Scholarlink

<p>This study introduces a new predictive framework for assessing fashion sustainability by combining advanced machine learning models with cutting-edge optimization techniques. Specifically, it utilizes Random Forest Classifier (RFC) and Decision Tree Classifier (DTC) alongside Bald Eagle Search Optimization (BESO) and Runge-Kutta Optimization (RKO) for to prevent convergence to local minima. This integration results in optimized models, including RFBE, RFRK, DTBE, and DTRK. The method also uses recursive feature elimination to identify key predictors, improving interpretability and accuracy. Experimental results show that the Runge-Kutta-optimized Random Forest (RFRK) achieves 97.8% training accuracy, 96.3% testing accuracy, and an overall accuracy of 96.5%, with a Matthews Correlation Coefficient (MCC) of 0.971, greatly exceeding baseline RFC and other methods. Similarly, the Runge-Kutta-optimized Decision Tree (DTRK) reaches 95.9% accuracy, outperforming the baseline DTC’s 93.3%. These results highlight the benefits of combining feature selection with advanced global optimization techniques, providing high-precision predictions for fashion sustainability and valuable insights for industry stakeholders and policymakers.</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

Leveraging Optimization Algorithms and Feature Selection to Predict Fashion Sustainability by Analyzing Random Forest and Decision Tree Models for Improved Classification Performance

  • 期刊论文

  • 发表日期：

    2026年1月7日

查看全文
  • Dan Li,
  • Qingwan Zeng,
  • Yongru Liu

摘要

This study introduces a new predictive framework for assessing fashion sustainability by combining advanced machine learning models with cutting-edge optimization techniques. Specifically, it utilizes Random Forest Classifier (RFC) and Decision Tree Classifier (DTC) alongside Bald Eagle Search Optimization (BESO) and Runge-Kutta Optimization (RKO) for to prevent convergence to local minima. This integration results in optimized models, including RFBE, RFRK, DTBE, and DTRK. The method also uses recursive feature elimination to identify key predictors, improving interpretability and accuracy. Experimental results show that the Runge-Kutta-optimized Random Forest (RFRK) achieves 97.8% training accuracy, 96.3% testing accuracy, and an overall accuracy of 96.5%, with a Matthews Correlation Coefficient (MCC) of 0.971, greatly exceeding baseline RFC and other methods. Similarly, the Runge-Kutta-optimized Decision Tree (DTRK) reaches 95.9% accuracy, outperforming the baseline DTC’s 93.3%. These results highlight the benefits of combining feature selection with advanced global optimization techniques, providing high-precision predictions for fashion sustainability and valuable insights for industry stakeholders and policymakers.