This research employs LightGBM and Explainable Artificial Intelligence (XAI) to forecast entrepreneurial competencies in university students, utilizing a dataset comprising 219 responses from university students. The research processed a diverse dataset by integrating standardization and One-Hot Encoding, resulting in 22 predictive features. The LightGBM model demonstrated a promising accuracy of 71%, with distinct precision and recall metrics. Applying XAI methods like LIME and SHAP, the study ensures interpretability and transparency in model predictions, which is crucial for educational policy and curriculum development. The research underscores the significance of data-centric methodologies in evaluating entrepreneurial skills, which is essential for adapting academic frameworks to current economic landscapes. The outcomes are designed to guide educators and policymakers, supporting and cultivating entrepreneurial talent conducive to innovation and societal advancement.