Geographically Weighted Regression on dengue epidemic in Peninsular Malaysia

Abstract

Dengue has been a global epidemic since World War II, with millions of individuals being infected every year. Repetitive dengue epidemic is one of the main health problems that, due to its rapid spread and geographically widespread, has become a major concern for the government authorities in dealing with this disease. In Malaysia, cases of dengue are reported annually. To keep cases under control, it is important to examine the possible factors that help the growth of the virus. Climatological factors such as rainfall, temperature, wind speed, and humidity are expected to have high potential to increase the growth of the virus in this study, and their spatial variation is associated with cases of dengue. The result revealed that Ordinary Least Square was not an effective method for modelling the relationships between dengue cases and climate variables, as climate variables in different spatial regions act differently. During the analysis, there could be some issues of non-stationarity since the geographical aspect and spatial data were involved. Hence, the Geographically Weighted Regression (GWR) is implemented due to its capability to identify the spatial non-stationarity behavior of influencing factors on dengue incidence and integrate the geographical location and altitude for the spatial analysis. GWR analysis found that the influenced factors exhibited a significant relationship with dengue incidence. GWR also shows a significant improvement in Akaike Information Criteria (AIC) values with the lowest value and the highest adjusted R square. It is expected that the developed model can help the local hygienic authorities design better strategies for preventing and controlling this epidemic in Malaysia.