Klasifikasi Tingkat Kedisiplinan Siswa Menggunakan Algoritma Machine Learning: Decision Tree, KNN, dan Naive Bayes

Hutabalian, Damri Mulia
Hutabarat, Pebruarianto
Mhd Prasetyo
Irnanda, Mhd Agung
Dalimunthe, Naufal Dhiya Putra
Rosnelly, Rika

Jurnal Komputer Teknologi Informasi Sistem Komputer (JUKTISI)
Vol. 4 No. 3 (2026): Februari 2026

Discipline is a crucial factor influencing the effectiveness of learning processes and the quality of graduates in vocational education. SMK Swasta RK Bintang Timur Pematangsiantar maintains records of student attendance and academic performance that have the potential to be analyzed as indicators of student discipline. However, these data have not been optimally utilized as a basis for decision-making to provide early detection of students who are at risk of declining discipline. This research aims to develop a predictive model of student discipline by identifying patterns of attendance and academic achievement using a data mining approach.The study employs the CRISP-DM framework, consisting of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The dataset includes daily attendance records, semester academic grades, and documented disciplinary behavior used as class labels. Several classification algorithms—Decision Tree (C4.5), KNN, Naive Bayes were implemented to compare model performance. Model evaluation was conducted using confusion matrix, accuracy, precision, recall, and F1-score, with k fold cross-validation.The results show that attendance and academic performance patterns significantly influence the prediction of student discipline levels. The Random Forest algorithm produced the highest performance results, with consistent F1-scores for at-risk student categories. The most influential features include attendance percentage, the number of unexcused absences, and average academic scores. The resulting model is implemented as a decision support prototype dashboard to assist counseling teachers and homeroom teachers in monitoring potential disciplinary violations and planning early intervention. This research is expected to support the development of data-driven discipline monitoring systems in schools and provide practical benefit in preventive actions to improve student behavior quality at SMK Swasta RK Bintang Timur Pematangsiantar.