Using Machine Learning for Air Quality Prediction in Alabama: An Environmental Justice Case Study

Environmental justice encompasses the fair treatment of all people regardless of race, color, national origin, education level, or income, in the context of environmental impacts. This paper investigates the application of machine learning (ML) to address environmental justice issues arising from air quality disparities in Jefferson County, Alabama, particularly around Birmingham. The study makes several key contributions including curation of ML-ready datasets from environmental sensor data for air quality analysis; fusion of domain specific datasets for analyzing the effects of air quality degradation across demographics; utilization of novel advanced ML approaches using a Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM) architecture to identify, predict, and analyze areas with significant environmental injustice; and projection for air quality degradation with demographics to assist in decision making and future policy making. This research demonstrates the utility of ML in rapidly identifying environmental justice hotspots and offers predictive capabilities for timely mitigation measures. Additionally, this study provides a pathway for extending the methodology to similar issues globally. The work also offers social value by providing policymakers with actionable insights for mitigating air quality disparities and empowering marginalized communities with access to precise environmental predictions.