Assessing machine learning models for fine particulate matter (pm2.5) concentrations prediction in iğdır, türkiye

Abstract

Air pollution is directly related to various atmospheric and terrestrial parameters, even in arid regions. Iğdır is a province in eastern Türkiye and has an arid climate. This study aims to predict air pollution levels in Iğdır by examining the relationship between fine particulate matter (PM₂.5) concentrations and meteorological variables. Accurate prediction of air quality is important for environmental sustainability and public health. In this study, daily PM₂.₅ concentration data obtained from the air quality monitoring station in Iğdır were used alongside meteorological parameters such as temperature, relative humidity, wind speed, and horizontal visibility acquired from meteorological stations. Machine learning models, including Linear Regression, Support Vector Regression, K-Nearest Neighbors, Random Forest, Adaptive Boosting, and Extreme Gradient Boosting algorithms, were applied, and their performance was evaluated using error metrics. The analysis results indicated that meteorological variables play a significant role in predicting PM₂.₅ concentrations. The study demonstrates the effectiveness of machine learning methods in air pollution prediction and provides valuable insights for air quality management. The findings may contribute to the development of air pollution forecasting systems.