Benchmark of Clean Water Classification Based on Water Quality Index Using Deep Learning

Poor water quality poses a significant health risk, contributing to the global spread of waterborne diseases. Challenges in water quality monitoring include data inconsistency, limited access to testing tools, and technological barriers. This study investigates the use of deep learning and machine learning models to classify water quality based on tabular Water Quality Index (WQI) parameters. A diverse dataset was aggregated from national monitoring sources and processed through cleaning, standardization, imputation, and categorical label assignment. Four models were evaluated: Random Forest as a baseline, and three deep learning models—Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), and TabNet. While MLP achieved the highest classification accuracy and robustness under noisy and incomplete inputs, TabNet showed promising results as a deep architecture designed specifically for tabular data, though it did not outperform MLP in this study. These findings highlight the potential of data-driven modeling as a scalable alternative to conventional laboratory-based water quality assessment.