Machine Learning Models for Clustering and Validation of Historical Thinking Skills: A Multi-Dimensional Approach

Open

Hera Hastuti, Tri Zahra Ningsih, Suci Kurnia Putri, Aisiah, Najmi, Agus Trilaksana

2026 International Journal on Informatics Visualization Vol. 10 Issue 2 Article Cited by 0 SDG 4SDG 17 Quartile

Abstract

—Identifying Change and Continuity of Upper Secondary Students Using a Machine Learning Approach: Validation. This research aims to investigate the historical thinking skills of upper secondary students to identify change and continuity in history using a machine-learning approach, employing clustering techniques and validation. The population of this study consists of 167 students aged 16-18 years old in West Sumatra. The data was collected by using the rubric to measure students’ ability to identify change (MP) and continuity (MK). The K-Means Clustering method was used to identify clusters of students who can distinguish between change and continuity. In this research, the clustering results were validated using Random Forest and Neural Network models to assess their reliability. Based on the validation results, the Silhouette Score is 0.49, the Davies-Bouldin Index is 0.88, and the Calinski-Harabasz Index is 150.55. Also, the Random Forest method showed that the clustering results were 100% accurate. The mean squared error for Neural Network Validation is 0.006448. The analysis indicates considerable variation in students' skills, and the 'Identifying Change' dimension plays a more crucial role in distinguishing between groups. This analysis offers considerable practical implications for curriculum development, allowing the adoption of learning strategies based on specific deficiencies in each cluster; for instance, reinforcing continuity for those who are good at change, and vice versa. The methodological framework used in this investigation provides a model for applying machine learning methods to multidimensional educational evaluation. This method could be of considerable use in future research undertakings in healthcare, market segmentation, or educational diagnostics. This study illustrates that data-informed personalization within educational settings promotes a more equitable and introspective approach to history education. © 2026, Politeknik Negeri Padang. All rights reserved.

Affiliations

Department of History Education, Faculty of Social Sciences, Universitas Negeri Padang, West Sumatra, Padang, Indonesia; Department of Educational Research and Evaluation Graduate School, Universitas Negeri Yogyakarta, Yogyakarta, Indonesia; Department of Social Studies, Faculty of Social Sciences, Universitas Negeri Yogyakarta, Yogyakarta, Indonesia; Study Program of History Education, Universitas Negeri Surabaya, East Java, Surabaya, Indonesia

Research at a Glance

Premium content — register to unlock

Research at a Glance

Register to unlock

Topics & SDG Alignment

Premium content — register to unlock

Topics & SDG Alignment

Register to unlock

Collaboration

Premium content — register to unlock

Collaboration

Register to unlock

Author Profile (Selected)

Premium content — register to unlock

Author Profile (Selected)

Register to unlock

References Overview

Premium content — register to unlock

References Overview

Register to unlock

Journal & Source

Premium content — register to unlock

Journal & Source

Register to unlock

Metadata & Integrity

Premium content — register to unlock

Metadata & Integrity

Register to unlock