Abstract
This study introduces a novel risk assessment model for university student innovation and entrepreneurship, grounded in decision tree (DT) methodology. It tackles the challenges faced by traditional models in merging multi-source data and understanding nonlinear relationships. This advanced approach aims to enhance both the precision and reliability of risk evaluations in the context of student-led entrepreneurial ventures. From the four dimensions of entrepreneurial environment, entrepreneurial education, entrepreneurial groups, and entrepreneurs, relevant college student innovation and entrepreneurship data was collected, and the collected data was preprocessed to select the most relevant feature from all available features. The C4.5 algorithm was optimized by cross validation to determine the depth of the number and the minimum sample size of leaf nodes, and a post-pruning strategy was adopted. The optimized C4.5 model was compared with Iterative Dichotomiser 3 (ID3), Classification and Regression Trees (CART), and C4.5 model, and risk assessment was applied to three entrepreneurial plan instances. The experimental findings indicated that the optimized C4.5 model had an average accuracy rate of 90.7% for the risk classification of college students’ innovation and entrepreneurship, and could accurately assess the risk of multiple entrepreneurial conditions in a comprehensive entrepreneurial plan.