Dataset

Preset Add class (click canvas)

k-NN

k = 5 Distance metric

Euclidean (L2) Manhattan (L1) Chebyshev (L∞)

Distance-weighted vote (1/d²) Render decision regions

Stats

Train points0

Classes used0

MetricEuclidean

LOOCV accuracy—

Query class—

Neighbour votes—

How to use. Pick a class, click on the canvas to add training points. Hover anywhere to query a test point — the k nearest neighbours light up and vote. Try k=1 vs k=30 to see overfitting vs underfitting. Switch the metric to reshape the boundary.