Datasets
Curated entries for mlearning.studio. This collection points at sources, live demos, and benchmark paths — it does not ship multi‑GB data dumps.
| Dataset | Path | Focus |
|---|---|---|
| GTFS | GTFS/ |
Public transit feeds |
| ↳ Madrid | madrid-gtfs/ |
Transit GTFS → reactive .me graph; Smart Cities demos + Zenodo / benchmark links |
| BigDaMa | BigDaMa/ |
Raha/Baran data-cleaning benchmarks (TU Berlin) |
| ↳ Hospital | hospital/ |
Dirty/clean pair → .me DataQuality checks (link + sha256) |
Principles
- Link, don’t dump — Zenodo / upstream packages stay the source of truth for scale‑1 / scale‑10 / scale‑100 CHANGE archives.
- Live demos first — tiny synthetic subsets run in the browser on the published
.mekernel. - Benchmark path — reproducible workloads live under
.me’sTypescript/tests/Benchmarks/(e.g. GTFS‑Madrid; Hospital / DataQuality‑Hospital when published).
Part of mlearning.studio · neurons.me · GitHub