k-NN starts from labeled points in a coordinate plane. This lesson only fixes the exact training set so later distance calculations have a visible source.
highlighted = computed this step
Class labels live on points
The training set is a collection of exact integer points with class labels. For example, (0, 0) has class A and (3, 3) has class B.
training point (0,0)↦A
Why neighbors matter
The mechanism compares a query point to these labeled points. Nearby labels supply the vote once a query is displayed.
labels sit on exact coordinates
Summary
No query point has been classified yet. This step only fixes the displayed training positions that later arithmetic must reuse.