🌿 SCoPE-Net

Scope 3 Embodied Carbon & Supply Chain Transparency Platform for Sime Darby Property — Team HUNE, Sime Darby Property University Hackathon (Township of Tomorrow).

Features

• Ingestion & NLP/OCR simulation – regex extraction + fuzzy matching of messy invoice/BOM lines to emission factors (tCO₂e/unit).
• Carbon engine – emissions = quantity × factor.
• Anomaly screening – Isolation Forest + per-material z-score + unmatched-material rule.
• CRUD – create (sidebar form), read (township / vendor tier / status filters), update (e.g. clear a flag to Verified), delete.
• Compliance – IFRS S1/S2 and NSRF requirement status from Dataset 3 with evidence, CSV export.

Architecture

app.py    Streamlit UI (tabs, CSS, state)
utils.py  loading, OCR/NLP, carbon engine, ML, CRUD, compliance
data/     three master CSVs (see data/README.md); demo data auto-generated if absent

State lives in st.session_state; swap in a database for persistence in production.

Setup

python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
# copy the 3 CSVs into data/
streamlit run app.py

Google Colab: pip install -r requirements.txt, then run Streamlit via a localtunnel.

Notes

Compliance statuses are heuristic readiness indicators, not assurance. Edits are in-memory (export CSV to keep them).