This research is motivated by the number of transactions in the last five years (2019–2023), 80% of students have a level of discipline violation, indicating the need for more strategic discipline. This study aims to analyze the systemic relationship between various types of participants to identify the causes of the problem. The method used in data mining is the FP-Growth algorithm, which is applied to 1,500 historical data points with a minimum support of 0.1 and a confidence level of 0.7. The analysis results show 15 significant pattern associations, with the strongest correlation between "Late → Not doing assignments" (confidence 0.83) and "Truancy → Smoking in school areas" (confidence 0.75, lift 2.5). This forms the basis for data-driven intervention recommendations, such as the implementation of the "Morning Check-in" program and the implementation of supervision in vulnerable areas, which will provide practical support to improve the effectiveness of school discipline management across schools.