Green AI for Sustainable Employee Attrition Prediction: Balancing Energy Efficiency and Predictive Accuracy

This study investigates the application of Green Artificial Intelligence (AI) principles to employee attrition prediction models, aiming to reduce computational energy consumption while maintaining comparable predictive accuracy to conventional approaches. The research addresses a critical gap at the intersection of sustainable AI and human resource management. A mixed factorial design was employed, evaluating six machine learning algorithms (Logistic Regression, Random Forest, XGBoost, Support Vector Machine, K-Nearest Neighbors, Decision Tree) in both conventional and Green AI-optimized versions, across three feature selection methods. Experiments were conducted on the IBM HR Analytics Employee Attrition & Performance dataset (N=1,470), with energy consumption (kWh/Wh) measured using OpenZmeter, CodeCarbon, and CarbonTracker, alongside standard performance metrics (Accuracy, F1-score, AUC-ROC). Results indicate that Green AI models achieved a significant average energy reduction of 44.8% during training and 35.6% during inference compared to conventional models. Crucially, these substantial energy savings were realized with only minimal and statistically non-significant differences in predictive performance (e.g., F1-score: Green AI M=0.62 vs. Conventional M=0.64, p=.07). Specific Green AI strategies, including model parameter optimization and feature selection, effectively reduced computational load while preserving performance. Optimized XGBoost models notably demonstrated a strong balance of high accuracy and reduced energy consumption. This research provides compelling empirical evidence that sustainable AI practices are feasible and effective in HR analytics, offering a viable pathway for organizations to mitigate their carbon footprint and operational costs without compromising the quality of predictive insights.