Comparative Analysis of Machine Learning Algorithms for Predicting Air Quality Index

Many scientists and researchers have been worried over the past few decades about the issue of air quality analysis and forecasting, because air pollution in the modern world has become a terrible environmental issue. There have been several health, environmental, and climatic changes owing to polluted air. The major causes leading to poor air quality are urbanization and industrialization. The major and significant air pollutants that affect air quality include NOx, SOx, CO, PM2.5, and PM10. Government agencies employ an Air Quality Index (AQI) to communicate to the public how contaminated the air is now or how polluted it is expected to become. The major emphasis of this work is the analysis of the aforementioned concentration of air pollutants, and classifying the different pollutant concentration levels that adversely affect the maintenance of favourable air quality based on a machine learning approach. We used the dataset, which consists of data on several types of air pollutants taken hourly, at different stations across various cities in India. The hourly data were collected from 15 January 2015 to 1 July 2020. There were 16 attributes. We employed the Decision Tree, Support Vector Machine (SVM), and Random Forest. The most accurate classification is done by a random forest classification algorithm. It outperformed the other approaches with a maximum accuracy of 74%. These results will assist to enhance present research, and guide the future.