Understanding Employee Turnover Intentions Using Machine Learning: A Multi-Factor Approach

Abstract

Employee turnover continues to challenge organizations and understanding the psychological and workplace factors that shape turnover intention is essential for effective retention efforts. This study integrates Industrial/Organizational Psychology with modern Data Science to examine how attitudes such as job satisfaction, organizational commitment, work engagement, perceived organizational support, perceived organizational justice, work environment, work life balance and employee net promoter score (eNPS) influence employees’ intentions to leave. Using survey data, four modeling approaches were compared, logistic regression, random forests, and a Bayesian Additive Regression Tree (BART) model, alongside an exploratory Automated Machine Learning (AutoML) pipeline that tested additional algorithmic configurations. Variable importance rankings from the machine-learning models were used to identify the most influential predictors contributing to turnover intention. Results show that organizational commitment, perceived support, and engagement are consistently the strongest drivers of turnover intention, and that machine-learning models, supported by AutoML exploration, uncover patterns that traditional methods often overlook. By combining psychological theory with interpretable machine learning, this thesis provides a clearer understanding of turnover behavior and offers HR leaders practical, data-informed guidance for designing targeted retention strategies.