Graduate employability remains an issue with high employment rates in Malaysia, where underemployment and skill mismatch being the main issues. This study used ensemble learning models for the predictive analytics in forecasting the employability graduates. It focuses on individual profiles, academic performance, soft skills, and ICT skills using data from the Skim Pengesahan Graduan (SKPG) for the period of 2016-2023. Feature selection was used prior to modeling and compared with different machine learning algorithms, namely Naïve Bayes, Decision Trees, and Gradient Boosting. Among these comparisons, Extreme Gradient Boosting (XGBoost) outperformed the others with the highest predictive accuracy. The XGBoost model selects important factors of employability and makes practical recommendations to bridge the gap in skills prevalent among graduates. These findings demonstrate the importance of data-driven approaches in policy making and graduate readiness initiatives to align education acquisition with the demands of the labor market. © 2025 IEEE.
Author Keywords
Index Keywords
