Forecasting Typhoid Fever Cases using Decomposition Method, Holt-Winter's Method, and Sarima Method

ABSTRACT:This study compares the forecasting accuracy of the three univariate time series models, Decomposition, Winter’s and Seasonal autoregressive integrated moving average (SARIMA) method to predict also monthly typhoid fever cases. Secondary data is collected from Federal Medical Centre Jalingo and explored with the three time series models. The study applied time series decomposition method, Holt-Winter’s method, and SARIMA method to compare their forecasting accuracy using Root Mean Square Error (RMSE) as the selection criterion. The results for typhoid fever cases revealed that SARIMA (4,0,0) (1,0,0) best fits and predict the data more than the Decomposition, and the two Holt-Winter’s model. The monthly Typhoid fever indices indicates that the month of April, May and July to October has higher typhoid fever cases compared to the annual average. The highest index is in the month of August accounting for nearly 6.6% increase over the annual average.