When is Wasserstein distance appropriate for drift?
hardAnswer
- Wasserstein-1 = 'earth mover's distance' — minimal work to move mass from P to Q.
- Advantages: (1) sensitive to shifts (unlike KL that focuses on mode overlap).
- (2) Works even when supports differ.
- (3) Metric (triangle inequality).
- (4) Meaningful units (same scale as feature).
- Use for continuous features with ordinal/spatial meaning (prices, temperatures).
- Drawback: expensive in high-D — use sliced Wasserstein or 1D-per-feature.
Check yourself — multiple choice
- Random
- Earth-mover's distance; sensitive to shifts even with different supports; metric with meaningful units; use for ordinal/spatial features; sliced version for high-D
- Same as KL
- Not real
Wasserstein: EMD; metric with meaningful units; sensitive to shifts.
#monitoring#data-quality
Practise MLOps & Data Quality
215 interview questions in this topic.
Related questions
- What is data drift and how do you detect it?
- How is concept drift different from data drift?
- What is schema drift and how do you detect it?
- What is label drift and why does it matter?
- How is Population Stability Index (PSI) computed and interpreted?
- Why prefer Jensen-Shannon Divergence over KL for drift?