Scholarlink

<p>Water quality assessment is essential for environmental protection, yet traditional methods are often time-consuming. This study introduces a hybrid framework that integrates classical ensemble learning, combining bagging and boosting strategies, with Quantum Machine Learning (QML) to predict the Water Quality Index (WQI) and classify Water Quality Classes (WQC). Seven river parameters were analyzed, and Principal Component Analysis (PCA) was applied to retain the three most informative features for consistent use across classical and quantum models. Ten regression algorithms were evaluated, and the hybrid Extra Trees (ET) + Adaptive Boosting (AdaBoost) model achieved the highest performance with an R<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(^{\varvec{2}}\)</EquationSource> </InlineEquation> of 0.9975. For classification, Quantum Support Vector Machine (QSVM), the Variational Quantum Classifier (VQC), and the proposed Classical–Quantum K-Nearest Neighbors (CQKNN) were compared with classical SVM, Naïve Bayes (NB), and KNN. CQKNN achieved the best overall results, including 95.35% accuracy, 92.50% precision, and 95.18% recall. Its performance reflects the benefits of quantum-fidelity based similarity computation together with reduced quantum resource usage made possible by qubit reuse and mid-circuit measurement. These findings show that integrating hybrid bagging and boosting regression with a compact quantum classifier improves predictive reliability and supports the development of efficient data-driven systems for water quality monitoring.</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

Hybrid Ensemble and Quantum Machine Learning Framework for River Water Quality Prediction and Classification

  • 期刊论文

  • 发表日期：

    2026年1月10日

查看全文
  • Amine Zeguendry,
  • Zahi Jarir,
  • Mohamed Quafafou

摘要

Water quality assessment is essential for environmental protection, yet traditional methods are often time-consuming. This study introduces a hybrid framework that integrates classical ensemble learning, combining bagging and boosting strategies, with Quantum Machine Learning (QML) to predict the Water Quality Index (WQI) and classify Water Quality Classes (WQC). Seven river parameters were analyzed, and Principal Component Analysis (PCA) was applied to retain the three most informative features for consistent use across classical and quantum models. Ten regression algorithms were evaluated, and the hybrid Extra Trees (ET) + Adaptive Boosting (AdaBoost) model achieved the highest performance with an R \(^{\varvec{2}}\) of 0.9975. For classification, Quantum Support Vector Machine (QSVM), the Variational Quantum Classifier (VQC), and the proposed Classical–Quantum K-Nearest Neighbors (CQKNN) were compared with classical SVM, Naïve Bayes (NB), and KNN. CQKNN achieved the best overall results, including 95.35% accuracy, 92.50% precision, and 95.18% recall. Its performance reflects the benefits of quantum-fidelity based similarity computation together with reduced quantum resource usage made possible by qubit reuse and mid-circuit measurement. These findings show that integrating hybrid bagging and boosting regression with a compact quantum classifier improves predictive reliability and supports the development of efficient data-driven systems for water quality monitoring.