Investigating the applicable position of established cloud servers using the hybrid partitional clustering

This paper proposes a hybrid partition clustering, which is the combination of three techniques: partitional clustering, proportional allocation, and center-of-gravity. The objective of this hybrid clustering is to identify the applicable position of established cloud servers to support the testing of a high-speed internet platform. These cloud servers will be installed in the Eastern region of Thailand. Moreover, hybrid clustering is developed using RStudio and R packages. The latitude-longitude position and the number of populations from each district that has internet service providers are considered as main factors to do the cluster. Regarding the experiment, we found that K-Means provide a better solution than K-Medoid with a silhouette width value equal to 0.45. In addition, the results of this hybrid clustering will also be compared with the existing solution. The difference in sample users between these two methods is approximately 25 percent.