Utilizing Machine Learning for air pollution prediction, comprehensive impact assessment, and effective solutions in Kolkata, India

Abstract

Escalating air pollution in urban areas is a matter of concern, and deteriorating air quality is having numerous impacts on human health and the environment. Kolkata is one of the most densely populated and highly polluted cities in India. The aim of this work is to predict the concentration of ambient PM_2.5 using different air pollutants and meteorological parameters as predictor variables by using statistical and different Machine Learning techniques as well as to understand the influence of other air pollutants and meteorological factors in ambient PM_2.5 prediction. Different advanced machine learning algorithms like Random Forest Regression, decision trees, k-nearest Neighbour, Support Vector Regression, Ridge Regression, Lasso Regression, and XGBoost have been used, and the results show that the XGBoost model exhibits higher linearity between predictions and observations, among other models. Moreover seasonal variation of the most influential factor for prediction of PM_2.5 is also noticed during the analysis. This work adds to the broader comprehension of the convergence of environmental science, public health, and machine learning and it offers significant perspectives for sustainable urban planning and pollution control tactics in rapidly expanding metropolitan areas such as Kolkata.