How do you choose the number of clusters k?
easyAnswer
- Use multiple signals, not one: the elbow method on inertia (look for the bend), the silhouette score (higher is better, up to 1), the gap statistic, and domain knowledge.
- For density-based clustering (DBSCAN, HDBSCAN) you don't set k at all — you set density parameters instead.
Check yourself — multiple choice
- The elbow method alone always gives the answer
- Silhouette + elbow + domain knowledge combined
- Always pick k = sqrt(n)
- DBSCAN requires you to pick k
No single metric is perfect — combine elbow, silhouette, and business context.
#clustering#evaluation
Practise Unsupervised Learning
214 interview questions in this topic.