README.md

---
title: ClimaIQ Kisan
emoji: 🌾
colorFrom: green
colorTo: gray
sdk: streamlit
app_file: app.py
pinned: false
license: apache-2.0
---

# ClimaIQ Kisan

Climate-adjusted credit scoring for Indian agricultural lending: **Kisan view**, **loan officer** scorecard and PDF-ready report, and **portfolio stress** scenarios. Bilingual (English / Hindi) narratives via **Google Gemma (cloud)** or **local Ollama**.

## Dataset (in this Space)

`data/climaiq_india_agricultural_credit_1000.csv` — 1,000 synthetic loans (Maharashtra / Punjab, crops, SPI, rainfall deficit, engineered features, default label, ClimaIQ score and risk band), aligned with the same generator and model as the app.

## Optional secret

Add **`GEMMA_API_KEY`** (Google AI Studio) under Space **Settings → Secrets** for cloud Gemma. Without it, open the app and use **Local (Ollama)** only when you run Streamlit on your own machine; the hosted Space still runs the **ClimaIQ model** and charts without the key.

## Local run

```bash
pip install -r requirements.txt
streamlit run app.py
```