Application of imputation methods for missing values of PM₁₀ and O₃ data: Interpolation, moving average and K-nearest neighbor methods

Background: PIn air quality studies, it is very often to have missing data due reasons such as machine failure or human error. The approach used in dealing with can affect the results of analysis. main aim this study was review types mechanism, imputation methods, application some them PM10 and O3 Tabriz, compare their efficiency. Methods: Methods mean, EM algorithm, regression, classification regression tree, predictive mean matching (PMM), interpolation, moving average, K-nearest neighbor (KNN) were used. PMM investigated by considering spatial temporal dependencies model. Missing randomly simulated 10, 20, 30% values. efficiency methods compared using coefficient determination (R2 ), absolute error (MAE) root square (RMSE). Results: Based on for all indicators, KNN had best performance, respectively. did not perform well without spatio-temporal information. Conclusion: Given that nature pollution always depends next previous information, computational based before after information indicated better performance than others, so case pollutant data, recommended use these methods.