AN INTEGRATED MODEL OF MACHINE LEARNING AND FUZZY LOGIC FOR QUALITY ASSESSMENT OF DRINKING WATER

Abstract

Ensuring access to safe drinking water is essential for environmental sustainability and public health. However, existing water quality assessment methods often encounter challenges in accurately predicting water quality parameters due to the inherent uncertainties and complexities associated with water quality data. This research addresses this gap by proposing a novel hybrid machine learning approach to enhance water quality prediction. While traditional machine learning models exhibit strong predictive capabilities, they often struggle to effectively manage the imprecise and ambiguous nature of water quality data. To address this limitation, this study investigates the integration of Random Forest with Fuzzy Logic to improve predictive performance. Specifically, Random Forest enhances the model’s classification accuracy, while Fuzzy Logic enables the nuanced interpretation of qualitative parameters. The proposed hybrid model was trained and evaluated on a comprehensive water quality dataset. The experimental results indicate that the integrated Random Forest-Fuzzy Logic model achieves a high level of predictive  performance, with an accuracy of 99.92%, precision of 99.47%, recall of 100%, and an F1-score of 99.73%. These findings highlight the effectiveness of the proposed approach in improving water quality monitoring and management. The integration of machine learning and fuzzy logic offers a robust framework for addressing uncertainties in water quality assessment, with significant implications for water resource management, public health protection, and evidence-based policy development.