Spot the Metric That Does Not Evaluate a Classifier

EasyClassical ML

Reported in interviews at Proxify

Problem statement

You are reviewing an evaluation report for a binary classifier that flags fraudulent transactions. A teammate listed four quantities they plan to report. Three of them measure how well a classification model performs; one does not belong in a classifier evaluation at all.

Which one is NOT a metric for evaluating a classification model's performance?

A. Matthews correlation coefficient

B. Logarithmic loss

C. Cook's distance

D. Lift curve

Show hint

What family of evaluation does each name belong to: threshold metrics, probability metrics, ranking metrics, or something else?

Relevant roles: ML Engineer, Applied Scientist and Data Scientist.

Related free problems