Influenza can be easily spread among humans by coughs or sneezes. It is one of the major public health problems caused by viruses. An influenza epidemic occurs in Thailand every year and produces social burdens. Public health forecasts show societal information in advance and can point to the future magnitude of various public health issues. Therefore, this study was to perform the model in order to explain and predict influenza incidence using a seasonal autoregressive moving average model with Box-Jenkins (SARIMA). The monthly influenza virus infection cases in Public Health Region 8, Udonthani, Thailand from January 2016 to December 2018 were used to develop the model. The best fit model was determined by Akaike’s Information Criteria (AIC), Bayesian Information Criteria (BIC) and Root Mean Square Error (RMSE). The results showed that SARIMA was the best model for forecasting influenza incidence. This model had the lowest AIC (59.24), BIC (67.16) and RMSE (0.4574). Based on the comparison of actual and forecast values, the mean absolute percentage error (MAPE) was 24.15%. It shows that the model could be used to predict and demonstrate the influenza incidence.
Keywords: forecasting; influenza; SARIMA model
*Corresponding author: Tel.: (+66) 23548530 Fax: (+66) 23548534
E-mail: jutatip.sil@mahidol.ac.th
