← back

ClusterLab

An interactive k-means clustering visualizer in Rust. Runs natively on desktop and in the browser via WebAssembly, animating every iteration so you can watch centroids move, point assignments flip, and Voronoi cells reshape — in 2D or 3D.

Features

• 2D and 3D modes, each fully independent — pan/zoom/click-to-add in 2D, orbit camera in 3D
• Voronoi visualization — edges in 2D, bisector planes in 3D
• Run controls: single step, auto-run, run-to-convergence, and an infinite loop mode
• Built-in datasets — random Gaussian blobs, two-moons, and the Fisher iris dataset (PCA-projected in 3D)
• Point editing in 2D — add, delete, or hand-place centroids
• Pure-logic core kept separate from rendering for testability

Engineering

89 tests across the k-means, geometry, and animation modules. CI runs fmt, clippy (deny warnings), and the full test suite, with coverage uploaded to Codecov. The codebase enforces a boundary where only the viz/app layers may import the rendering library.