Real time air quality monitoring and forecasting using AI and nature-based recommender system for climate-resilient air quality management

Abstract

Air pollution is a significant environmental and public health challenge, particularly in developing countries, where increased urbanization and uncontrolled emissions have exacerbated atmospheric pollution levels. This study aims to design an AI-based expert system for real time air quality monitoring and forecasting. It also employs a nature-based recommender system integrating multi source environmental and meteorological data. The real-time data has been collected using Google Application Programming Interface keys, The data contains the concentration of notable air pollutants including PM₂.₅, PM₁₀, NO₂, SO₂, CO and O₃ along with weather variables such as wind speed, humidity, temperature and atmospheric pressure. Machine learning algorithms including Random Forest Regressor and Artificial Neural Network (ANN) have been used to forecast pollutant concentrations and resulting Air Quality Index (AQI). The model achieved a high accuracy of 80-85% with low Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) values and a strong R² score, indicating reliable forecasting performance. Results showed that PM₂.₅ and PM₁₀ are the most significant pollutants affecting air quality followed by NO₂, O₃ and temperature. A deep reinforcement learning (DRL) based recommender module assisted in generating alerts during hazardous pollution events, by providing adaptive, nature-based solutions. The system visualized live AQI, 24 hours forecasts and health alerts with smooth functionality. The outcomes of this study is to contribute toward intelligent environmental management through sustainable early warning solutions for air quality management.