Why does K-Means fail on non-spherical clusters?

Updated May 15, 2026

Short answer

K-Means assumes spherical clusters, so it fails on complex shapes.

Deep explanation

It minimizes distance to centroids, which cannot represent curved or density-based clusters.

Real-world example

Failing to cluster crescent-shaped customer groups.

Common mistakes

  • Using KMeans for all clustering problems.

Follow-up questions

  • What is better alternative?
  • Why does shape matter?

More Unsupervised Learning interview questions

View all →