abstract
- Dengue fever is one of the major health related issues as reported in World Health Organization (WHO). As recently, Malaysia has been reported with dengue epidemic, that could rises up to 120, 000 cases recorded per year. This serious issue need a vital look to prevent the dengue occurrences as it has no medicine yet to be found. Therefore, a study is needed on the factors that influencing dengue incidences. Using this study on factors that influencing dengue incidences, it is important for the government to create a predictive system so that precaution steps could be taken. Dengue incidence prediction models are very important at present as the disease is becoming a major health issue in tropical and subtropical countries. This study builds a dengue incidence prediction model to avoid epidemic using climate models in online. MariaDB database engine been used to store all the data and it is imported to R-software in localhost to build the prediction model. It also presents a high accuracy dengue occurrences prediction model which could forecast the dengue occurrences accurately. Manifold learning procedures are carried out to reduce the dimension of climate variables into a single dimension by maintaining the geodesic distances between all points. Next, machine learning procedures such as clustering (based on the K-means technique) and linear regression are carried out to develop a prediction model. Averaged silhouette width method was used to determine the number of clusters, K, for the K-means technique. Finally, percentage of accuracy were reported based on the model’s R-squared. Our prediction model achieves accuracies up to 82.58%. This proves that dimension reduction techniques can increase the accuracy of a developed prediction model for dengue forecast.