Big data-based framework for prediction of employee attrition by using deep data people analytics

Abstract

The phenomenon of Employee turnover has a negative impact on profit management accuracy across many industries within the business sector. The utilisation of contemporary advanced computing technology enables the development of a predictive model for staff attrition, hence facilitating cost savings for business owners. Although there is a lack of evaluation of these models in real-world scenarios, various implementations were created and utilisedin the IBM HR Employee Attrition list to examine the potential integration of these models they are put into a decision support system and how they affect making strategy decisions. In this research, a neural network based on Transformers was utilised as a computational method to analyse staff turnover. The network was distinguished by its ability to adapt contextual embeddings to tabular data. The experimental findings provided evidence that this particular model had superior performance compared to other contemporary models in terms of predictive accuracy. Moreover, the present research has revealed that deep learning, specifically Transformer-based networks, exhibit considerable potential in addressing the challenges associated with the presence of tabular and imbalanced data.