Technology-enhanced data analytics for student assessment using PCA and clustering to support SDG 4

¹ Directorate of PSDKU Kampus 5, Universitas Negeri Surabaya, Surabaya, Indonesia
² Norwegian Institute of Bioeconomy Research, P.O.Box 115, NO-1431 Ås, Norway

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Achieving SDG 4 on improving quality education requires higher education institutions to adopt technology-enhanced and data-driven approaches. Conventional summative scores tend to reduce multidimensional rubric-based tests to simple categories, which hides subtle trends in student performances. This paper aims to determine hidden performance student profiles through technology-enhanced data analytics. To achieve this, we applied unsupervised machine learning techniques, including Principal Component Analysis (PCA) for dimensionality reduction and two clustering methods (K-Means and Bisecting K-Means) to identify distinct student performance profiles. A total of 136 student records with ten rubric elements were evaluated with these unsupervised machine learning techniques. The results describe that there were two best student clusters suggested by internal measurement matrices, (Silhouette Score, Calinski-Harabasz Index, and Davies-Bouldin Index). Cluster 0 had consistently high balanced performance and scores across all elements, while Cluster 1 had students with uneven mastery. These results indicate that the PCA-Clustering approach is a powerful tool used to discover significant student portraits and promote more equitable, evidence-based assessment activities in the SDG 4 direction. Future work will include increasing dataset size and variation, and exploring adaptive AI-based feedback systems to support personalized and sustainable learning improvement.