| Title |
Prediction of Radon Concentration using Seasonal ARIMA Model |
| Authors |
정재원(Jai-Won Chung) ; 윤덕경(Deog-Gyeong Yoon) ; 김규식(Gyu-Sik Kim) |
| DOI |
https://doi.org/10.5573/ieie.2021.58.4.53 |
| Keywords |
Indoor radon concentration; Seasonal ARIMA; Time series cross-validation; Forecast model |
| Abstract |
Modern people who live mainly in enclosed rooms faced the danger of radiation exposure by indoor radon. The model for prediction of the indoor radon concentration could be used to effectively reduce the indoor radon concentration, but few studies have looked at the model. In this paper, we proposed the model for prediction of the indoor radon concentration, which was derived using the indoor radon concentration data measured for a total of 921 hours and seasonal ARIMA analysis. As a result of construction of the models using a total of 691 hours of the data as the training set and the time series cross-validation method, ARIMA(3,1,4)(3,1,4)24, ARIMA(3,1,4)(2,1,4)24, ARIMA(3,1,4)(4,1,4)24, ARIMA(4,1,4)(2,1,4)24 were found to be the optimal models. As a result of evaluating the performance by calculating R2, RMSE, and MAPE values ??using 230 hours of the data used as the test set for the four ARIMA models, the ARIMA(4,14)(2,1,4)24 model showed the best performance, and predicted indoor radon concentration with an error of about 18%. It is expected that indoor radon concentration could be effectively reduced by opening a window or operating a ventilator before the indoor radon concentration increases using the ARIMA model. |