Aldian A. Surbakti
Institut Teknologi dan Bisnis Indonesia, Karo

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

Penerapan Algoritma C4.5 untuk Klasifikasi Tingkat Pemahaman Mahasiswa pada Mata Kuliah Pemrograman Berdasarkan Hasil Evaluasi Akademik Aldian A. Surbakti; Teuku Amanda Rivaldi; Seri Ninta Br Bangun; M. Agustian; David JM Sembiring; Raheliya Br Ginting
Jurnal Sains dan Teknologi Informasi Vol 5 No 4 (2026): September 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/jussi.v5i4.11338

Abstract

The use of academic data through data mining techniques can help identify patterns in student learning outcomes systematically. This study aims to apply the C4.5 algorithm to classify students’ levels of understanding in a programming course based on academic evaluation results. The attributes include attendance percentage, assignment scores, practical scores, midterm examination scores, and final examination scores, with target classes categorized as low, medium, and high. The research employs a classification approach using the C4.5 algorithm through the calculation of Entropy, Information Gain, and Gain Ratio to construct a decision tree. A simulated dataset consisting of 250 records was divided into 200 training records (80%) and 50 testing records (20%). Model performance was evaluated using a confusion matrix, accuracy, precision, recall, and F1-score. The reported test results indicate an accuracy of 78.00%, a macro-average precision of 75.83%, a macro-average recall of 80.61%, and a macro-average F1-score of 77.50%. Of the 50 testing records, 39 were correctly classified, while 11 were misclassified. The low category achieved a recall of 87.50%, the medium category 73.08%, and the high category 81.25%. These findings indicate that C4.5 has the potential to produce an interpretable classification model through decision rules. However, because the study uses a simulated dataset and the target labels are derived from the same evaluation attributes used as predictors, the results cannot yet be generalized to measure students’ actual understanding. Further testing using real academic data and validated class definitions is required to strengthen the validity of the findings.