Data Science for Business: What You Need to Know about Data Mining and Data-Analytic Thinking

by Foster Provost

Cosine distance is particularly useful when you want to ignore differences in scale across instances — technically, when you want to ignore the magnitude of the vectors.