Scholarlink

<p>Precision water quality forecasts are crucial to public health, thus prediction methods must improve. This research offers a new hyper-parameter tuning method that combines XGBoost and Krill Herd Algorithms (KHA) to improve drinking water quality prediction.&#xa0;The study baseline model for forecasting drinking water quality is XGBoost, a popular machine learning algorithm. Due to the necessity of optimizing model hyper-parameters, we give the Krill Herd Algorithm, a natural optimization technique, to explore XGBoost's parameter space.&#xa0;Integrating KHA with XGBoost requires an objective function to steer optimization. Due to its efficiency in exploring high-dimensional spaces and adaptability, the method is used to improve XGBoost and encourage the identification of globally optimal hyper-parameters.&#xa0;To evaluate the model, a huge dataset of water quality conditions is used. To avoid overfitting and ensure generalizability, cross-validation is performed to evaluate model resilience. The baseline XGBoost is compared to the optimized model following KHA-based hyper-parameter tweaking to determine its improvement.&#xa0;Results reveal that the integrated technique predicts drinking water quality more accurately. XGBoost model achieved 86.12% accuracy, Linear Regression gained 74% accuracy and Random Forest achieved 78.85% accuracy. The proposed model gained the accuracy level 96.4%. A sensitivity analysis helps determine how hyper-parameters affect model performance and their importance.&#xa0;Finally, this study increases water quality prediction by introducing a hybrid model that combines the advantages of XGBoost and the Krill Herd Algorithm. The upgraded model shows how nature-inspired strategies can improve machine learning models' performance for crucial tasks like water potability evaluation. Results from this study could help public health treatments and environmental 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

Optimized XGBoost Hyper-Parameter Tuned Model with Krill Herd Algorithm (KHA) for Accurate Drinking Water Quality Prediction

  • 期刊论文

  • 发表日期：

    2025年3月10日

查看全文
  • Nikhil Malik,
  • Arpna Kalonia,
  • Surjeet Dalal,
  • Dac-Nhuong Le

摘要

Precision water quality forecasts are crucial to public health, thus prediction methods must improve. This research offers a new hyper-parameter tuning method that combines XGBoost and Krill Herd Algorithms (KHA) to improve drinking water quality prediction. The study baseline model for forecasting drinking water quality is XGBoost, a popular machine learning algorithm. Due to the necessity of optimizing model hyper-parameters, we give the Krill Herd Algorithm, a natural optimization technique, to explore XGBoost's parameter space. Integrating KHA with XGBoost requires an objective function to steer optimization. Due to its efficiency in exploring high-dimensional spaces and adaptability, the method is used to improve XGBoost and encourage the identification of globally optimal hyper-parameters. To evaluate the model, a huge dataset of water quality conditions is used. To avoid overfitting and ensure generalizability, cross-validation is performed to evaluate model resilience. The baseline XGBoost is compared to the optimized model following KHA-based hyper-parameter tweaking to determine its improvement. Results reveal that the integrated technique predicts drinking water quality more accurately. XGBoost model achieved 86.12% accuracy, Linear Regression gained 74% accuracy and Random Forest achieved 78.85% accuracy. The proposed model gained the accuracy level 96.4%. A sensitivity analysis helps determine how hyper-parameters affect model performance and their importance. Finally, this study increases water quality prediction by introducing a hybrid model that combines the advantages of XGBoost and the Krill Herd Algorithm. The upgraded model shows how nature-inspired strategies can improve machine learning models' performance for crucial tasks like water potability evaluation. Results from this study could help public health treatments and environmental monitoring.