Advances in analytics have created opportunities for entrepreneurship researchers to apply machine learning techniques to address entrepreneurship questions, contributing to entrepreneurship theory and practice. Although these opportunities have been recognised in the business and entrepreneurship literature, challenges remain which limit their adoption in entrepreneurship research. This study aims to elaborate on these challenges, and to illustrate some of the opportunities available from the application of machine learning techniques to entrepreneurship research questions. Drawing on data from the Global Entrepreneurship Monitor (GEM), this study adopts a machine learning methodology to examine the relative importance of the determinants of entrepreneurial intentions (EI) and total early stage entrepreneurship (TEA), and the ability to predict these outcomes. The machine learning approaches are compared with a more traditional regression model. The results show that the more advanced algorithm has higher predictive accuracy, and also provides variable importance measures, which enable us to examine the dominance of determinants. Across all models, self-perceptions, experience and age are found to be relatively more important determinants, with cultural factors and other demographics less important. Overall, TEA can be modelled more accurately than EI, but it remains challenging to accurately predict both EI and TEA. The results contribute to our understanding of the determinants of EI and TEA, as well as highlighting the application of the machine learning methodology.
|Publication status||Published - 25 Aug 2021|
|Event||Irish Academy of Management Conference 2021|
- Dublin, Ireland
Duration: 25 Aug 2021 → 26 Aug 2021
|Conference||Irish Academy of Management Conference 2021|
|Period||25/08/2021 → 26/08/2021|