Check your understanding
Quiz
You have learned to classify into more than two classes and measure it fairly, rebalance a rare class without leaking, set the decision threshold by what errors cost, read ROC, PR, lift and gains curves, calibrate predicted probabilities, and see what a single AUC hides. This quiz checks what stuck. The last two steps are live R you can run.
Question 1 of 6
The accuracy trap
A fraud model where 1% of cases are fraud reports 99% accuracy. The likely explanation is:
Correct: with 1% positives, always guessing the majority scores 99% while catching zero fraud.
The imbalance is what makes accuracy misleading here.