Academic excellence in innovation ecosystems: A predictive approach to university rankings and startup ecosystem performance

Abstract

This study investigates the extent to which institutional attributes derived from global university rankings (QS and THE) influence the performance of territorial innovation ecosystems, as measured by the Global Startup Ecosystem Report (GSER). By integrating 2,145 institutional records linked to dozens of cities featured in all three rankings, the analysis applies feature selection techniques, support vector machine (SVM) regression models, and clustering methods. The results indicate that employability, academic reputation, and internationalization are strongly associated with the dynamism of the startup ecosystem. However, unmodeled contextual factors, such as local innovation policies, venture capital networks, and technological infrastructure, also exert significant influence. The adopted methodological approach combines statistical rigor with predictive capacity, offering valuable insights for data-driven innovation ecosystem planning and institutional strategies aimed at developing startups. From a practical perspective, the findings provide clear guidance for policymakers, university leaders, and innovation stakeholders to design targeted strategies that enhance graduate employability, strengthen institutional reputation, and foster international collaborations, thereby improving the global competitiveness of cities' startup ecosystems. The study further outlines practical directions for policymakers and university leaders, particularly in emerging cities, and recommends future model enhancements incorporating data on technological output, international co-authorship networks, and regional R&D investment.