The role of students is crucial in determining the success of education. However, not all students can complete their studies on time, leading to the risk of dropping out. Dropping out, or the termination of student status, occurs when a student's status is revoked for certain reasons determined by the university. This phenomenon poses a challenge for institutions as it leads to significant losses for both students and the institution. To address this issue, preventive measures are needed, such as predicting potential dropouts using classification techniques like XGBoost and SHAP Values. XGBoost is a classification algorithm that applies boosting techniques to enhance the performance of weak models until they become stronger. After forming the XGBoost model, interpretation is carried out by analyzing the average contribution of each variable using SHAP Values. The study, which utilized data from 13,552 students, revealed that the number of semesters completed, preparatory GPA, and overall GPA are significant factors in determining whether a student will drop out. The optimal RMSE achieved was 0.01084, resulting from hyperparameter tuning. The XGBoost model demonstrated excellent performance with 100% accuracy, 100% sensitivity, and 100% specificity. The dashboard for dropout prediction includes four menus: A dashboard to view student characteristics with imbalanced and balanced data, a menu displaying overall student data, a predictions menu for input variables to predict student status, and a variable contribution menu to assess the influence of variables on dropout prediction. This interactive tool provides valuable insights for early intervention and prevention of student dropouts. © 2025 Author(s).