🌎 MVP3 — Climate–Economic Risk Platform (Brazil)

Integrating Climate Data + Machine Learning Socioeconomic Models for State-Level Vulnerability Assessment

OpenWeather Challenge Edition

---

🔍 Overview

The MVP3 – Climate-Economic Risk Platform is an interactive application that integrates real-time climate data from OpenWeather with a Machine Learning–based socioeconomic vulnerability model to generate a combined climate-economic risk index for each Brazilian state (UF).

This project was developed as part of the OpenWeather Challenge, focusing on originality, impact, and technical transparency.

Access the live application here:
👉 https://mvp3-climate-economic-risk-brazil.streamlit.app/

---

📌 Key Features

• Forecast-based climate risk (heat, rain & floods, wind, and air quality)
• Socioeconomic ML model trained on:
  • GDP per capita
  • Total GDP
  • Population
  • Density
  • Additional derived indicators
• Integrated risk index (60% climate + 40% socioeconomic)
• Explainability with SHAP (global + local importance)
• Interactive dashboards
• Dual-language interface (PT/EN)
• OpenWeather API integration
• Deployed on Streamlit Cloud

---

📂 Project Structure

mvp3-climate-economic-risk-brazil/ │ ├── data/ │ └── mvp3_dataset.csv # ML dataset │ ├── models_ml/ │ ├── socio_model.pkl # Trained ML model │ └── socio_shap_values.pkl # SHAP explainability │ ├── app_streamlit_mvp3.py # Main application ├── climate_risk_engine.py # Climate risk computation ├── socioecon_risk_engine.py # ML preprocessing ├── combined_risk_engine.py # Climate + socioeconomic combination ├── risk_engine.py # High-level orchestration ├── openweather_client.py # API consumption ├── weather_maps.py # Weather Maps integration (OpenWeather) ├── theme.py # UI theme configuration ├── requirements.txt # Dependencies └── README.md

---

⚠️ Important — API Key Requirement

To run this project locally, you must create a .env file in the root directory containing:

👉 You can obtain a free API key at
https://home.openweathermap.org/api_keys

The .env file is intentionally excluded from the repository
(via .gitignore) to protect private credentials.

---

🛠️ Installation

1. Clone the repository

git clone https://github.com/marconiv/mvp3-climate-economic-risk-brazil.git
cd mvp3-climate-economic-risk-brazil

### 2. Create your .env
OPENWEATHER_KEY=YOUR_PERSONAL_KEY

3. Install dependencies
pip install -r requirements.txt

4. Run the app
streamlit run app_streamlit_mvp3.py

🧠 Machine Learning Model (Socioeconomic Component)

Model type: Random Forest Regressor

Task: Predict socioeconomic vulnerability (0–100)

Input variables:

GDP per capita

Total GDP

Population

Density

Output:

Vulnerability score

Category label (Baixo / Moderado / Alto / Crítico)

Explainability:

SHAP summary plot

State-level local SHAP values

☁️ OpenWeather Integrations

Current Weather

5-day / 3-hour Forecast

Air Pollution API

Weather Maps (Tile layers)
All climate indices are normalized between 0 and 100.

🎯 Purpose & Impact

This platform provides a first-of-its-kind integrated view of climate and socioeconomic vulnerability for Brazil, enabling:

Public sector decision-making

Contingency planning

Investment prioritization

Early risk identification

Socio-climatic resilience insights

🌐 Live Demo

💻 Streamlit Cloud:
👉 https://mvp3-climate-economic-risk-brazil.streamlit.app/

📝 License

This project is released under the MIT License.

🙌 Acknowledgments

OpenWeather — APIs and Challenge platform

Scikit-learn — Machine learning

SHAP — AI explainability

Streamlit — Application framework

✨ About the Author

Marconi Fábio Vieira
InfoChoice Tecnologia Ltda
40+ years of experience in IT, AI, Data Science & Project Management
Brazil 🌎

Post about the app: https://infochoice.com.br/site/index.php/2025/11/21/mvp3-climate-economic-risk-platform-a-new-integrated-view-of-state-level-vulnerability-in-brazil/