The dengue disease has been reported as a major cause of morbidity and mortality for the last 40 years worldwide including in Malaysia. According to the Malaysian Statists Department, total reported number of dengue cases for the year 2008 to 2010 is 136,992 and it is increasing every year. According to Ministry of Health (MoH) of Malaysia, in urban area 77 of people are suffering from dengue virus compared to 23 people suffer in rural areas. Several studies on dengue are based on their serotype, epidemiology, weather forecasting, but this thesis look into the prediction modelling based on clustering algorithm. Since, the motivation to work on this projects comes from old existing studies which only represent the spatial map with undefined number of incidences and no clustering model exist. Since, in our study we have proposed the spatial- temporal mapping along with the Clustering techniques which comes with k-mean as the initial step to generate the clusters of incidences after that to optimize K-means, K-NN techniques is applied to find best fit K values. After that Gaussian Mixture Model is applied to find density of the dengue incidences, since where K-means is used to find the centroid of the incidences. To process the Gaussian Mixture Model, Estimation Maximization (EM) algorithm is used to relate the cluster with their respective clusters. To optimize the EM algorithm Bayesian Information Criteria (BIC) is performed which gives the best fitting model of BIC and optimizes the EM algorithm. In the end, Geographical Information System (GIS) technique is use to visualize vulnerability mapping to locate the accurate prediction location for dengue incidences in state Selangor of Malaysia (area of study). This research work discusses and implements the visualization and prediction modelling based on machine learning concepts, for the vector borne diseases (dengue). The results are tested for a region (Petaling district of Selangor state) in Malaysia and they showed good performance in predicting the dengue incidences. Thus, the proposed method is able to localize the nature of dengue incidence that can further be utilized for vector disease controlled process. The results confirms location of the predicted coordinates based on the previous data for the year 2014. © 2018 IEEE.