Tools
Tools
Inayat, Ismail (2025) Analysis of Employee Turnover factors using AI models. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (923kB) | Preview |
Abstract
Turnover is a major problem for organizations since hiring and training new staff may be expensive. It is crucial to recognize and handle new hires who may decide to change employment. Turnover-influencing traits have been identified by earlier studies evaluating turnover intention. With the goal of overcoming the limitations of previous research, the current study demonstrated a turnover intention prediction technique based on machines with deep learning. A Human Resources Analytics dataset, currently accessible to the public, was utilized in this study. Four models are tested, such as Ensemble decision tree, tuned KNN, deep scaled MLP, and TabNet. The ensemble decision tree obtained 86.05% accuracy, and the tuned KNN achieved 86.39%, above all existing KNN models. Subsequently, deep scaled MLP behaves sustainably and achieved 96.68%, indicating deep learning strength for extracting representations without manual feature extraction. The TabNet integrated attention mechanisms with feature selection using a deep learning architecture. It achieved 99.57% accuracy highest among all available TabNet algorithms. The highest performance explores an attention-based architecture suited for organized tabular data. Overall results confirmed that deep learning outperformed machine learning algorithms with the IBM HR dataset. Whereas machine learning models are less explainable and simpler. The deep learning models with improved insight features are considered suitable for HR decision making.
Actions (login required)
 |
View Item |