Explainable Machine Learning for Employee Promotion Prediction: A SHAP-Based Framework for Strategic Human Resource Management

Song Jin, Rong Li, 江贤朋

2025

Data confidence 70 medium

How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

Employee promotion is a critical human resource management decision that impacts organizational performance, employee motivation, and talent retention. This study proposes an explainable machine learning framework for predicting employee promotion eligibility in a large multinational corporation. We compare eight ensemble tree-based models, including XGBoost, LightGBM, Gradient Boosting, Random Forest, AdaBoost, Extra Trees, Decision Tree, and CatBoost, achieving test accuracies ranging from 75.97% to 94.33%. XGBoost demonstrates superior performance with a test AUC of 0.8173 and test accuracy of 0.9433. To enhance model interpretability and support strategic HR decision-making, we employ SHAP (SHapley Additive exPlanations) analysis to identify key promotion determinants. Results reveal that average training score, previous year rating, and department affiliation are the most influential factors. This transparent, data-driven approach provides actionable insights for HR managers to optimize promotion processes and accelerate talent development cycles.