Predicting the dengue incidence in Singapore using univariate time series models.

Abstract

Dengue is endemic in Singapore with year-around transmission. Prediction of dengue incidence is important for effective use of limited resources for vector-control and contingency measures. In the work, we develop a set of time series models based on the observed weekly dengue incidence since 2000. The dengue incidence data of Singapore from 2000 - 2011 is used to develop and fit the predictive models. For testing and validation, we use the 2012 data at two levels: A) real versus predicted incidence and B) real versus predicted outbreak severity. The statistical measures of validation show that the models predict both the dengue incidence and the outbreak severity level with acceptable level of accuracy.