Quality index metrics with Bi-GRU-based water quality prediction

Abstract

ABSTRACT The precise assessment of groundwater resources is significant in the face of growing water demands, environmental pollution, and degradation. However, traditional methods of water quality prediction are inadequate in dealing with large volumes of data or missing data. Machine learning-based prediction methods are being explored to address this issue. Currently, single-indicator techniques are commonly used, but they may not accurately forecast multimodal water quality or capture inter-indicator connections. In this study, a Bi-GRU is proposed to predicting water quality based on quality index data. To evaluate the performance of our proposed method, we utilize Kaggle Datasets, which provide a diverse range of water quality data. We compare our results with existing methods such as MLP, LSTM, and GRU, using metrics including accuracy, precision, recall, and f-score. Overall, our study demonstrates that the Bi-GRU model is highly effective in predicting water quality based on quality index data. The results of our experiments indicate that our proposed method surpasses traditional approaches like MLP, LSTM, and GRU in terms of accuracy, precision, recall, and f-score. These findings have significant implications for improving water quality monitoring and pollution prevention efforts, enabling better management of groundwater resources in the face of growing water demands and environmental challenges.