Exploring Socioeconomic Drivers of Air Quality in Singapore with Machine Learning and Explainable AI Techniques

Air quality is essential for public health and environmental sustainability, as poor air conditions are associated with significant health risks and ecological damage. Although the impact of socioeconomic factors on air quality is increasingly recognized, their incorporation into predictive models has been underexplored. This study seeks to integrate socioeconomic attributes into air quality modeling, leveraging tree-based ensemble learning techniques and explainable AI methods to identify key determinants affecting air quality. Using Singapore's Pollutant Standards Index (PSI) as a case study, we analyzed 34 socioeconomic attributes through various models, including Random Forest, ExtraTrees, XGBoost, and LightGBM. Our feature engineering, informed by the Pearson Correlation Coefficient, revealed hidden relationships among these attributes. Model performance was evaluated using 10-fold cross-validation, supplemented by Shapley Additive Explanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) to enhance the interpretability of our predictions. Random Forest emerged as the top-performing model on the original dataset, achieving mean values for Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE) of 14.12%, 22.85%, and 14.09%, respectively. In contrast, a single-layer stacking model excelled on the feature-engineered dataset, yielding results of 13.68%, 22.73%, and 13.59%. The analyses conducted with SHAP and LIME highlighted key drivers of air quality, such as the Arrivals and Departures Ratio, Visitor Average Length of Stay, and Price of LPG, emphasizing the significant effects of tourism, industrial activity, and transportation policies on air quality. This study effectively demonstrates that integrating socioeconomic data with explainable AI can provide valuable insights, make it easier for targeted interventions and informed policy decisions that support sustainable urban development.