Student Retention Risk Prediction Using Machine Learning and Educational Analytics

This research presents a machine learning-based approach for predicting student retention risk using academic performance data. Educational institutions often face challenges in identifying students who are likely to experience academic difficulties or drop out. Early detection of such students enables institutions to provide timely support and improve overall retention rates. The proposed system analyzes several academic indicators including attendance percentage, semester grade point averages (GPA), backlog count, and student participation in extracurricular events. These attributes are processed using data preprocessing and feature engineering techniques to generate meaningful predictive features. A Random Forest Regression model is implemented to estimate a risk probability score for each student based on their academic performance. The predicted risk score ranges between 0 and 1 and is further categorized into Low Risk, Medium Risk, and High Risk groups. The model is evaluated using statistical metrics such as R² Score, Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and cross-validation to measure prediction accuracy and generalization capability. In addition to the predictive model, an interactive academic risk dashboard is developed to visualize student performance trends, attendance impact, backlog influence, and feature importance. This dashboard allows educators to easily monitor student progress and identify at-risk students who may require academic intervention. The proposed system demonstrates that machine learning and educational analytics can significantly support data-driven decision making in higher education institutions. By identifying potential academic risks early, institutions can implement targeted interventions and improve student retention outcomes.