Implementation of Boosting Algorithms in Predicting Air Quality Index of South Indian Cities

This paper addresses the critical issue of atmospheric pollution in India, underscoring the necessity for precise predictive analytics of Air Quality Index (AQI) data for effective pollution control. The study delineates the etiological factors and substantial health hazards correlated with air pollution, encompassing elevated mortality rates, respiratory and cardiovascular diseases, and mental health complications. The AQI is presented as a necessary component for converting complex air quality data into a single, easily understandable metric. This research aims to facilitate effective pollution control through real-time AQI monitoring and precise future predictions for timely interventions. To attain this objective, the research employs the use of boosting algorithms, like extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), and an ensemble stack of XGBoost and LightGBM for AQI prediction of South Indian cities. The performance of these models was found noteworthy, with high R² scores and low root-mean-squared error (RMSE) scores, exhibiting its efficacy in providing highly accurate results. By merging technological innovation with machine learning capabilities, the research aims to equip decision-makers with actionable insights for informed pollution mitigation strategies, promoting a more sustainable environment. Keywords: Air Pollution, Air Quality Index (AQI), XGBoost, LightGBM, Ensemble Stack