ANTHEM: ASSESSING THE EXPOSURE TO AIR POLLUTANTS IN AN URBAN ENVIRONMENT

This repository contains the code and resources for the ANTHEM project, a research initiative focused on understanding and mitigating exposure to air pollutants in urban environments. The project combines geospatial data analysis, machine learning, and an interactive web application to provide a comprehensive tool for researchers and the public.

Project Overview

The ANTHEM project aims to:

• Develop a high-resolution model of CO₂ concentrations in an urban environment using a variety of data sources, including sensor data, OpenStreetMap, and weather data.
• Create an interactive web application that allows users to visualize CO₂ concentrations, plan routes with minimal exposure, and understand the factors that contribute to air pollution.
• Provide a platform for researchers to explore the relationship between urban form, human activity, and air quality.

Directory Structure

The project is organized into the following main directories:

• app/: Contains the source code for the Streamlit web application.
• data/: Contains the raw and processed data used in the project. (Note: This directory is in .gitignore to protect sensitive data).
• notebooks/: Contains Jupyter Notebooks for data exploration, feature engineering, and model development.
• src/: Contains the source code for the data processing, modeling, and visualization pipelines.
• output/: Contains the output of the data processing and modeling pipelines, including predictions and plots.
• models/: Contains the trained machine learning models.

Main Pipeline

The core of the project is a data processing and modeling pipeline that generates the CO₂ predictions used in the web application. The pipeline consists of the following main steps:

1. Data Ingestion and Preprocessing: Raw data from various sources is loaded, cleaned, and preprocessed.
2. Feature Engineering: A rich set of features is engineered from OpenStreetMap data, weather data, and other sources.
3. Model Training and Optimization: A sophisticated ensemble model is trained and optimized to predict CO₂ concentrations.
4. Prediction and Visualization: The trained model is used to generate high-resolution predictions of CO₂ concentrations, which are then visualized on a map.

Usage

The project has two main components that can be run independently:

1. Web Application

The interactive web application can be run using Streamlit:

streamlit run app/app.py

2. Data Processing and Modeling Pipeline

The data processing and modeling pipeline can be run from the command line:

# To generate the street network predictions
python src/visualization/street_graph_prediction.py

# To run the ensemble model optimization
python src/ensemble/main.py

Contributing

1. Clone the repository
2. Make sure you have all the dependencies installed (see requirements.txt)
3. Create a new branch from develop (e.g., feature/manova or feature/pca)
4. Make your changes
5. If you install a new package, run pip freeze > requirements.txt to update the requirements file
6. Commit and Push your changes
7. Open a Pull Request to merge your branch into develop

Data Privacy

Please do not upload any sensitive data to this repository. Every contributor is working under a Non-Disclosure Agreement (NDA) and should not share any data outside the project. For this reason, the data/ folder is included in the .gitignore file.

To download the data, please refer to the shared OneDrive folder or ask the project manager for the data files.

We cannot test code using remote GitHub Actions or cloud-based services that might expose the data. All CI/CD pipelines must be run locally. If you want to approve pull requests into the develop branch, please use the act tool and ensure that the scripts work locally before merging.

License

This project is licensed under the MIT License. See the LICENSE file for details.