An Explainable Artificial Intelligence Framework for Predicting Green Entrepreneurial Intention Through Policy Support, Higher Education, and Entrepreneurial Self-Efficacy

Green entrepreneurship is increasingly recognized as a pathway toward sustainable economic development, yet the predictive contributions of institutional, educational, and psychological factors remain insufficiently understood. This study evaluated an explainable artificial intelligence framework for predicting green entrepreneurial intention through government policy support, higher education exposure, and entrepreneurial self-efficacy. Questionnaire data from 1,046 respondents across Nigeria, Ghana, and South Africa were examined using descriptive statistics, reliability assessment, correlation analysis, exploratory factor analysis, and machine learning. Predictive performance was assessed through five-fold cross-validation using Elastic Net, Random Forest, Histogram Gradient Boosting, and Support Vector Regression. Although green entrepreneurial intention exhibited the highest mean score (3.8713), internal consistency was inadequate across all constructs, with Cronbach's alpha ranging from −0.1092 to 0.0503. Factorability diagnostics revealed weak sampling adequacy (KMO = 0.5088) and a nonsignificant Bartlett's test (p = 0.8399). All predictive models produced negative cross-validated R² values, indicating no improvement over baseline prediction. Cross-country validation further demonstrated limited generalizability. These findings underscore the importance of establishing measurement validity before applying explainable artificial intelligence to entrepreneurial behavior. Future research should prioritize validated instruments, cross-national measurement equivalence, and external predictive validation to support reliable, interpretable, and evidence-based green entrepreneurship decision-making.