Developing highly accurate machine learning models for optimizing water quality management decisions in tilapia aquaculture

Ashwaq M. Alnemari, Wael Mohamed Elmessery, Amjad Saleh Qazaq, Moustapha E. Moustapha, S.Zh. Rakhimgalieva, Mohamed Fawzi Abdalshefie Abuhussein, Sadeq K. Alhag, Laila Ali M. Al-Shuraym, Farahat Saad Moghanm, Péter Szűcs

2 more authors

Mohamed Hamdy Eid, Abdallah Elshawadfy Elwakeel

Scientific Reports, 2025

Full text Open access: Gold Metadata (CSL-JSON)
DOI 10.1038/s41598-025-16939-w
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

The optimization of water quality management is crucial for the success and sustainability of tilapia aquaculture. This study presents a novel approach for developing a decision-support system by comparing various machine learning models to predict optimal water quality management actions based on key environmental parameters. The novelty of this work lies in its focus on automating management decisions, moving beyond simple parameter prediction. A synthetic dataset, representing 20 critical water quality scenarios, was generated and used for model development. This dataset was preprocessed using class balancing with SMOTETomek and feature scaling. Several machine learning algorithms, namely Random Forest, Gradient Boosting, XGBoost, Support Vector Machines, Logistic Regression, and Neural Networks, were trained and evaluated. Additionally, a Voting Classifier ensemble model was employed to leverage the strengths of these individual models. Performance was assessed using accuracy, precision, recall, and F1-score, with cross-validation conducted to ensure robustness. The results demonstrated that multiple models including the ensemble Voting Classifier, Random Forest, Gradient Boosting, XGBoost, and Neural Network models, achieved perfect accuracy on the held-out test set. Cross-validation confirmed high performance across all top models, with the Neural Network achieving the highest mean accuracy of 98.99% ± 1.64%. Rather than identifying a single optimal model, this study demonstrates that model selection should be guided by specific deployment requirements, with each approach offering distinct advantages for different operational priorities. The proposed machine learning approach offers a promising tool for optimizing water quality management in Tilapia aquaculture, providing a foundation for data-driven systems that can improve efficiency, productivity, and sustainability in the industry.