Scholarlink

<p>Water resources are those relating to the very life of humankind, economic development, health concerns, and sustainable environments. Accurate water-quality predictions need optimal parameters, reducing costs and improving machine learning (ML) efficiency for better management and pollution control. The current study analyzes the effectiveness of integrating two newly developed optimization algorithms, namely AVO and Non-Monopolize Search (NMS), along with traditional ML models, with a view to improving the results of water quality predictions. The base models used are CAT-Boost Classification (CATC) and LGBC. The following hybridization took place with the optimizers to give four frameworks, namely the CATC model combined with AVO-CAAV, the CATC model coupled with NMS-CANM, the LGBC model combined with NMS-LGNM, and the LGBC model coupled with AVO-LGAV. Accuracy Metric for the Testing Section - CANM performed best among them in the accuracy metric, having an accuracy of 0.985, while CAAV placed third with a value of 0.967. At the Testing phase, the Precision value is shown by the LGNM model as the second best, having a value of 0.971, while the worst is with the CATC model, with a value of 0.934.</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

Improving water quality prediction: a comparative analysis of CAT boost and LightGBM with AVO and NMS optimization

  • 期刊论文

  • 发表日期：

    2026年3月23日

查看全文
  • Li Chen

摘要

Water resources are those relating to the very life of humankind, economic development, health concerns, and sustainable environments. Accurate water-quality predictions need optimal parameters, reducing costs and improving machine learning (ML) efficiency for better management and pollution control. The current study analyzes the effectiveness of integrating two newly developed optimization algorithms, namely AVO and Non-Monopolize Search (NMS), along with traditional ML models, with a view to improving the results of water quality predictions. The base models used are CAT-Boost Classification (CATC) and LGBC. The following hybridization took place with the optimizers to give four frameworks, namely the CATC model combined with AVO-CAAV, the CATC model coupled with NMS-CANM, the LGBC model combined with NMS-LGNM, and the LGBC model coupled with AVO-LGAV. Accuracy Metric for the Testing Section - CANM performed best among them in the accuracy metric, having an accuracy of 0.985, while CAAV placed third with a value of 0.967. At the Testing phase, the Precision value is shown by the LGNM model as the second best, having a value of 0.971, while the worst is with the CATC model, with a value of 0.934.