Exploiting Machine Learning Techniques in Human Resource Management: A Descriptive Research

الملخص

This study provides a descriptive review of the role of artificial intelligence (AI) techniques in human resource (HR) management, with a specific focus on machine learning algorithms used for prediction and classification tasks. By comprehensively analyzing 8 studies published between 2020 and 2024 across diverse geographic contexts (e.g., Bangladesh, China, Portugal, and international platforms like Kaggle), the research compares the effectiveness of various algorithms in HR tasks such as performance evaluation, turnover prediction, and promotion decisions.

The Random Forest (RF) algorithm emerged as the most effective tool, achieving the highest predictive accuracy (ranging from 0.713 to 0.982) across all scenarios, outperforming other algorithms like Logistic Regression (LR), Support Vector Machines (SVM), and Artificial Neural Networks (ANN). Key findings include:

• In a Bangladesh-based study (sample: 1,109 employees), RF achieved 98.2% accuracy in performance evaluation.
• In a Portugal-based study (sample: 199 employees), RF outperformed other models with 78.0% accuracy for turnover prediction.
• In a China-based study (sample: 287,229 employees), RF maintained 71.3% accuracy despite dataset complexity.

The study offers practical recommendations for HR professionals, such as:

• Adopting RF-based models to enhance decision-making accuracy.
• Improving data quality and reducing bias through standardized data collection.
• Training HR teams to leverage analytical tools effectively.

Future research directions are identified, including:

• Testing models across diverse industries (e.g., healthcare vs. technology).
• Addressing ethical challenges like data privacy and algorithmic bias.
• Integrating advanced techniques (e.g., deep learning) for imbalanced datasets.

By clarifying the scope of the reviewed literature (size, timeframe, and geographic diversity), this study enhances its methodological credibility while acknowledging limitations, such as small sample sizes in some studies (e.g., 199 employees in Portugal).