Machine Learning

KNN: How Many Neighbors Matter?

Adjust K and watch the decision boundary transform. Click anywhere on the plot to add a test point and see its K nearest neighbors highlighted.

Add a query point

K range explored: 0/10

Class 0

Class 1

Query point

K (number of neighbors)K = 5

✓ Balanced boundary

Training vs Validation Accuracy (Computed Live)

Training Validation│ current K

Both curves are computed from the actual points: training accuracy scores the visible points against themselves (at K=1 each point finds itself, so it is always 100%), validation accuracy scores 50 held-out points drawn from the same two clusters.

Click to Select Query Point

Click on the scatter plot to add a test point. Its K nearest neighbors will highlight with circles.

← Previous Guide Next Guide →