# Izere
**Mobile Money history → lender-ready credit decision.**
Rwandan micro and small businesses generate real revenue through MoMo but lack the
documentation banks require — part of a ~$1.8B SME financing gap. Izere turns a MoMo
statement into structured books and a transparent, explainable credit assessment that a
bank, MFI, or SACCO can act on.
## What it does
1. **Ingests** a MoMo statement CSV (drag-and-drop, or use the built-in samples).
2. **Verifies integrity before anything else** — balance-chain arithmetic, duplicate
transaction IDs, date order. A doctored statement is caught to the exact row and no
score is issued (try the "Tampered statement" sample).
3. **Builds the books** — monthly income, expenses, net, selling days, top customers,
and an interactive cash-flow chart (hover any month for its full numbers). Owner
cash top-ups are never counted as revenue.
4. **Scores** — five explainable factors (inflow, regularity, balance floor, volatility,
expense ratio) blended into a 0–100 score with plain-language reasons. Businesses with
under 3 months of history are "not scoreable yet", never falsely declined.
5. **Sizes the loan** — safe monthly repayment capped at 30% of net inflow, recommended
limit of 2.5 months of profit, a loan-request checker with counter-offers, and a
stress test showing whether the decision survives a revenue drop.
6. **Exports** a clean two-page report (print to PDF).
Fully bilingual: **Kinyarwanda and English**, with contextual help on every concept so
non-technical users can navigate it.
## Run it
```bash
npm install
npm run dev # open
localhost
npm test # 13 unit tests: integrity, revenue rules, verdict bands, loan matching
npm run build # production build
```
## Sample data
| File | What it shows |
| --- | --- |
| `public/samples/sample_healthy.csv` | Kigali retail shop, 6 months → score 94, Approve |
| `public/samples/sample_seasonal.csv` | Seasonal trader → Review (irregular income) |
| `publ …