Exploring entrepreneurial phases with machine learning models : evidence from Hungary

Objective: The article aims to explore the potential differences between the two phases of entrepreneurship, i.e., total early-stage entrepreneurial activity and established business, as defined by the Global Entrepreneurship Monitor (GEM). The study aimed to classify entrepreneurs using various machine learning models and to evaluate their classification performance comparatively. Research Design & Methods: Using the Hungarian GEM datasets from 2021 to 2023, we analysed a subsample of 964 entrepreneurs. Due to inconsistent results from traditional analyses (e.g., correlations, regressions, principal component analyses), we employed machine learning approaches (supervised learning classification methods) to uncover latent relationships between variables. Findings: The study utilized seven machine learning classification methods to examine the feasibility of grouping companies within the sample using Hungarian GEM data. Findings indicate that machine learning techniques are particularly effective for classifying businesses, although the performance of each method varies significantly. Implications & Recommendations: These results provide valuable insights for researchers in selecting methodologies to identify various business phases. Moreover, they offer practical benefits for market research professionals, suggesting that machine learning techniques can enhance the classification and understanding of entrepreneurial phases. Contribution & Value Added: The study adds to the existing body of knowledge by demonstrating the effectiveness of machine learning methods in classifying business phases. It highlights the variability in performance across different machine learning techniques, thereby guiding future research and practical applications in market research and entrepreneurship studies.