Hybrid Clustering and Classification Approaches in Motivation Profiling: A Comparative Study

Understanding students' motivation is essential for educators to personalize learning experiences effectively. Motivation significantly influences students' effort, persistence, and engagement, which are critical for academic success. However, capturing the nuanced and dynamic patterns of student motivation remains a challenging task, necessitating advanced analytical approaches. This research investigates hybrid machine learning methods to profile student motivation more accurately by integrating clustering and classification techniques. A comparative analysis of multiple clustering and classification algorithms was conducted to identify the most effective combinations for motivation profiling. The clustering methods included KM, DBSCAN, and AHC, while the classification methods involved NB, DT, SVM, XGB, and MLP. The findings highlight that DBSCAN combined with MLP achieves high classification performance, attaining 100% accuracy and F1-score in mental effort, 82.86% accuracy with an 80.94% F1-score in persistence, and 75.22% accuracy with a 74.59% F1-score in active choice. The novelty lies in integrating DBSCAN and MLP to optimize student motivation profiling.