Performance optimization through Hybrid Modelling: An application to Dengue Disease

Abstract

Although, Dengue virus could be prevented through responsible human actions, it has become a serious threat to mankind. This study was intended to increase the prediction accuracy of dengue transmission using hybryd models. After forecasting with Grey Forecasting Model, Growth Curve Model, Alpha Sutte Indicator and Generalized Additive Model, the models with the best prediction accuracy were determined through lowest Mean Absolute Percentage Error (MAPE) recorded in error calculation. Accordingly, a hybrid model was developed, by using a weighted average method as a coupling technique. Through the calculations and the analysis carried out, Alpha Sutte Indicator and the Generalized Additive Model were chosen to develop the Hybrid Model. The model enhances the prediction accuracy for most of the regions in Sri Lanka. Forecasting dengue transmission accurately is important to allocate medical personnel and equipment, conduct effective environmental management and awareness programs and chemical vector controlling in correspondence to the rising figures of dengue patients.