Abstract
Purpose : The present study is an effort to improve entrepreneurship research by shifting from traditional theory-based and socio-cultural explanations to a data-driven, interpretable decision-modeling framework. It demonstrated the complex behavioral patterns that influence entrepreneurship skills using simple and transparent analytical techniques. Methodology : In this research, a new framework known as the hybrid fuzzy cognitive system–entrepreneurial decision model (FCSM-EDM) has been used. This framework used fuzzy logic to create rules based on entrepreneurship traits such as risk appetite, innovativeness, leadership skills, resilience, and business knowledge. It also utilized Shapley Additive exPlanations (SHAP)-based interpretability, Sankey visual reasoning, and explainable AI for drawing clear insights for entrepreneur decision-making. Findings : Results from the working model revealed a fuzzy accuracy of 79.99%, an interpretability score of 0.88, a precision of 0.75, and a high interpretability-precision tradeoff score (IPTS) of 0.809. The model exhibited RUC-AUC of 68%, a bootstrapped mean accuracy of 0.66, and a tiny standard deviation of 0.1999, which shows that it was both stable and reliable. Practical Implications : The model presented a precise and comprehensible decision-support system that allows educators, politicians, incubators, and financial institutions to evaluate entrepreneurial potential. It improved the design of evidence-based interventions, the identification of talent, and entrepreneurship training suited to various behavioral profiles. Originality : This study introduced a novel integration of fuzzy decision modeling, SHAP-based interpretability, and Sankey visual analytics in the field of entrepreneurship research. It offered a scalable and interpretable alternative for entrepreneurial and corporate decision-making applications by outperforming Bayesian fuzzy systems, fuzzy-ANFIS hybrids, random forests, and SVM-RBF models.

This publication has 31 references indexed in Scilit: