and ROC value.
The ability to identify fraudulent online transactions should be given higher priority than the
classification accuracy rate. The ability to identify fraudulent transactions is usually being evaluated
using the Type I and Type II error rate. Type I error is defined as false positive error and Type II error
is defined as false negative error.
Figure 9 and 10 show the Type I and Type II error rate for the examined algorithms.