This study proposes an artificial intelligence–based decision-support framework to optimize innovation and entrepreneurship resource allocation in higher vocational colleges by improving institutional success prediction accuracy. Traditional allocation approaches are largely experience-based and static, lacking objective, data-driven evaluation mechanisms. An Artificial Fish Swarm Algorithm–Enhanced K-Nearest Neighbors (AFSA-EKNN) model is developed by integrating swarm-based parameter optimization with recursive feature elimination, Z-score normalization, and 10-fold cross-validation to enhance prediction robustness and adaptability. Experiments were conducted using a Kaggle dataset comprising aggregated institutional records from 500 higher vocational colleges, including financial, operational, and entrepreneurship-related indicators. The proposed AFSA-EKNN framework achieved a prediction accuracy of 93.0% with a cross-validation loss of 0.070. Feature importance analysis identified annual budget and resource allocation as the most influential factors affecting entrepreneurial success. The proposed framework provides effective data-driven decision support for higher vocational institutions, enabling objective resource allocation, reducing subjective bias, and improving the overall effectiveness of innovation and entrepreneurship education.
Zhou et al. (Sun,) studied this question.