A Comparative Study of Machine Learning Models for Air Quality Index (AQI) Prediction Using Multi-Pollutant and Meteorological Parameters in Maharashtra, India

Abstract

Air pollution has emerged as a critical environmental and public health concern in rapidly urbanizing regions of India. Maharashtra, hosting several highly industrialized and densely populated metropolitan centers, exhibits considerable spatiotemporal variation in air quality influenced by pollutant concentrations, meteorological conditions, seasonal cycles, and anthropogenic activities. This study proposes an explainable machine learning framework for predicting the Air Quality Index (AQI) in Maharashtra, India, utilizing comprehensive pollutant and meteorological parameters. The framework incorporates particulate matter (PM₂.₅ and PM₁₀), nitrogen dioxide (NO₂), sulfur dioxide (SO₂), carbon monoxide (CO), ozone (O₃), temperature, relative humidity, wind speed, and other relevant environmental variables. Multiple machine learning algorithms, including Linear Regression, Decision Tree, Random Forest, Support Vector Regression, XGBoost, LightGBM, Artificial Neural Network, and Long Short-Term Memory (LSTM), are comparatively evaluated. Model performance is assessed using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and coefficient of determination (R²). Beyond prediction accuracy, Explainable Artificial Intelligence (XAI) techniques, particularly SHAP (SHapley Additive exPlanations), are employed to identify the relative contribution of individual pollutants and meteorological parameters to AQI predictions. The proposed approach aims to provide both accurate AQI prediction and interpretable insights regarding the major factors influencing air pollution in Maharashtra. The results are expected to support data-driven air-quality management and provide an interpretable framework for environmental decision-making.