Cosine similarity vs Euclidean — when do you use cosine?

medium

Answer

$y\; / \;( \mid \mid x \mid \mid \; \mid \mid y \mid \mid )$
$Use when magnitude is irrelevant: TF-IDF vectors (doc length varies), sentence embeddings, user-item preferences.$
$Use Euclidean when magnitude matters (physical measurements, absolute values).$
$In practice: L2-normalize embeddings then use Euclidean \to equivalent to cosine but works in standard k-means / FAISS indexes.$

Check yourself — multiple choice

Same thing
Cosine: angle only (magnitude-invariant); Euclidean: absolute distance. L2-normalize then Euclidean ≡ cosine — enables k-means / FAISS
Random
Only for text

Cosine: angle-only; Euclidean after L2-norm ≡ cosine (usable in k-means).

#clustering#similarity

214 interview questions in this topic.