Beyond the Average Entrepreneur: A Machine Learning and Latent Class Approach to Understanding Entrepreneurial Diversity Using Kauffman Data

Entrepreneurship is increasingly recognized as a complex and heterogeneous phenomenon involving diverse pathways, characteristics, and resource configurations. This study moves beyond the traditional focus on the average entrepreneurial profile by identifying hidden entrepreneurial profiles and patterns using the Kauffman Index of Entrepreneurial Activity (KIEA). The study applies an advanced analytical framework combining Exploratory Factor Analysis (EFA), Latent Class Analysis (LCA), machine learning classification techniques, Explainable Artificial Intelligence (SHAP), and Generalized Additive Models (GAM) to uncover the complexity of entrepreneurial participation. The findings reveal that entrepreneurs are not a homogeneous group but consist of distinct categories shaped by differences in human capital, financial resources, demographic characteristics, and social conditions. The analysis further demonstrates that advanced data-driven approaches provide deeper insights into entrepreneurial pathways by identifying hidden structures, improving prediction, and explaining individual-level entrepreneurial outcomes. The study contributes to entrepreneurship research by presenting a multidimensional perspective that recognizes entrepreneurship as a combination of different profiles rather than a single universal process.