Air pollution in South India’s industrial clusters remains poorly studied, despite their growing importance as sources of pollutants in the region. This study presents an in-depth examination of air quality in Gummidipoondi, Tamil Nadu, encompassing statistical analysis, feature importance, and machine learning models. The results showed that PM2.5 and PM10 are the main features that affect the Air Quality Index (AQI), accounting for more than 99% of its variations. Gaseous pollutants and weather conditions had only a lesser effect. XGBoost was the most accurate machine learning model (R² = 0.9965, RMSE = 0.0573), compared to regression and neural models. The SHAP summary plot confirmed that PM2.5 is the main feature that had a significant impact on AQI. Temporal patterns indicated significant pollution impacts during winter and post-monsoon seasons, attributed to slow boundary layer conditions and daily industrial activities, while monsoon offered natural alleviation through dispersion. These results broaden air quality studies outside urban areas and illustrate the effectiveness of explainable machine learning for real-time surveillance and early warning systems. The study underscores the immediate necessity for focused particulate matter regulation to mitigate health hazards in industrialising corridors.