Video tour
Watch the k-d tree work
Three guided walkthroughs showing how the explorer finds the nearest business, why insertion order matters, and how the lost C program was rebuilt from its test kit.
Journey 01
Find the nearest
Click anywhere on the Melbourne CBD map and watch the algorithm work. See the k-d tree descent, the pruning decisions, and the nearest businesses highlighted step by step.
What you'll see
- 1Click on the Melbourne CBD map
- 2Watch the k-d tree descent and pruning animation
- 3See the nearest businesses highlighted
Journey 02
Order matters
The same data, five different insertion orders. Compare tree depth, query costs, and bootstrap confidence intervals to see why median-first construction wins.
What you'll see
- 1Run random vs sorted vs median-based tree builds
- 2Compare tree depth and comparisons per query
- 3View bootstrap confidence intervals on the results
Journey 03
Under the hood
The C program was lost, so it was rebuilt from the test kit that survived. See the rebuild process, parity tests, and the 2026 web implementation.
What you'll see
- 1View the C source code in /rebuild
- 2Check the parity test results
- 3See the "rebuilt in 2026" labels
Try it yourself
The explorer is live at comp20003-kd-tree.vercel.app. Click anywhere on the map, run the benchmark suite, or explore the rebuild story.
Production note
These showcase videos are generated by Playwright scripts that drive the live app with deterministic seeds and on-screen step captions. The scripts live in web/e2e/showcase.spec.ts; the media conversion pipeline is in web/scripts/convert-showcase-media.mjs.