MACHINE LEARNING MODELS FOR EMPLOYEE PERFORMANCE PREDICTION: INTEGRATING PSYCHOMETRICS AND MANAGERIAL DECISION-MAKING

Abstract

Predicting employee performance has emerged as a critical priority for modern organizations seeking to enhance productivity, optimize talent management, and strengthen strategic decision-making. Traditional evaluation methods, often subjective and inconsistent, fail to capture the complex interplay of cognitive, behavioral, and environmental factors influencing performance. This study introduces an integrated machine learning (ML) framework that combines psychometric assessment data covering personality traits, motivation, and emotional intelligence with managerial decision parameters such as leadership evaluation and peer feedback. A dataset of employee performance indicators was analyzed using supervised ML algorithms including Random Forest, XGBoost, and Artificial Neural Networks (ANN) to predict performance categories. Feature importance and SHAP interpretability were applied to assess model transparency and fairness. The hybrid model demonstrated superior accuracy (up to 89%) and enhanced interpretability compared to traditional regression-based approaches. Findings suggest that psychometric variables significantly contribute to performance prediction, accounting for 42% of the predictive power, while managerial assessments add contextual refinement. The study underscores that integrating data-driven learning with human judgment offers a robust and ethical pathway for talent forecasting, minimizing bias and improving organizational decision-making reliability.