Scholarlink

<p>Cyanobacteria is an indicator for freshwater ecosystem health. This study aims to (a) simulate the spatial distribution of cyanobacteria dynamics; and (b) optimize machine learning (ML) models with explainable artificial intelligence for cyanobacteria prediction in tropical freshwater lake. The water quality parameters (e.g. cyanobacteria concentration (μg/L), chlorophyll concentration (μg/L), turbidity, and cell count) were analysed from eleven sampling stations in Tasik Kenyir. The cyanobacteria dynamics was modelled via GeoPhyton analysis. Linear Regression (LR), Artificial Neural Network (ANN), and Extreme Gradient Boosting (XGBoost), were optimized to assess predictive performance. Strong temporal variability observed major peak in August and October 2024. High predictive capabilities were obtained by LR (training R<sup>2</sup> = 0.9435; testing R<sup>2</sup> = 0.9902), ANN (training R<sup>2</sup> = 0.9660; testing R<sup>2</sup> = 0.9739), and XGBoost (training R<sup>2</sup> = 1.000; testing R<sup>2</sup> = 0.7468). Feature importance analysis using SHapley Additive exPlanations (SHAP) identified cell count and chlorophyll as the dominant predictors.</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

Spatiotemporal Distribution of Cyanobacteria in Tropical Freshwater Reservoir and Machine Learning Prediction using Explainable Artificial Intelligence

  • 期刊论文

  • 发表日期：

    2026年5月19日

查看全文
  • Muhammad Sarfraz Ahmad Mangat,
  • Teh Sabariah binti Abd Manan,
  • Sharifah Sakinah Syed Abd Mutalib,
  • Zarimah Mohd Hanafiah,
  • Nur Azwa Muhamad Bashar,
  • Salmia Beddu,
  • Nur Liyana Mohd Kamal,
  • Daud Mohamad,
  • Shatha Aser Hamad Aldala’in,
  • Adel Gohari,
  • Wan Hanna Melini Wan Mohtar,
  • Muhammad Raza Ul Mustafa,
  • Mohamed Hasnain Isa,
  • Hamidi Abdul Aziz,
  • Mohd Suffian Yusoff

摘要

Cyanobacteria is an indicator for freshwater ecosystem health. This study aims to (a) simulate the spatial distribution of cyanobacteria dynamics; and (b) optimize machine learning (ML) models with explainable artificial intelligence for cyanobacteria prediction in tropical freshwater lake. The water quality parameters (e.g. cyanobacteria concentration (μg/L), chlorophyll concentration (μg/L), turbidity, and cell count) were analysed from eleven sampling stations in Tasik Kenyir. The cyanobacteria dynamics was modelled via GeoPhyton analysis. Linear Regression (LR), Artificial Neural Network (ANN), and Extreme Gradient Boosting (XGBoost), were optimized to assess predictive performance. Strong temporal variability observed major peak in August and October 2024. High predictive capabilities were obtained by LR (training R2 = 0.9435; testing R2 = 0.9902), ANN (training R2 = 0.9660; testing R2 = 0.9739), and XGBoost (training R2 = 1.000; testing R2 = 0.7468). Feature importance analysis using SHapley Additive exPlanations (SHAP) identified cell count and chlorophyll as the dominant predictors.