Predicting entrepreneurial intention in Colombian academics: a machine learning approach

Abstract

Purpose

This study aims to analyze the entrepreneurial intention (EI) of Colombian researchers using machine learning (ML) techniques, considering their academic activity, contexts and social norms (SN).

Design/methodology/approach

Unsupervised classification techniques were applied, including principal component analysis, hierarchical clustering with the Ward method and a logistic model to evaluate the classification. This was done to group researchers according to their characteristics and EI.

Findings

The methodology used allowed the identification of three groups of academics with distinct characteristics, of which two showed a high presence of EI. The results indicate that EI is influenced by the connection with the private sector (consulting, intellectual property and applied research) and by the lack of institutional support from universities. Regarding SN, only the preference for entrepreneurial activity over being an employee and the social appreciation of entrepreneurial dedication were identified as predictors of EI.

Originality/value

The use of ML techniques to study the EI of researchers is uncommon. This study highlights the ability of the methodology used to identify differences between two groups of academics with similar characteristics but different levels of EI. One group was identified that, despite rejecting values associated with entrepreneurs, has a high predisposition to develop a career as an entrepreneur. This provides valuable information for designing policies that promote EI among Colombian researchers.