AI-Based Prediction of Student Dropout Risk Using Behavioural Patterns

– Prediction of student dropout is one of the major challenges faced by higher education institutions with improvements in academic performance, student retention, and institutional ranking. Most of the traditional approaches rely heavily on academic indicators such as grades, attendance, and performance in exams, which fail to capture early changes in behaviour. Here in, this paper describes an AI-based predictive model based on behavioural patterns related to LMS usage, trends in assignment submission, consistency in login, duration of sessions, and metrics related to academic engagement that can enhance the accuracy in the prediction of student dropout. This study compares the algorithms of Logistic Regression, Random Forest, and XGBoost using a real dataset collected from ERP and LMS platforms. The findings reveal that behavioural analytics enhances the early detection of the dropout risk significantly, thus helping in implementing timely academic interventions and strategies for student support.