Employing decision trees to predict cyberbullying victimization among Chinese adolescents and identify subgroups and their shared characteristics

Early research has revealed distinct subgroups of cyberbullying victims. However, due to the limitations of traditional statistical methods, the characterization of features in the subgroups has been relatively limited, making it challenging to gain a relatively comprehensive understanding of different subgroup members. Decision trees and machine learning techniques offer notable advantages in addressing such issues. The primary aim of this study is to develop a high-performing classifier based on self-reported data from 814 middle school students to accurately predict cyberbullying victimization and uncover the most influential factors. On this basis, the study attempts to identify different subgroups of cyberbullying victims and their shared characteristics. The results indicated that the classification tree achieved a prediction accuracy of approximately 80