Enhancing workforce retention in the engineering sector: machine learning-driven turnover prediction models
Rania Sawalma et al.
Abstract
Purpose Employee turnover continues to pose a significant challenge for engineering organizations, particularly in emerging economies where talent shortages and project-based work structures heighten retention pressures. While Machine Learning (ML) techniques have demonstrated strong potential in forecasting turnover, empirical benchmarks tailored to specific sectors remain scarce, especially within developing regions. The purpose of this study is to design and rigorously validate ML models capable of predicting voluntary turnover intention in Jordan's engineering sector, and to translate these predictive insights into practical, evidence-based retention strategies. Design/methodology/approach Data were gathered through a structured questionnaire administered to 241 employees across engineering firms in Jordan. Several supervised ML classifiers: Logistic Regression, Random Forest, Extra Trees and LightGBM, were trained and evaluated using a holdout validation strategy (70/30 split) and a suite of performance metrics. Prior to model development, the reliability and construct validity of the survey instrument were thoroughly assessed. Findings LightGBM emerged as the most accurate prediction model, achieving an accuracy of 75.3%, a recall of 81.4% and an F1 score of 79.6%. Key predictors of turnover included organizational commitment, training opportunities and career development. This indicates that ensemble-based models delivered the strongest predictive performance. Organizational Commitment (OC), career development pathways and access to training opportunities emerged as the most influential determinants of turnover intention. Originality/value This study contributes to a validated turnover-related dataset specific to the engineering sector in an emerging economy, a robust ML benchmarking framework grounded in cross-validation and a translation of predictive outputs into actionable guidelines to support strategic HR decision-making.
Evidence weight
Balanced mode · F 0.40 / M 0.15 / V 0.05 / R 0.40
| F · citation impact | 0.50 × 0.4 = 0.20 |
| M · momentum | 0.50 × 0.15 = 0.07 |
| V · venue signal | 0.50 × 0.05 = 0.03 |
| R · text relevance † | 0.50 × 0.4 = 0.20 |
† Text relevance is estimated at 0.50 on the detail page — for your query’s actual relevance score, open this paper from a search result.