Intelligent Student Dropout Analytics System Using Random Forest, SHAP Explainability and AI-Based Personalized Intervention

Student dropout continues to be a significant challenge for higher educational institutions, making early identification of at-risk students an essential requirement for improving academic success and institutional retention. This paper presented an Intelligent Student Dropout Analytics System that combines machine learning, explainable artificial intelligence, and personalized intervention strategies within a unified educational analytics framework. The proposed system employs a Random Forest classifier to predict student dropout risk using both academic and behavioural attributes obtained from the Kaggle Student Dropout Prediction Dataset. Unlike traditional prediction systems, the proposed framework integrates SHAP Explainability to provide transparent and interpretable predictions, enabling faculty members to understand the factors contributing to each student's risk level. The incorporation of an AI-Based Intervention Advisor and a What-If Simulation module further extends the practical usefulness of the system by supporting personalized counselling and evidence-based academic planning. The developed Streamlit dashboard provides an interactive environment for individual student analysis, batch prediction, cohort monitoring, and early warning visualization. Experimental evaluation demonstrated that the proposed model achieved 91.4% prediction accuracy with an AUC-ROC score of 0.94, confirming its effectiveness in identifying students who require timely academic support. Overall, the proposed system serves as an efficient decision-support tool that enables educational institutions to adopt proactive retention strategies and improve student success.

[1]   J. Pattanaphanchai, K. Leelertpanyakul, and N. Theppalak, “The Investigation of Student Dropout Prediction Model in Thai Higher Education Using Educational Data Mining: A Case Study of Faculty of Science, Prince of Songkla University,” Journal of University of Babylon for Pure and Applied Sciences, vol. 27, no. 1, pp. 356–367, 2019. doi: 10.29196/jubpas.v27i1.2191. [2]      R. S. Baker and K. Yacef, “The State of Educational Data Mining in 2009: A Review and Future Visions,” International Journal of Educational Technology in Higher Education, vol. 6, pp. 1–14, 2009. [3]  S. Lee and J. Y. Chung, “The Machine Learning-Based Dropout Early Warning System for Improving the Performance of Dropout Prediction,” Applied Sciences, vol. 9, no. 15, 2019. doi: 10.3390/app9153093. [4]    K. Arnold and M. Pistilli, “Course Signals at Purdue: Using Learning Analytics to Increase Student Success,” in Proceedings of the 2nd International Conference on Learning Analytics and Knowledge (LAK), 2012, pp. 267–270. [5]   K. Danilowicz-Gösele, K. Lerche, J. Meya, and R. Schwager, “Determinants of Student Success at University,” Education Economics, vol. 25, no. 5, pp. 513–532, 2017. [6]       L. Breiman, “Random Forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001. [7]     L. Kemper, G. Vorhoff, and B. U. Wigger, “Predicting Student Dropout: A Machine Learning Approach,” European Journal of Higher Education, 2018. [8]  A. Tayebi, J. Gomez, and C. Delgado, “Analysis on the Lack of Motivation and Dropout in Engineering Students in Spain,” IEEE Access, vol. 9,   pp.   66253–66265,   2021.   doi: 10.1109/ACCESS.2021.3076751. [9]  L. Paura and I. Arhipova, “Student Dropout Rate in Engineering Education Study Program,” Engineering for Rural Development Conference Proceedings, 2016. [10]    B. Sekeroglu, K. Dimililer, and K. Tuncal, “Student Performance Prediction and Classification Using Machine Learning Algorithms,” in Proceedings of the ACM International Conference, 2019, pp. 7–11. doi: 10.1145/3318396.3318419.