<![CDATA[Purpose – This study develops an Internet of Things (IoT)-based Smart Water Quality framework for monitoring water-quality parameters and evaluates a Gaussian Naive Bayes classifier for clean-water feasibility assessment. Methods – A prototype-development study used an ESP32 integrated with pH, total dissolved solids (TDS), turbidity, and temperature sensors at a single school borehole well during a one-month monitoring period. One prototype session produced 500 telemetry records over 61 minutes for acquisition-stability and threshold analysis. A separate labeled dataset retained 485 records, divided into 388 training and 97 held-out test observations; five-fold cross-validation was performed on the training subset. Findings – Telemetry was acquired at a mean interval of 7.38 seconds. Recorded pH, TDS, and turbidity exceeded the permissible limits and were categorized as Not Suitable under the adopted Indonesian health criteria. Gaussian Naive Bayes achieved mean cross-validation accuracy of 95.46%, precision of 95.51%, recall of 95.46%, and F1-score of 95.47%. On the 97-record held-out test set, the model achieved 94.85% accuracy, 94.00% precision, 95.92% recall, and 94.95% F1-score, with 47 true positives, 45 true negatives, three false positives, and two false negatives. Research implications – The proposed system provides a functional architecture for IoT-based water-quality monitoring and highlights the need for representative labeled datasets and independently verified ground truth. Originality – This study integrates an IoT telemetry prototype with separately evaluated Gaussian Naive Bayes classification for clean-water feasibility assessment without claiming a fully validated end-to-end classifier.]]>
Copyrights © 2026
