Classical models built from the math you already know
Two questions have been sitting unanswered since k-Nearest-Neighbours entered the course, first for regression and then for classification. First: nothing so far explained how to actually pick k, only that it behaves like the same capacity dial met throughout this course. Second, and more surprising: the whole idea of "nearest" quietly assumes that distance is a meaningful way to compare points. That assumption holds up fine with two or three features. It does not hold up forever.
🔒 This is a Pro lesson — the interactive figure, worked examples, quiz and practice open with Pro access.
▶ Choosing k & the Curse of Dimensionality