kuda2026/-mundaai-

AI-powered crop risk intelligence for Zimbabwe's smallholder farmers AI4I Challenge 2026` # MundaAI 🌾 **MundaAI** is an AI-driven agricultural risk prediction platform built for smallholder farmers in Zimbabwe. By integrating predictive machine learning with high-accessibility communication channels (WhatsApp, USSD, Web), MundaAI delivers real-time crop risk assessments to help farmers mitigate the impacts of climate change and shifting market signals. Developed under the **POTRAZ AI4I Challenge 2026 (Development Track)**. --- ## 🛠️ Interactive Demo To view and test our live prediction pipeline, click the **"Open in Colab"** badge above or use the direct link below. The notebook features a **no-code interactive form** where reviewers can adjust parameters (rainfall, soil moisture, agro-ecological region) to watch the Random Forest classifier predict crop risk in real-time. 🔗 **Direct Colab Link:** MundaAI Interactive Demo --- ## 📂 Repository Architecture The project has been refactored into modular directories to support robust pipeline deployment: * **`api/`**: Contains the FastAPI backend code (`main.py`) serving prediction endpoints. * **`data/`**: Holds our agricultural, climate, and market data, including the baseline 360-record dataset. * **`docs/`**: Includes architectural diagrams, dataset overviews, and our formal proposal. * **`model/`**: Houses the model training history and the interactive demo: * `01_mundaai_training.ipynb`: Original training pipeline (Random Forest + CTGAN). * `02_mundaai_interactive_demo.ipynb`: Live demo notebook with interactive prediction sliders. * **`frontend/` / `whatsapp/` / `ussd/`**: Contain the respective codebase directories for user interfaces. --- ## 🚀 How to Run the API Locally 1. Clone this repository: ```bash git clone github.com