Home

For the GMU Systematic AQ Study, the materials can be found across two platforms: the website at and the GitHub.

This platform contains detailed documentation and explanations of the study. You’ll find descriptions of the project’s objectives, methodologies, and results. This includes sections covering the various deep learning architectures, datasets used, and an overview of the tools required. Both platforms complement each other, with the website offering guidance and GitHub providing the technical resources required to conduct and reproduce the study.

Climate change, urbanization, and energy consumption are contributing factors to worsening air pollution and related health issues (Masood, 2021). Monitoring air quality (AQ) is vital, especially in urban areas where pollution tends to be higher. Traditional AQ monitoring systems are accurate but often limited by high costs and geographic reach. In contrast, low-cost sensors (LCS) like Purple Air (PA) sensors offer broader coverage and real-time data but face accuracy challenges under varying environmental conditions.

This study addresses these challenges by focusing on the calibration of particulate matter (PM2.5) data—tiny pollutants that can pose serious health risks, including respiratory and cardiovascular diseases. Our systematic comparison explores eleven machine learning models to enhance PA sensor accuracy and reliability for both scientific and regulatory use.

VIEW ON GITHUB:

The GitHub repository houses the actual code and scripts. It is organized into top-level folders, such as DNN/, RNN/, etc., with each folder containing subfolders for different libraries like TensorFlow, PyTorch, and RStudio. These subfolders contain the specific model scripts, configuration files, and metrics necessary for running the experiments. The Training Data folder in GitHub includes the CSV datasets used in the study, while README files within each architecture folder guide you through installation, setup, and execution of the models.