EventsThe 10th International Electronic Conference on Water Sciences
Published
This submission belongs to the session S7. Remote Sensing, Artificial Intelligence, and New Technologies in Water Sciences of the event The 10th International Electronic Conference on Water Sciences
Published date
19 Nov, 2026
Academic Editor
author-avatarNikiforos Samarinas
Contribution&Funding
Conceptualization,
[Sarah Wathiq Taha] ; [Mustafa Muthanna Najm Shahrabani]
Methodology,
[Sarah Wathiq Taha] ; [Mustafa Muthanna Najm Shahrabani]
Software,
[Sarah Wathiq Taha] ; [Mustafa Muthanna Najm Shahrabani]
Validation,
[Mustafa Muthanna Najm Shahrabani]
Formal analysis,
[Sarah Wathiq Taha]
Investigation,
[Mustafa Muthanna Najm Shahrabani]
Resources,
[Mustafa Muthanna Najm Shahrabani]
Data curation,
[Sarah Wathiq Taha] ; [Mustafa Muthanna Najm Shahrabani]
Writing - original draft,
[Sarah Wathiq Taha]
Writing - review,
[editing] ; [Mustafa Muthanna Najm Shahrabani] ; [Sarah Wathiq Taha]
Visualization,
[Sarah Wathiq Taha]
Project administration,
[Sarah Wathiq Taha]
Funding: --
Citation
Sarah Wathiq Taha, Mustafa Muthanna Najm Shahrabani, Using Artificial Neural Network Approach for Water Quality Classification, in Proceedings of The 10th International Electronic Conference on Water Sciences, 24 November–27 November 2026, MDPI: Basel, Switzerland
Share
Email
Facebook
Twitter
LinkedIn

Using Artificial Neural Network Approach for Water Quality Classification

1. Civil Engineering, Al-Maarif University, Al-Anbar Governorate, Iraq
2. Informatics, Engineering and Technologie, Kauno Kolegija Higher Institute, Kaunas, 50468, Lithuania
Abstract

Water quality is an essential aspect of public health and environmental sustainability. This study employed Artificial Neural Networks to classify water quality in the Danube River, Budapest. Four key parameters were utilised as an input layer: pH, dissolved oxygen, nitrate (NO₃-N), and total suspended solids (TSS). The data were pre-processed for consistency by handling missing values and normalising the parameters, while World Health Organisation guidelines for drinking water quality were used as benchmarks for determining water safety. A Multilayer Perceptron ANN was trained and tested to classify water quality into two groups: safe or unsafe. The model achieved an overall accuracy of 94%, indicating its ability to identify complex patterns in water quality data. The safe category achieved a recall of 100% and an F1-score of 0.97, demonstrating that the model effectively identified safe water samples with minor misclassifications. In contrast, the unsafe category showed lower performance, with a recall of 40% and an F1-score of 0.57. The confusion matrix revealed that a number of unsafe samples were misclassified as safe, indicating model bias toward the safe category due to the imbalance in the dataset. The training accuracy increased rapidly during the early epochs and eventually stabilised at more than 92%, suggesting a strong learning capability. Despite the limitation associated with imbalanced data, the findings confirm the potential of ANN-based classification for data-driven water quality assessment. The outcomes of this study could support environmental monitoring, public health, and water management by providing a rapid approach that may reduce reliance on time-consuming laboratory analysis and improve real-time decision-making. This approach could also be integrated into early warning systems to detect pollution events and mitigate risks before contamination spreads. Future studies should consider a wider range of physicochemical parameters and address dataset imbalance to improve the model’s sensitivity to unsafe water conditions.

Keywords
Water Quality
Artificial Neural Network
Monitoring
Classification
Danube River
A Portable NB-IoT/LTE-M Network Diagnostic Tool for Hydrometeorological Data Transmission Reliability in Santa Catarina, Brazil
Long-term surface water temperature changes in large rivers: a satellite-based assessment of the Yangtze River, China