Akademik Veri Yönetim Sistemi
Journal of the Knowledge Economy, 2026 (Scopus)
Drawing on the national innovation systems (NIS) perspective, this study uses the Global Innovation Index (GII) as an integrated measure of countries’ innovation capabilities and outcomes to examine how different innovation enablers jointly shape innovation performance. Using GII data for 2013–2022, we apply ensemble machine learning algorithms to analyze how the GII input and output dimensions, treated as NIS elements, predict overall GII scores. Countries are grouped into five subsamples (all countries, G20, European countries, all countries during the COVID-19 period, and low-income countries) to capture contextual variation in innovation systems. The results show that innovation performance is driven not by single factors but by context-specific combinations of enablers. Research and development and creative/intangible outputs form a powerful configuration for most countries, whereas information and communication technologies, credit availability and innovation linkages become particularly salient during COVID-19. In low-income countries, tertiary education and creative outputs emerge as a distinctive configuration associated with higher innovation performance. These findings demonstrate that the relative importance and effective combinations of NIS elements vary systematically across country groups and over time. By mapping these patterns, the study advances NIS research methodologically and substantively, showing how machine learning can uncover non-linear, context-dependent configurations of innovation enablers and informing tailored policy interventions for different country groups.