Explain how you would evaluate a classification model beyond accuracy.
Classic MLLevel: IntermediateClassical machine learning
Explain how you would evaluate a classification model beyond accuracy.
Hints
- State the task, inputs and prediction target.
- Compare with a simple baseline: assumptions, objective and independent validation.
Answer
Use a confusion matrix to see the breakdown of TP/FP/TN/FN, then compute precision, recall, F1, and ROC-AUC/PR-AUC for threshold-independent performance. For imbalanced data prefer PR-AUC over ROC-AUC, since ROC can look overly optimistic when negatives vastly outnumber positives. Also examine calibration (do predicted probabilities match observed frequencies?) if downstream decisions depend on probability estimates, not just class labels.
Source and licence: Ishan Dutta — Awesome Data Science Interview Q&A · MIT