Machine Learning based Predictive Analytics for Student Dropout and Academic Success

Student attrition in higher education is acknowledged as a major concern that influences both institutional sustainability and the long-term academic and professional paths of learners. In many institutions, rising dropout rates place pressure on academic planning, financial stability, and student support systems, which makes the early identification of vulnerable students increasingly important. In this context, a comprehensive machine learning framework is applied in the present study to predict student dropout and academic success using the well-established dataset from the UCI Machine Learning Repository.A wide range of classification algorithms is systematically examined, including Decision Trees, Random Forest, Support Vector Machines, Logistic Regression, and several ensemble approaches. Each model is evaluated to understand its predictive capability, generalization potential, and suitability for real-world academic monitoring systems. The ensemble model that demonstrated the highest effectiveness achieved an accuracy of 89.7 percent, a precision of 87.3 percent, a recall of 91.2 percent, and an F1-score of 89.1 percent, indicating a strong balance between identifying true dropouts and minimizing misclassification. The feature correlation analysis further indicates that curricular units credited in the first semester at 0.68, admission grade at 0.54, and tuition fee payment status at 0.47 show the most influential predictive associations. These factors reflect academic preparedness, financial consistency, and early academic engagement, all of which contribute meaningfully to student persistence. The overall findings provide practical and data-driven insights that can support the development of early intervention systems. By applying these predictive indicators, institutions can design proactive measures, strengthen academic advising, and offer targeted assistance that helps reduce dropout rates and fosters sustained student success.