A Hybrid Machine Learning Framework for Prediction, Classification, and Regional Clustering of Air Quality Index

Abstract

Air pollution has emerged as one of the most critical environmental and public health challenges worldwide. Poor air quality is linked to respiratory diseases, cardiovascular disorders, a short life span and significant economic loss. Thus, correct prediction and classification of air quality is crucial for any effective monitoring and decision making in the environment. The application of machine learning, in predicting and classifying the Air Quality Index (AQI), based on the concentration of atmospheric pollutants and meteorological parameters is investigated in this study. The machine learning algorithms used for the analysis of air quality patterns were: Linear Regression, Polynomial Regression, K-Nearest Neighbors (KNN), and K-Means Clustering. The regression models were used to predict the AQI, the KNN and K-Means were used for classification and clustering of pollution levels. Data consisted of PM2.5, PM10, NO₂, SO₂, CO, O₃, temperature and humidity. The experimental results showed that machine learning models can effectively predict the AQI and categorize pollution. PolyNomial Regression model performed best among the evaluated models because it was able to capture the nonlinear relationships among the environmental variables. The results indicate that machine learning-based AQI forecasting systems can assist the environmental authorities in pollution control and management strategies.