Bullying is one of the main risk factors for mental health disorders in adolescents and is significantly correlated with increased suicidal ideation. This study aims to develop a predictive model of suicidal ideation risk in bullying victims using a questionnaire data-based machine learning approach as the basis for the development of an early warning system. The study used a predictive quantitative design involving 350 respondents who had experienced bullying. The variables analyzed included demographic factors, experiences of bullying, as well as psychological indicators such as depression, anxiety, stress, self-esteem, and social support. Four classification algorithms were compared, namely Logistic Regression, Random Forest, Support Vector Machine, and XGBoost. The results show that XGBoost has the best performance with an accuracy of 91% and a ROC-AUC of 0.94. The most influential variables on risk prediction were depression scores, social support, anxiety, and bullying frequency. These findings show that the machine learning approach is effective in supporting early detection of the risk of suicidal ideation and has the potential to be implemented as an early warning system in the educational environment.