Analyzing schools admission performance achievement using hierarchical clustering

Abstrak

In this study, an implementation of hierarchical clustering methods was conducted in schools’ admission data. We aim to demonstrate that the hierarchical clustering method can be used to help analyze the membership changes of each cluster based on its achievement number of new students from different months period observations. This method can be used by decision-makers to make a strategy for each school which has decreasing achievement from the previous period. In this paper, we employ the hierarchical clustering method to cluster admission performance achievement from fifty Telkom Schools. Instead of clustering admission in one period directly, this paper tried to analyze the movement of clustering membership from one period to another. We observed the movement membership of the group from three categories period, such as monthly, quarterly, and semesterly. The experimental results demonstrate that the monthly scenario was the best clustering result. The monthly scenario achieves the best score for all metrics such as the Dunn index, Silhouette score, Davies-Bouldin index, and Calinski-Harabasz compared to the quarter and semester scenario. There are four schools which are consistent in the first cluster and seven schools which are consistent in the second cluster in all scenarios and all periods.