Air quality index prediction using meteorological data using featured based weighted xgboost

Abstract

Over the recent years, air pollution or air contamination has become a concerning threat, being responsible for over 7 million deaths annually according to a survey conducted by “WHO”(World Health Organisation). The four air pollutants which are becoming a concerning threat to human health are namely respirable particulate matter, nitrogen oxides, particulate matter and sulphur dioxide. Hence to tackle this problem, efficient air quality prediction will enable us to foresee these undesirable changes made in the environment keeping the pollutant emission under check and control. Also inclusion of meteorological data for isolating the factors that contributes more to the Air Quality Index (AIQ) prediction is the need of the hour. A feature based weighted XGBoost model is built to predict the AIQ of Velachery, a fast developing commercial station in South India. The model resulted in low RMSE value when compared with other state of art techniques.