Paper Key : IRJ************653
Author: B.nandini,G.ravi Kumar,B.deesritha,K.kiran Chowdary,B.mahesh Babu,D.chandu
Date Published: 05 Apr 2025
Abstract
Air pollution represents a significant danger to both human health and ecological balance, especially in areas experiencing rapid urban growth. The ability to accurately forecast air quality is crucial for creating prompt intervention methods and keeping the public informed. This document investigates how machine learning techniques can be utilized to predict levels of air pollution, with a particular emphasis on the various Air Quality Index (AQI) classifications. By leveraging past data that includes multiple pollutants (PM2.5, PM10, NO, NO2, NOx, NH3, CO, SO2, O3, Benzene, Toluene, Xylene) alongside weather-related variables from monitoring stations, we construct and assess predictive models. This research assesses the efficacy of different classification methods, including Naive Bayes, Support Vector Machines (SVM), Logistic Regression, and Decision Trees, in addition to an ensemble method known as the Voting Classifier. These models are designed to classify AQI categories (e.g., Good, Satisfactory, Moderate, Poor, Very Poor, Severe) based on the concentrations of pollutants recorded. Initial findings suggest that machine learning can effectively forecast air quality levels. The goal of the comparison is to pinpoint the most effective models for this classification challenge, establishing a basis for a system that can provide timely air pollution forecasts to support both environmental management and public health safety. Future endeavors will aim to integrate a wider range of data sources and enhance model accuracy and responsiveness in real-time situations.