AN EXPLAINABLE AI-DRIVEN STACKED ENSEMBLE FRAMEWORK FOR PREDICTING UNIVERSITY STUDENTS' ENTREPRENEURIAL POTENTIAL

ABSTRACT

It is important to identify university students with authentic entrepreneurial potential, since that spurs innovation, enables self-employment, and helps the economy keep improving. Conventional statistical methods and many machine learning models either do not reflect the complex determinants of entrepreneurship or achieve very high prediction accuracy without clarifying the basis for their decisions. An ensemble that stacks multiple components brings Explainable Artificial Intelligence together with the task of estimating entrepreneurial potential, grounded in psychological, demographic, behavioral, and environmental proof. This framework integrates Recursive Feature Elimination, ANOVA F-test feature selection, SMOTE, stacked ensemble learning models, plus a Logistic Regression meta-learner. SHAP and LIME deliver global and local explanations that raise transparency and solidify user trust. A FastAPI-based web application deployment supports real-time decision-making, providing accurate and interpretable predictions for universities, policymakers, entrepreneurship educators, and investors.

Keywords: Entrepreneurial Potential, Explainable Artificial Intelligence, Stacked Ensemble Learning, SHAP, LIME, Machine Learning.

AN EXPLAINABLE AI-DRIVEN STACKED ENSEMBLE FRAMEWORK FOR PREDICTING UNIVERSITY STUDENTS’ ENTREPRENEURIAL POTENTIAL

AN EXPLAINABLE AI-DRIVEN STACKED ENSEMBLE FRAMEWORK FOR PREDICTING UNIVERSITY STUDENTS’ ENTREPRENEURIAL POTENTIAL

AN EXPLAINABLE AI-DRIVEN STACKED ENSEMBLE FRAMEWORK FOR PREDICTING UNIVERSITY STUDENTS' ENTREPRENEURIAL POTENTIAL