"You are who you're closest to." To classify a new point, find its k nearest neighbors in feature space and let them vote. Click anywhere on the canvas to query a new point.
k5
Vote at query—
Train accuracy (leave-one-out)—
5
k controls bias-variance: k=1 follows every quirk (low bias, high variance — jagged regions); k=25 averages broadly (smoother boundary, higher bias). The colored background shows the predicted class for every grid cell.