Interactive Learning Lab

K-Nearest Neighbors Visualizer

Place points, compare neighbors, and learn how each setting changes the classifier behavior.

Enhanced Mode

K = 3Dataset: Simple BinaryTest Points: 0

Click the canvas to place a test point.

Place your first test point

Click anywhere on the chart to run a prediction.

How KNN Works

KNN classifies each test point by checking the 3 nearest training points and taking a vote. In enhanced mode, the connections animate so the decision path is easier to follow.

Click to add and classify a test point.

Higher confidence means a cleaner local neighborhood.

Change K to compare smooth vs sensitive behavior.

Try boundary areas where classes overlap.

Learning Notes

Basics

Training points are known labeled examples.
Test points are clicks you place on the canvas.
Dashed or animated links show nearest neighbors.

Distance Types

Euclidean: direct straight-line distance between two points.
Manhattan: block-style distance using horizontal + vertical steps.

Confidence and Weighting

Confidence estimates how strongly neighbors agree.
Weighted KNN gives closer points more influence.

Core Controls

K Value: 3

115

Lower values react to local noise. Higher values smooth the decision.

Distance Metric

Euclidean uses straight lines. Manhattan uses grid-like steps.

Use weighted KNNShow confidence

Dataset

Two clearly separated classes

Tip: place points near class boundaries to see uncertain predictions.

Core Controls

K Value: 3

115

Lower values react to local noise. Higher values smooth the decision.

Distance Metric

Euclidean uses straight lines. Manhattan uses grid-like steps.

Use weighted KNNShow confidence

Dataset

Two clearly separated classes

Tip: place points near class boundaries to see uncertain predictions.

Advanced Options

Show decision boundaryAnimate distancesGame mode

Legend

Class A

Class B

Class C

Class D

Class E

Selected Neighbor

Distance Link