In: Statistics and Probability
What insight we get from EER, and the AUC? Explain as detailed as possible.
Consider the above graph to understand EER and AUC:
The AUC (Area under the curve) is simply the area under the ROC curve which is in the graph above. The more the AUC, it is a better model. AUC actually measures the degree of separability of the classes or how able is the model to distinguish between the different classes. For example, higher the AUC, better the model is to distinguish between patients with cancer and no cancer. When the AUC is 0.5, it means that the model can't distinguish between the classes. If AUC = 0.8, it means that there is 80% chance that the model will be able to distinguish between the classes.
EER (Equal Error Rate) is the point where the False Positive Rate and False Negative Rate are equal. The model is good if the EER is low. EER is vastly used in the biometric field. In the biometric field, EER is used to find the threshold values of false acceptance rate (Proportion of number of times the system grants permission to unauthorised person) and false rejection rate (Proportion of number of times the system denies permission to authorised person). When the rates are equal, it is the equal error rate. The lower the equal rate value, the higher the accuracy of the biometric system.