COMPARATIVE STUDY OF CLUSTERING ALGORITHM TO DETERMINE STUDENT PROFILES BASED ON ACADEMIC ABILITY

Abstract

Competition among similar study programs requires each Program to develop strategies to compete for new students, especially for private universities like Universitas Islam Bandung. One of the study programs at Universitas Islam Bandung, namely Statistics, has experienced a downward trend in new student enrollment over the last 3 years. One strategy that can be employed is to identify student profiles. Creating student profiles can serve as an evaluation tool for improving educational quality and the quality of student input. One method for finding patterns in data, in this case, student profiles, is clustering. In this research endeavor, a comparison of clustering algorithms for selecting the best model was conducted using K-Means, Complete Linkage, Average Linkage, and Ward’s Method. The outcomes show that Average Linkage with two clusters was selected as the best model with the minimum DBI value. The Average Linkage clustering reveals that the academic performance of students in cluster one surpasses that of students in cluster two. The promotional strategy can be focused on cluster 1 to improve the quality of student input.