Classification of Student Discipline Levels Using the C4.5 Al-gorithm Based on Violation Points in High School

Abstract

This study aims to classify student discipline levels based on accumulated violation points by implementing the C4.5 decision tree algorithm within the CRISP‑DM framework. The research follows a structured and iterative process, beginning with problem understanding, dataset exploration, and preparation, followed by model training and evaluation. The dataset consists of student demographic information, violation types, and total violation points collected over one academic year. The C4.5 algorithm was selected for its ability to process both categorical and numerical data and to generate interpretable classification rules. The model was trained using a split of training and testing data and further validated using cross‑validation to ensure reliability. The results indicate that the model effectively classifies students into high, medium, and low discipline levels, achieving strong predictive performance. The generated decision tree provides clear and interpretable rules, enabling educators to identify patterns in student behavior and prioritize targeted interventions. These findings highlight the potential of data‑driven approaches to enhance discipline management practices in educational institutions.