Ensemble tree-based machine learning models for data-driven AQI prediction

Introduction Air pollution has become a critical global health and environmental concern due to rapid urbanisation and industrialisation. Making accurate prediction of air quality has became crucial for effective prevention and control. Machine learning offers a robust approach by capturing complex pollutant interactions, enabling precise and timely air pollution prediction. However, existing approaches lack temporal-based chronological splitting, which aids in addressing bias in the data. Further, these AQI prediction methods are inadequate in handling missing data prior to imputation. Methods The proposed framework aims to integrate pre-processing and AQI prediction on Central Pollution Control Board dataset of India. Further, this work ensures the statistical validity of the imputation and adopts strict chronological validation on the benchmarking models to predict AQI. The dataset is formally analysed using statistical mechanisms such as MCAR, MNAR, and MAR, and the imputation methods such as mean, KNN, MICE, GAIN, and linear interpolation. Feature engineering is used to increase the number of features from 12 to 24 by adding rolling, lag, and temporal features. The data is split into 80:20, in strictly chronological order for the prediction models such as CatBoost, XGBoost, LightGBM, and Random Forest, using the metrics MAE, R2, and RMSE. Results Among the four models, CatBoost achieves the strongest overall performance, with the MAE of 10.94, RMSE of 16.93, and R2 of 0.96. Season-stratified evaluation reveals that summer is the most predictable season, with the RMSE of 11.25 and an R2 of 0.92. Discussion The integration of imputation methods and adoption of strict chronological validation for AQI prediction establishes a temporally validated benchmark across multiple cities in India. This enables policymakers to implement season-specific emission controls, equips with timely health awareness and supports environmental agencies in monitoring and regulating pollution.