Lalu Ganda Rady Putra, Didik Dwi Prasetya, Triyanna Widiyaningtyas
A concept map is a visual tool used to illustrate the relationships between concepts in a domain of knowledge. In an educational context, concept maps play an important role in helping students understand the structure of knowledge and the relationships between ideas. With the increasing adoption of technology in learning, analyzing concept map data is becoming increasingly relevant to evaluate student understanding and support the decision-making process. This study compares three machine learning methods, namely Random Forest (RF), Support Vector Machine (SVM), and SVM with parameter hypertuning (SVM-Hypertuning), in the classification of concept map data. The data used is represented in the form of structural and semantic features of student concept maps. Performance evaluation was conducted using accuracy, precision, recall, and F1-score metrics. The results show that the SVM-Hypertuning method has the best performance compared to RF and standard SVM, with accuracy reaching 91.2%. This research makes an important contribution to the development of analytic models to improve the effectiveness of data-driven educational evaluation. © 2025 IEEE.
Universitas Negeri Malang, Department of Electrical Engineering, Malang, Indonesia
