Documentation
Every project document, rendered for reading in the browser. The markdown files in the repository remain the source of truth — these pages are regenerated from them on every build.
Start here
| Document | What it covers |
|---|---|
| End-to-End Project Report | The standalone account: how the handbooks were reverse-engineered into a relational database, faculty-by-faculty findings, the justified final dataset, and the DuckDB/explorer design |
| User Manual | For reviewers and deans: what the dataset contains, the ideal-student definition, validation results, caveats, worked examples (Word copy for circulation) |
Method & process
| Document | What it covers |
|---|---|
| Replication Log & Hazard Catalogue | The authoritative process log: pipeline details, per-edition layout contracts, hazards H1–H41, onboarding procedure for new years and faculties |
| Final-Clean Dataset Method | How discrepancies are resolved: the R0/R3/R1/R2/R4 rule order, the adjudication registers, worked examples |
| Adjudication Register Guide | Schema and workflow for recording case-by-case adjudications |
| Original Design Document | The Commerce review and schema proposal the project began from |
Repository guides
| Document | What it covers |
|---|---|
| Repository README | Project overview, quick start, layout, status |
| Analysis Layer & Explorer | The DuckDB semantic layer, its views, the fee deflator, and how to run this site |
| Extractors Guide | The shared engine, per-faculty configs, and bespoke parsers |
| Development TODO | Open work, in value order — including the adjudication review workflow |