More images

Abstract

Machine learning offers a powerful approach to predicting entrepreneurial tendencies among university students, enabling targeted measures and better decision-making. This research explores a dataset of academic, demographic, and behavioral attributes to classify students into entrepreneurial categories using nine machine learning models. Random Forest emerged as the top-performing model, achieving the highest accuracy of 83%. For better interpretability, Local Interpretable Model-agnostic Explanations (LIME) was applied, revealing key features such as gender, CGPA, study hours, and program involvement that influenced predictions. By integrating predictive accuracy with explainability, the study provides transparent insights into entrepreneurial behavior, helping educators and policymakers in fostering entrepreneurship. Future research can enhance this work by integrating additional features and applying the model to broader student populations.

Contributions

• Designed and conducted the complete research workflow, from problem formulation to final paper preparation.
• Collected the dataset manually through structured surveys using Google Forms and prepared the dataset for analysis.
• Performed data preprocessing and exploratory analysis, including feature preparation and statistical visualization.
• Implemented and evaluated nine machine learning classification models to predict entrepreneurial tendencies.
• Achieved the best performance using Random Forest with 83% accuracy.
• Applied Explainable AI (LIME) to interpret model predictions and identify key influential features.
• Created visualizations and analytical diagrams to better communicate patterns and model performance.
• Authored and formatted the complete research paper, including methodology, experiments, results, and discussion.

Technologies