Compare ROC-AUC vs PR-AUC
Company: Databricks
Role: Data Scientist
Category: Machine Learning
Difficulty: easy
Interview Round: Technical Screen
Explain the difference between **ROC-AUC** and **PR-AUC** (Average Precision / area under the Precision–Recall curve).
In your answer, cover:
1. What each curve plots and the key formulas (TPR/FPR vs Precision/Recall).
2. How class imbalance (rare positives) affects each metric and its baseline.
3. Practical tradeoffs: when you would prefer ROC-AUC vs PR-AUC for model evaluation/selection.
4. Common pitfalls (e.g., interpretability, threshold choice, calibration, and how AUC relates to ranking).
Quick Answer: This question evaluates understanding of classification performance metrics by comparing ROC-AUC and PR-AUC, specifically what each curve plots and their core formulas (TPR/FPR vs Precision/Recall) and how class imbalance and baseline rates affect each metric.