Approximate Belief Rule-Based Modeling for Coupling and Coordination Evaluation of Employee Competency
提出一种基于近似置信规则模型的员工胜任力评价方法，融合专家经验与有限历史数据，通过耦合协调指数反映胜任力发展水平，经多种对比验证，模型精度和鲁棒性优于传统方法。
Employee competency evaluation is a fundamental prerequisite for achieving job fit. Due to the complex hierarchical structure, multiple dimensions, and the uncertainty of indicators, it is difficult for existing methods to accurately evaluate the comprehensive and developmental levels of employee competency. Thus, a coupling and coordination evaluation method for competency based on the approximate belief rule-based (ABRB) model is proposed. This method can effectively integrate expert experience and limited historical data to describe and infer multivariate uncertain indicator information, and reflect the development level of employee competency by introducing a coupling coordination index. To ensure research rigour, robustness checks are conducted through multiple approaches: 1) comparative analysis with three mainstream methods (Back Propagation Neural Network, Extreme Learning Machine, Fuzzy Reasoning System); 2) validation under different training set proportions (10%, 30%, 50%, 60%); 3) iterative optimization of model parameters via the projection-based covariance matrix adaptive evolution strategy (P-CMA-ES) with interpretable constraints. Taking the evaluation of the competency of employees in a certain group as an example, the results show that the proposed method has better modeling accuracy and optimization robustness than traditional methods, verifying the reliability of the research.