Kristen3381/Phenominals-Kenya-AI

# Shamba Score — Farmer Reliability Scoring Tool > Kenya AI Challenge — Mercy Corps AgriFin Track > **Team:** Phenominals Kenya AI --- ## What This Is Smallholder farmers in Kenya have no verifiable record of their selling activity, so lenders treat them as too risky for credit — even reliable sellers get rejected for lacking a paper trail. **Shamba Score** solves this by modeling a farmer's marketplace transaction history (who they sold to, how often, whether buyers return) as a relationship graph in Neo4j, and computing a transparent, explainable reliability score from it. New farmers with thin history are flagged "insufficient data," not scored low. --- ## Current Status (June 2026) Working frontend demo with mock data. Backend (Node/Express + Neo4j) being built in parallel. ### What's Built | Screen | Description | |--------|-------------| | **Farmer Dashboard** | Searchable list of all farmers with score rings, location, transaction stats | | **Farmer Detail** | Score gauge, evidence breakdown (txns, repeat/unique buyers, recency), consistency note | | **Transaction Graph** | SVG node-link diagram showing farmer → buyer relationships, repeat buyers highlighted green | ### Demo Data 5 sample farmers with varied profiles: - **Grace Mwangi** — Score 72, 18 transactions over 4 months - **David Omondi** — Score 88, 35 transactions, weekly deliveries for 8 months - **Amina Hassan** — Score 45, 7 transactions, limited history - **Samuel Kiprop** — Flagged **Insufficient Data** (1 transaction, cold-start case) - **Njoki Wanjiku** — Score 94, 52 transactions, year-round supplier --- ## Tech Stack - **Frontend:** React (Vite), React Router v7 - **Styling:** Tailwind CSS v4 - **Backend:** Node/Express + Neo4j (in progress — frontend uses mock data) - **Theme:** Dark forest green (#1E5631), mid green accent (#27AE60), light green bg (#D5F5E3) --- ## Project Structure ``` src/ ├── main.jsx # Entry point, BrowserRouter setup ├── App.jsx …