Unveiling Key Determinants of Science and Technology Policy Effectiveness in Iran: A Machine Learning-Based Analysis Using Soft Computing Techniques

Shababi, Hooman

doi:10.22059/ijms.2025.398164.677769

Unveiling Key Determinants of Science and Technology Policy Effectiveness in Iran: A Machine Learning-Based Analysis Using Soft Computing Techniques

Articles in Press

Document Type : Research Paper

Author

Hooman Shababi

Faculty Member at Rahedanesh Institute of Higher Education

10.22059/ijms.2025.398164.677769

Abstract

Complex national innovation systems (NIS) in developing countries need advanced analytical tools to identify and optimize key drivers of science, technology, and innovation (STI) policy outcomes. This study applies a hybrid machine learning framework—combining Random Forest, Support Vector Machines, K-means clustering, and Principal Component Analysis—to evaluate Iran's STI policy performance from 2010 to 2024. Drawing on multi-source data, including R&D expenditures, university-industry collaboration indices, human capital metrics, and innovation outputs, the proposed model uncovers nonlinear relationships among policy variables and prioritizes the most influential factors. The results reveal that R&D intensity, institutional collaboration quality, policy implementation coherence, and human capital development are the dominant predictors of policy success. By integrating soft computing methods and empirical policy data, this work offers a replicable approach for evidence-based policymaking in emerging economies. The findings align with recent advancements in AI-driven decision support systems and contribute to the growing body of research advocating for data-driven innovation governance.

Keywords

Machine learning

Soft computing

Science and technology policy

National innovation systems

D investment

Subjects