This analysis reveals GNN outperforms other machine learning models in predicting entrepreneurial success in students, suggesting effective support strategies.
In the modern world of innovation, startups play pivotal roles as agents of economic growth, employment, and technological disruption. With learning institutions promoting entrepreneurship, there is an increasing imperative to detect student projects with a long-term potential for success. This research puts forth a predictive framework that applies a high-dimensional set of educational performance, soft skills, behavioral characteristics, and institutional support factors to predict startup success among students. The research compares machine learning models, such as Random Forest, XGBoost, LightGBM, Artificial Neural Networks (ANN), and Graph Neural Networks (GNN), to determine which best represents the complex nature of entrepreneurial success. Interestingly, the GNN model performed the best, with the highest accuracy by successfully learning feature correlations and inter-dependencies. The results have real-world application for universities, incubators, and funding agencies as they facilitate early identification of high-potential student entrepreneurs. This allows for intentional mentoring, focused resource allocation, and the creation of stronger campus-based startup ecosystems.
No takes yet. Share an insight, caveat, or question.
A 2025 study studied this question.