Machine Learning-Based Classification of Employee Turnover Factors

Tee, Yen Xuan (2025) Machine Learning-Based Classification of Employee Turnover Factors. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text RMM_Tee Yen Xuan_Full Text.pdf Restricted to Registered users only Download (4MB)

Abstract

Employee attrition, particularly voluntary turnover, presents a significant challenge for organizations, leading to operational disruptions, increased costs, and reduced productivity. Despite its critical impact, many organizations struggle to understand the underlying causes of attrition and predict which employees are most likely to leave. This study addresses this gap by developing a predictive model using statistical and machine learning techniques to identify key factors driving employee turnover. The primary objectives are to explore the dataset to uncover the factors most strongly associated with attrition, evaluate various machine learning algorithms, and build an effective model for predicting employee departure. The methodology includes data cleaning, exploratory data analysis (EDA), outlier treatment, feature encoding, and feature selection using SHAP. Several machine learning algorithms such as Logistic Regression, XGBoost, SVM, Random Forest, LDA, and AdaBoost are applied, with hyperparameter tuning performed through Grid Search and Bayesian optimization. AdaBoost outperformed the other models tested, achieving the highest accuracy, precision, specificity, and AUC score, demonstrating superior predictive performance. Key factors influencing attrition include marital status, job level, work-life balance, and remote work. The study concludes by deploying the AdaBoost model in a user-friendly interface for HR teams, enabling proactive retention strategies. This research provides organizations with actionable insights to reduce attrition, improve employee satisfaction, and optimize workforce management, ultimately enhancing overall business performance. Keywords: Employee attrition, voluntary turnover, machine learning, predictive model, turnover prediction, AdaBoost, feature selection, HR analytics.