Open Science & Reproducibility — Practice Guide

Reproducibility is a credibility multiplier and increasingly a funder/journal requirement. AEC's reliance on simulations, bespoke datasets and proprietary tools makes this especially important — and often neglected.

1. FAIR data principles

• Findable — deposit with a persistent identifier (DOI) and rich metadata; index it.
• Accessible — retrievable via a standard protocol; state access conditions (open vs controlled).
• Interoperable — use standard formats/vocabularies (open formats, documented schemas; for built-env data consider IFC, CityGML, standard CSV + data dictionary).
• Reusable — clear license, provenance, and enough documentation to reuse correctly.

FAIR ≠ "open": sensitive/proprietary data (site, personal, commercial) can be FAIR yet access-controlled.

2. Data repositories

• General: Zenodo (CERN; DOIs, GitHub integration), Figshare, Dryad, OSF, Harvard Dataverse.
• Institutional repositories; domain repositories where they exist.
• Choose one that issues a DOI, supports your license, and offers versioning. Avoid sharing data only as "available on request" — it usually isn't.