Scholarlink

<p>Employee turnover has become a crucial issue in organizations, with a direct impact on marketing, workforce stability, and long-term human resource planning. Correctly forecasting employee turnover can also help organizations plan proactive retention strategies, employee-specific Human Resource (HR) advice, and internal marketing policies based on engagement. Nevertheless, noisy data, redundant features, and uncertainty in employees’ behavior limit the performance of traditional prediction models. To overcome these issues, a hybrid, innovative structure called the Bayesian Variational Coupled Modular Neural Network with Leech Growth Algorithm (BVCMNN-LGA) is proposed to predict employee turnover. Initially, the preprocessing phase is done using Improved Adaptive Iteratively Reweighted Penalized Least Squares (IAIRPLS), and then followed by the Basketball Team-Wolf Bird Optimization Algorithm (BT-WBOA). Finally, optimal feature selection is accomplished, producing a smaller number of highly discriminatory features. These characteristics are then used to feed the Bayesian Variational Coupled Modular Neural Network (BVCMNN), which is used for predicting complicated behavior exhibited by employees. Finally, the Leech Growth Algorithm (LGA) is used to minimize the loss incurred in BVCMNN, thereby ensuring stability in convergence. The predicted turnover probabilities are further translated into actionable IM-HR recommendations to improve employee retention and workforce engagement. Experimentation with Employee Turnover data indicates that the proposed BVCMNN-LGA has better performance with an accuracy of 99.97%, precision of 99.95%, recall of 99.96%, F1-score of 99.94%, specificity of 99.93%, and a low error rate of 0.01% and lower computation time of 0.02s, which confirms its ability to be used in intelligent HR decision support.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Employee turnover prediction in internal marketing and human resource recommendations using a leech growth optimized Bayesian variational coupled modular neural network

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • G. Gokul Kumari,
  • Layla Nasser,
  • Eman Alyami,
  • Sulafah M. Basahel,
  • Bhabajyoti Saikia

摘要

Employee turnover has become a crucial issue in organizations, with a direct impact on marketing, workforce stability, and long-term human resource planning. Correctly forecasting employee turnover can also help organizations plan proactive retention strategies, employee-specific Human Resource (HR) advice, and internal marketing policies based on engagement. Nevertheless, noisy data, redundant features, and uncertainty in employees’ behavior limit the performance of traditional prediction models. To overcome these issues, a hybrid, innovative structure called the Bayesian Variational Coupled Modular Neural Network with Leech Growth Algorithm (BVCMNN-LGA) is proposed to predict employee turnover. Initially, the preprocessing phase is done using Improved Adaptive Iteratively Reweighted Penalized Least Squares (IAIRPLS), and then followed by the Basketball Team-Wolf Bird Optimization Algorithm (BT-WBOA). Finally, optimal feature selection is accomplished, producing a smaller number of highly discriminatory features. These characteristics are then used to feed the Bayesian Variational Coupled Modular Neural Network (BVCMNN), which is used for predicting complicated behavior exhibited by employees. Finally, the Leech Growth Algorithm (LGA) is used to minimize the loss incurred in BVCMNN, thereby ensuring stability in convergence. The predicted turnover probabilities are further translated into actionable IM-HR recommendations to improve employee retention and workforce engagement. Experimentation with Employee Turnover data indicates that the proposed BVCMNN-LGA has better performance with an accuracy of 99.97%, precision of 99.95%, recall of 99.96%, F1-score of 99.94%, specificity of 99.93%, and a low error rate of 0.01% and lower computation time of 0.02s, which confirms its ability to be used in intelligent HR decision support.