Scholarlink

<p>To solve bankruptcy prediction&#xa0;tasks, we proposed an improved rime optimization&#xa0;technique (RMRIME). The proposed RMRIME&#xa0;algorithm first employs roulette wheel selection step, introducing random individuals into the position updating process to expand the search space and boost the RMRIME’s exploration power. Also, a mutation idea is utilized, which uses info from elite agents to generate mutated individuals. This method aids in increasing population diversity and improving the algorithm’s convergence accuracy. To evaluate RMRIME, it was compared with nine advanced algorithms on the IEEE CEC 2017 benchmark functions. The experimental results show we could significantly improve overall accuracy, with better convergence on the benchmark functions. Finally, a fuzzy k-nearest neighbor-based feature selection, based on the RMRIME, is proposed to tackle the bankruptcy prediction. Experiments performed on three bankruptcy datasets versus the advanced algorithms. The results reveal that the proposed model could&#xa0;achieve an accuracy rate of 91.02%, 97.08% on the dataset, respectively.</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

Advancing bankruptcy prediction: a study on an improved rime optimization algorithm and its application in feature selection

  • 期刊论文

  • 发表日期：

    2025年1月24日

查看全文
  • Yaoxian Ji,
  • Chenglang Lu,
  • Lei Liu,
  • Ali Asghar Heidari,
  • Chengwen Wu,
  • Huiling Chen

摘要

To solve bankruptcy prediction tasks, we proposed an improved rime optimization technique (RMRIME). The proposed RMRIME algorithm first employs roulette wheel selection step, introducing random individuals into the position updating process to expand the search space and boost the RMRIME’s exploration power. Also, a mutation idea is utilized, which uses info from elite agents to generate mutated individuals. This method aids in increasing population diversity and improving the algorithm’s convergence accuracy. To evaluate RMRIME, it was compared with nine advanced algorithms on the IEEE CEC 2017 benchmark functions. The experimental results show we could significantly improve overall accuracy, with better convergence on the benchmark functions. Finally, a fuzzy k-nearest neighbor-based feature selection, based on the RMRIME, is proposed to tackle the bankruptcy prediction. Experiments performed on three bankruptcy datasets versus the advanced algorithms. The results reveal that the proposed model could achieve an accuracy rate of 91.02%, 97.08% on the dataset, respectively.