K-Nearest Neighbors for Predicting Ozone Concentrations: A Machine Learning Approach for Air Quality Assessment

Ozone (O<sub>3</sub>) is a significant air contaminant that poses severe health risks, particularly in urban areas. Accurate prediction of the ozone concentration level is crucial for increasing public consciousness and giving important data to governments for public health alerts and air quality management. This study explores the application of machine learning techniques, the K-Nearest Neighbors (KNN) method, for predicting ozone (O<sub>3</sub>) concentrations based on meteorological variables collected from three monitoring stations in the Klang Valley region. The research involves data preprocessing procedure that includes handling missing values through imputation and applying the KNN algorithm to predict ozone concentrations. The model was trained and tested using cross-validation and its performance was assessed using evaluation metrics, such as Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). The KNN model achieved the most accurate predictions at Petaling Jaya station with an MAE of 0.00350 and RMSE of 0.00447, followed by Cheras station (MAE: 0.00402, RMSE: 0.00520) and Batu Muda station (MAE: 0.00406, RMSE: 0.00527). These results indicate that the KNN model achieved relatively low prediction error in this study, demonstrating its potential reliability in forecasting ozone concentrations. These findings suggest that the KNN model can be applied to predict ozone concentrations across different urban locations, offering valuable insights for public health planning and supporting strategies for mitigating ozone pollution.