Interview: 'no labels — how do you know your clustering is any good?'

hard

Answer

$(1) Internal indices: silhouette + Calinski-Harabasz + Davies-Bouldin across k.$
$(2) Stability: bootstrap Jaccard, ARI between init runs.$
$(3) Downstream utility: does adding cluster label improve any adjacent supervised task?$
$(4) Domain sanity: do clusters agree with known heuristics?$
$(5) Visual: 2D projection colored by cluster — coherent groups?$
$(6) Business sanity: are cluster profiles actionable + interpretable?$
$Reality: unsupervised eval is always partial — combine 3-4 signals + stakeholder buy-in.$

Check yourself — multiple choice

Impossible
Internal indices (silhouette / CH / DB) + bootstrap stability + downstream utility + domain sanity + 2D projection + business interpretability — always partial, combine signals
Random
Only silhouette

No-label eval: multiple internal indices + stability + downstream + domain + viz.

#interview#clustering#evaluation

214 interview questions in this topic.