Modelling and prediction of dengue cases at two progressive regions in Malaysia

Abstract

Time series analysis method has been used to model and forecast the weekly number of dengue cases in Malaysia. Dengue fever is one of the pressing public-health problems in Malaysia and causes substantial economic burden. Forecasting results of dengue cases could be useful in preparedness programme to facilitate the planning of public health interventions. The analysis is conducted in Selangor and Wilayah Persekutuan, two states in Peninsular Malaysia with high dengue occurrences. Two time series models, Double Exponential Smoothing and Holt-Winters Method are fitted to dengue data (2010-2015) from both regions to determine the best model to represent the cases. Both models are able to represent the data very well, however closer inspections using MAPE, MAP and MSD indicated that Holt-Winters method is most appropriate. Forecasting results using Holt-Winters method exhibit a substantial increase for Selangor and a more gradual increase for Wilayah Persekutuan.

Keywords: Dengue, Exponential Smoothing Models, Forecasting