Evaluating the Interrelationships within the Technological Ecosystem: An SEM-PLS Approach to Innovation Management

Abstract

The objective was to identify the factors that influence entrepreneurial intention among university students in Peru, considering the low youth participation in business activities. A quantitative and basic approach was adopted, with a non-experimental, cross-sectional, and explanatory design, to examine the relationships between the influencing factors (independent variables) and entrepreneurial intention (dependent variable). The target population consisted of university students from a region of Peru. The data analysis employed multivariate regression techniques, correlational analysis, and machine learning methods such as Random Forest and Decision Tree, accompanied by interpretative techniques such as SHAP to evaluate the importance of the factors. Statistical significance criteria (p < 0.05) and coefficients of determination (R2) were used to interpret the results. The results indicated that F5 (access to capital) had the greatest impact on entrepreneurial intention, followed by F2 (internet usage capability) and F3 (entrepreneurial orientation). The Random Forest and Decision Tree models achieved exceptional performance, with an R2 of 0.9943 for the decision tree and an R2 of 0.9880 for Random Forest. On the other hand, ordinal logistic regression showed limited performance, with an AUC-ROC of 0.5337, indicating that it did not effectively capture the relationships among the variables. In conclusion, the factors of entrepreneurial experience and computer capability were key to entrepreneurial intention, and non-parametric models such as Decision Tree and Random Forest proved to be more effective in predicting this intention compared to ordinal logistic regression, validating the importance of integrating training in technical and entrepreneurial skills into educational programs