Abstract:
This study presents a machine-learning framework
to predict employee performance levels, empowering HR
professionals with data-driven insights for eTective talent
management. Leveraging a comprehensive dataset
encompassing demographics, job roles, engagement metrics,
training history, and historical performance ratings, the
research explores multiple algorithms, including LightGBM,
XGBoost, XGBoost with SMOTE, and Random Forest. To
address class imbalance, the Synthetic Minority Over
sampling Technique (SMOTE) was implemented, generating
synthetic samples to enhance prediction accuracy across all
classes. Feature selection and importance analysis identified
key performance predictors, such as tenure, engagement
scores, work-life balance, and satisfaction levels. Among the
evaluated models, Random Forest achieved the highest
accuracy (94%) with balanced class performance, making it
the preferred choice for deployment. This research
underscores the transformative role of machine learning in
HR practices, providing actionable insights to design targeted
development programs, optimize employee performance, and
improve organizational outcomes.
