<![CDATA[Access to safe water remains a significant issue, with around 5.8 billion people lacking access to potable water globally. Rapid and accurate identification of water safety is thereby essential to reduce public waterborne diseases. However, conventional laboratory-based testing is typically time-consuming and expensive. On the other hand, machine learning provides time- and cost-effective assessments based on physicochemical properties. Unfortunately, most studies only evaluate a single model type in a small dataset, resulting in limited insight that makes it hard to determine the actual effectiveness of these models. To address this limitation, the present study conducts a comparative analysis of three machine learning paradigms: ensemble-based, neural network-based, and instance-based models. Using a publicly available dataset of 7,999 samples, each model is evaluated using key performance metrics, including accuracy, precision, and confusion matrix analysis. The evaluation results show that the ensemble-based model achieves the highest accuracy of 96.62% and precision of 96.53%, outperforming the neural network-based model, which achieves an accuracy of 94.75% and precision of 70.47%. Additionally, the instance-based model achieves an accuracy of 91.12% and a precision of 83.04%. These results indicate the effectiveness of the ensemble-based model for real-time water quality monitoring.]]>
Copyrights © 2025
