k-Nearest Neighbors (k-NN)

Classify a point by looking at the labels of its nearest examples under a chosen distance metric.

Math Foundation

Distance metrics such as euclidean, manhattan, and cosine distance.

Key Concepts

• k-NN is lazy: almost all work happens at prediction time
• The choice of k controls noise sensitivity versus smoothness
• Distance metrics encode what similar means for the problem
• High-dimensional spaces make distance less informative without careful feature design

Practice Path

• Vary k and watch the boundary become jagged or smooth
• Try Euclidean versus cosine distance
• Standardize features before comparing distances

Checkpoint

Why is k-NN sensitive to feature scaling, and what happens when k is too small or too large?

Archive Note

This card is backfilled from Sensei's public 30 AI Algorithms curriculum and the learning-progress signal that this algorithm had been delivered. Private progress, engagement analytics, delivery metadata, and local filesystem paths were intentionally excluded.

Prerequisites: support-vector-machine