Modelling techniques

Unsupervised models interview questions

Without labels there’s no score to hide behind, so unsupervised questions test judgement. If you segment customers with k-means, how do you choose the number of clusters, and how would you know the segments are useful rather than just tidy?

Interviewers also ask when k-means is the wrong tool, what PCA keeps and what it throws away, and how you’d find anomalies before you know what an anomaly looks like. A strong answer starts from what the output will be used for, because that’s the only real test of an unsupervised model.

Try one now

Standard

You're using k-means to segment customers. How do you choose k, and when would k-means be the wrong tool?

Free, no sign-up. The interviewer asks follow-ups on your reasoning, then rates your answer.

What the 9 questions cover

  • Anomaly detection4
  • K-means clustering3
  • Hierarchical clustering2
  • PCA2
  • DBSCAN1
  • Gaussian mixture models1
  • t-SNE and UMAP1

A question can cover more than one topic, so the counts add up to more than 9.

How the practice works

  1. The AI interviewer asks the question and follows up on your answer, the way an interviewer would.
  2. Stuck? Ask for a hint. Each one gives a little more away, so you can get unstuck without being handed the answer.
  3. Get a written debrief. Four ratings, each from Needs work to Interview-ready, tied to what you actually said, and what to study next.

More modelling techniques questions

See all categories