ENHANCED ANALYSIS AND PREDICTION OF THE AIR QUALITY INDEX USING MACHINE LEARNING

Global Warming is a pressing concern in this AI era, and Air Pollution is one of the causes.Air Quality Index [AQI] is the metric to find the purity of the Air.We employ two modes, namely traditional mode and timeseries mode, to predict the relevance of air pollutants, especially in the Indian context.The traditional mode involves rigorous preprocessing and hyperparameter tuning with Regression models, including Random Forest Regressor, XGBRegressor, Gradient Boosting Regressor, and others, to determine the relevance of air pollutants or features in increasing AQI.The timeseries models include Prophet, SARIMA, Lasso/Ridge Regression, Random Forest Regressor, and more to identify relevant patterns between the AQI of the city, day, and month of the year.All the research has been conducted in the Indian city context, and to streamline the overall process, a web application using Streamlit has also been developed.The user can input the relevant features and get the predicted AQI from the best model loaded through pickle.The research also talks about the findings about the methodology and scope of feature enhancements that could improve the performance for a sustainable future.