| Abstract | Air quality in India is a field of trouble. There has been a steady decrease in the air quality from the previous years. The poor air quality straightforwardly influences the environment and majorly affects the human wellbeing. Sulphur dioxide, nitrogen dioxide, respirable suspended particulate matter, particulate matter are the main pollutants in the air which are responsible for the poor air quality. The air quality is measured with the help of air quality index (AQI) which gives a relative measure for surrounding air concentration. The study likewise illuminates the impact of parameters, for example, sulphur dioxide, nitrogen dioxide, respirable suspended particulate matter, particulate matter on to air contamination and the requirement for the measures against the poor air quality. The objective of this research study is to predict the air quality into six groups such as good, moderate, unhealthy for sensitive people, unhealthy, very unhealthy, hazardous. This study proposes an architecture for the air quality prediction which predicts the quality of air using data mining techniques using decision tree, random
forest, light GBM, SVM, naive bayes, ridge regression, XGboost, KNN, Multinomial
logistic regression and gradient boosting.on data consisting of 12 attributes from the years 1990-2015 for every states in India. Among all the algorithms XGBoost in regression and gradient boosting in classification gave the best accuracy. Air quality
prediction system is developed using the best models with three major functionalities
train, predict and visualize. After user enters all the information system gives the output
showing the AQI value and how dangerous it is to human health. Keywords: AQI, Air quality, Air quality prediction, Decision tree, Random forest, light GBM, SVM, Naive bayes, Ridge regression, XGboost, KNN, Multinomial logistic regression and Gradient
boosting |