Home / Computing / K-means clustering

K-means clustering

Vary cluster count and initial centers; watch assignments and center trails.

    CS12 · K-means clustering0.00 s · Ready

    Computing · AI foundations / Principles and exploration

    K-means clustering experiment guide

    Vary cluster count and initial centers; watch assignments and center trails.

    How does it work?

    Assignment and mean updates never increase squared error.

    Key relationship

    μⱼ = mean(xᵢ ∈ Cⱼ) · SSE = Σ‖xᵢ − μⱼ‖²

    Try this investigation

    1. Vary cluster count and initial centers; watch assignments and center trails.
    2. Run and record each of the three reference settings.
    3. Change one variable and compare with the reference records.

    Interpreting your results

    How is each center updated?

    Mean of assigned points · Assignment and mean updates never increase squared error.

    Assumptions and limitations

    Euclidean distances and seeded points are used. Empty clusters keep their centers. A local minimum is not a ground-truth class.