The 60-second answer
Start from geometry and product needs: expected cluster shape, whether K is known, noise/outlier behavior, scale, and whether a hierarchy matters. K-Means is fast for roughly spherical Euclidean clusters; hierarchical methods expose nested structure; DBSCAN handles arbitrary dense regions and noise but is sensitive to density parameters.
Build the answer in this order
Start from geometry and product needs: expected cluster shape, whether K is known, noise/outlier behavior, scale, and whether a hierarchy matters.
K-Means is fast for roughly spherical Euclidean clusters; hierarchical methods expose nested structure; DBSCAN handles arbitrary dense regions and noise but is sensitive to density parameters.
Standardize/transform features and choose a distance metric that reflects domain similarity before comparing algorithms.
Use intrinsic metrics cautiously and validate cluster stability, interpretability, and downstream utility with domain review.
A useful interview mental model
This is the shape of a strong answer—not a script to memorize.
Senior-level signal
- Senior answers discuss high-dimensional distance concentration and why embedding quality may dominate clustering choice.
- Include operational concerns: incremental updates, cluster-ID stability, and how new points are assigned.
What the interviewer is really testing
Likely follow-up questions
Common weak-answer patterns
- Reciting a definition without mechanism or assumptions.
- Claiming one technique is always better without a data regime.
- Stopping before failure modes, validation, or deployment implications.