jakeschege/kenya-rhomis-finance

A system meant to act as a credit scoring assistant # Kenya RHoMIS Finance Kenya smallholder credit scoring and farmer segmentation web app. ## Start **Terminal 1 — API (FastAPI)** ```bash cd api uvicorn main:app --reload --port 8000 ``` **Terminal 2 — Frontend (Next.js)** ```bash cd frontend npm run dev ``` App: localhost API docs: localhost ## Add Groq key (optional — for conversational intake + narratives) Edit `frontend/.env.local`: ``` GROQ_API_KEY=gsk_... NEXT_PUBLIC_GROQ_ENABLED=true ``` ## Pages | Route | Description | |---|---| | `/` | Overview — stats, segment cards, training countries | | `/score` | Manual multi-step scoring form | | `/intake` | Conversational intake (requires Groq key) | | `/segments` | Detailed Kenya segment profiles | | `/model` | Model comparison table — all results, feature list | ## API endpoints | Endpoint | Description | |---|---| | `POST /score` | Score a household from raw inputs | | `GET /metadata` | Full model metadata (features, segments, results) | | `GET /health` | Which models loaded successfully | ## Artifacts used Models loaded from `../artifacts_kenya/`: - **Primary model**: Random Forest (443 trees, 27 features) - **Comparison**: XGBoost, Logistic Regression, Decision Tree - **Segmentation**: K-Means k=4 (14 features, Kenya only) - **Excluded** (NumPy BitGenerator mismatch): GB, Stacking models