Refining Machine Learning for Employee Turnover Classification: A CAP-BOA-Based Probability Neural Network Approach

This paper introduces a novel approach, the CAP-BOA-based Probabilistic Neural Network (PNN), aimed at optimizing the smoothing factor of PNN to address the suboptimal performance issues encountered by traditional methods in handling complex classification problems. The CAP-BOA algorithm leverages the global and local search capabilities of the Butterfly Optimization Algorithm (BOA) and integrates Cauchy mutation, adaptive weight optimization, and dynamic switching probability strategy to enhance the rationality of computation and switching strategies during the optimization process. Compared to conventional PNN optimization methods, the CAP-BOA-based PNN demonstrates a significant improvement in convergence accuracy. Experimental results validate the efficacy of this approach in overcoming traditional limitations, particularly in the scenario of classifying employee turnover within enterprises. The findings showcase the effectiveness of the CAP-BOA-based PNN in achieving enhanced overall convergence accuracy.