Prospective New College Student Dashboard: Insights from K-Means Clustering with Principal Component Analysis

Abstract

Higher education institutions currently compete to attract prospective students, necessitating the implementation of effective and efficient promotion strategies. Universities can create effective promotion strategies by considering the characteristics of prospective students. The clustering method is An approach to understanding prospective students' characteristics. However, clustering analysis with numerous attributes faces the issue of the curse of dimensionality. This research aims to overcome the curse of dimensionality in clustering by applying the K-means clustering method, which is enhanced through dimensionality reduction using Principal Component Analysis (PCA). The enhanced K-means method was applied to clustering data on prospective students at Telkom University Surabaya. The data used in this research pertains to prospective students interested in Telkom University Surabaya (TUS). Attributes include school origin, school province, domicile district/city, type of registration pathway, school type, and choice of study program. This research indicates that utilizing K-means clustering with PCA yielded superior cluster outcomes when evaluated against the Davies-Bouldin Index and Calinski-Harabasz Index, surpassing the performance of ordinary K-means clustering. The cluster analysis also shows that the ideal number of clusters is 3, using three principal components (PCs). The outcomes of the K-means clustering with PCA are incorporated into a dashboard that visually displays comprehensive information about the clusters. This dashboard simplifies examining how potential new students are geographically spread out, alongside how clusters are distributed across various study programs, school types, registration routes, and locations in districts/cities. The analytical data exploration on the dashboard can be utilized to address Business Questions Formulation related to the characteristics of prospective new students based on clustering results at Telkom University Surabaya.