Predicting Student Performance in Architectural Visualization: A Random Forest Comparison of Core and Elective Student Groups

Abstract

This investigation examines whether student performance patterns in "Architectural Visualization" differ between core students (Multimedia and Creative Technology, MCT) for whom it is mandatory and elective students (Environmental Science and Disaster Management, ESDM) who self-select the course. Using a dataset of 65 students (37 MCT, 28 ESDM), department-wise Random Forest models were developed to identify key performance drivers. Cross-validated models (82.2% accuracy for ESDM, 59.3% for MCT, exceeding baseline by 2. 3 × and 2. 4 × ) revealed distinct learning pathways: ESDM students rely more on continuous engagement (Attendance: 22.1% importance), while MCT students depend on major examinations (Final Exam: 17.8% importance), despite achieving statistically similar final outcomes (p = 0. 605). This indicates students from different academic backgrounds achieve similar outcomes through distinct learning pathways. The analytical framework provides educators with insights for developing targeted pedagogical interventions for both core and elective student groups.