Classification of Medical Students' Skills Using the Random Forest Method Based on Practicum and Theory Scores

Abstract

This study classifies the skill levels of medical students based on their practical and theoretical scores using the Random Forest method. The data consist of secondary data obtained from the Faculty of Medicine, Universitas Muhammadiyah Sumatera Utara, co vering students from the Medical Education Study Program across three cohorts (2021, 2022, and 2023) in the block “Basic Organ System of Special Sense, Reproduction and Urinary Tract.” The dataset includes 1,074 students with five assessment features: block exam score (UAB), practical score (average of Histology, Anatomy, Physiology, and Biochemistry), tutorial score, SGD attitude score, and PIM attitude score. The Random Forest method is used to classify students into “Good” (final score ≥ 75) and “Poor” (final score < 75) skill categories. The results indicate that the model achieves an accuracy of 93.33%, with precision values of 94.12% for the “Good” category and 90.00% for the “Poor” category, as well as recall values of 97.56% and 78.26%, respectively. The most influential features are the block exam score (0.378), practical score (0.295), and tutorial score (0.192). The study also generates 11 expert-validated classification rules (average score 4.69/5.00) that can support early identification of students with lower skill levels. The Random Forest model demonstrates effectiveness and consistency, achieving accuracy above 92% across all cohorts, and supports the development of a machine learning–based evaluation system for medical students at Universitas Muhammadiyah Sumatera Utara.