Air Quality Prediction using Machine Learning: A Data-Driven Study of Ambient Pollutants in Delhi

Abstract

Air pollution is one of the major threats to the environment and health worldwide. This research analyzes an air quality dataset consisting of major pollutants like NO_2,SO_2 ,CO,O_3,NH_3,PM2.5,PM10 in the capital city of India- Delhi. Using Python-based exploratory data analysis (EDA) and a comparison and evaluation of various machine learning models like Linear Regression, Random Forest, XGBoost, and LightGBM were performed on the data to understand pollutant behavior, inter-correlations, and their influence on Air Quality Index (AQI). The performance of each model was evaluated based on the results like Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R^2). It is observed that ensemble-based models gave better performance than linear models. LightGBM gave the best result with the lowest RMSE and the highest R^2, followed by XGBoost with a very close result. Random Forest gave moderate results, whereas Linear Regression failed to give good results due to non-linear pollutant interactions. It is evident from the results that the ensemble and boosting algorithms work best for the prediction of AQI.