Evaluation interview questions
Evaluation questions are where interviewers find out whether you can connect a metric to a decision. Precision, recall, ROC AUC and log loss each answer a different question, and picking one means knowing what a false positive and a false negative cost.
Expect to be asked what you’d compare a model against (a simple baseline often settles more arguments than a better AUC), how you’d choose a classification threshold, why a model that ranks well can still give badly calibrated probabilities, and which regression metric to report when a few large errors matter more than many small ones.
Try one now
An airline's operations team wants a model to predict which flights will leave late. A colleague says a few simple rules would do. How do you decide?
Free, no sign-up. The interviewer asks follow-ups on your reasoning, then rates your answer.
What the 23 questions cover
- Choosing a classification threshold9
- Regression metrics (MAE, RMSE, MAPE, R²)6
- Calibration6
- Precision and recall5
- ROC AUC5
- Baseline models4
- Precision-recall curve4
- Confusion matrix2
- Log loss / cross-entropy2
- Lift and gain charts2
- KS statistic2
- Gini coefficient1
A question can cover more than one topic, so the counts add up to more than 23.
How the practice works
- The AI interviewer asks the question and follows up on your answer, the way an interviewer would.
- Stuck? Ask for a hint. Each one gives a little more away, so you can get unstuck without being handed the answer.
- Get a written debrief. Four ratings, each from Needs work to Interview-ready, tied to what you actually said, and what to study next.