A STUDY ON MACHINE LEARNING-BASED APPROACHES FOR PM2.5 PREDICTION

Abstract

Clean air and water is the fundamental need of humans. But people are exposed to polluted air produced due to several reasons such as combustion of fossil fuels, industrial discharge, dust and smoke which generates aerosols. Aerosols are tiny droplets or solid particles such as dust and smoke that floats in the atmosphere. The size of the aerosol also called as particulate matter ranges from 0.001–10 μm which when inhaled by human affects the respiratory organs. Air pollution affects the health of 9% of the people every year. It is observed as the most important risk factor that affects human health. There is a need for an efficient mechanism to forecast the quality of air to save the life of the people. Statistical methods and numerical model methods are largely employed for the predicting the value of PM2.5. Machine learning is an application of artificial intelligence that gives a system capability to learn automatically from the data, and hence, it can be applied for the successful prediction of air quality. In this paper, various machine learning methods available to predict the particulate matter 2.5 from time series data are discussed.