This paper aims at obtaining better model between seasonal ARIMA and simple seasonal exponential smoothing that will be used for forecasting number of diabetes patients in a given hospital. Monthly dataset from January 2009 to December 2019 from Enugu State Teaching Hospital was used for this research. Seasonal ARIMA was modelled using the techniques of Box-Jenkins, and simple seasonal exponential smoothing modelled using the least squares method. Bayesian Information Criterion (BIC) was employed to obtain the best seasonal ARIMA model, while the Theil’s U statistics and MAPE were used to obtain the best forecast model. ARIMA (1,1,2)(0,0,0)12 was selected as the best SARIMA model with the BIC of 7.873, and simple seasonal exponential smoothing was considered the best forecast model with a Theil’s U Statistic of 0.11241 and MAPE of 23.450. The fitted model was used to make out-sample forecast for the period January 2020-December 2025. The fitted model in this findings will help Enugu state government to plan efficiently, expand public sensitization, and allocate adequate resources for emergencies.
Citation: Christogonus Ifeanyichukwu Ugoh, Anyadiegwu Chinelo Ujunwa, Thomas Chinwe Urama, Ugwu Gibson Chiazortam (2022) Application of SARIMA Model and Simple Seasonal Exponential Smoothing on Diabetes Mellitus: A Case of Enugu State Teaching Hospital, Nigeria, European Journal of Statistics and Probability, Vol.10, No.1, pp., 21-32
Keywords: BIC, Diabetes Mellitus, MAPE, SARIMA, Simple Seasonal Exponential Smoothing, Theil’s U Statistic
