Examining the Relationship between Environmental Hazards and Socioeconomic Factors in the United States Using Machine Learning Methods

Abstract

This study investigates the intricate relationship between environmental factors and socioeconomic indicators across all U.S. states, utilizing data from the EPA EJScreen. With environmental challenges like climate change and air pollution continuing to persist, it is essential to understand not only their overall impact but also who is most affected. Disadvantaged groups, such as low-income individuals and people of color, often bear the brunt of these issues, experiencing heightened exposure to pollutants and other hazards. Through comprehensive analysis using advanced machine learning methods such as correlation analysis, Pearson correlation coefficient, linear regression, XGBoost, and neural networks, this research identifies statistically significant correlations between environmental and socioeconomic factors in certain states, potentially linked to historical patterns of systemic discrimination in policymaking. Conversely, in other states, this relationship appears less pronounced. Further investigation is warranted to elucidate the underlying factors contributing to these variations, including geographical features and demographic characteristics. While providing valuable insights, this study acknowledges limitations such as the reliance on state-level data and the use of correlation analysis, which does not establish causation. Moving forward, future research should explore these relationships at finer geographical scales and employ methodologies beyond correlation analysis to better understand causal relationships. By addressing these limitations and continuing to explore the complexities of environmental justice, policymakers can inform more targeted and effective interventions to address environmental disparities and advance equity for all communities.