<![CDATA[This study aimed to detect academic integrity anomalies among twelfth-grade students at an Islamic senior high school (Madrasah Aliyah Negeri) using a Python-based data mining approach. Academic monitoring had previously been conducted manually, which risked delaying the identification of students who required guidance. This research applied the Isolation Forest algorithm with an unsupervised learning approach to analyze midterm and final exam scores, project scores, report averages, attendance percentages, disciplinary violation records, and school counseling records from 650 students. The analysis followed the CRISP-DM methodology, covering six stages from business understanding to deployment. The anomaly detection results were further validated using a rule-based approach, producing a more objective and contextual hybrid validation process. The model identified 52 students (8%) as anomalous data. Hybrid validation classified 544 students (83.69%) as Safe, 97 students (14.92%) as Needing Monitoring, and 9 students (1.39%) as Needing Follow-up. The results were visualized through a Streamlit-based dashboard that can support school decision-making in objective, data-driven academic monitoring.]]>
Copyrights © 2026
