Water is the key that keeps living things alive and serves as a key factor in supporting overall health and well-being, balancing ecosystems, and supporting agriculture. However, when water becomes contaminated, it can lead to serious health consequences. Certain drinking water pollutants have been connected to cognitive impairments, reproductive issues, and gastrointestinal disorders. Children, elderly, pregnant women, and persons with low immunity are possibly vulnerable to these dangers. In response to the previous concerns, this paper presents an investigation of a method that leverages machine learning (ML) algorithms to enhance the classification of water quality based on observation of key physicochemical properties. It presents the implementation and comparison of the performance of seven benchmark supervised ML algorithms: Decision Tree (DT), Logistic Regression (LR), K-Nearest Neighbors (KNN), eXtreme Gradient Boosting (XGBoost), Random Forest (RF), ensemble Adaptive Boosting (AdaBoost) and Support Vector Machine (SVM) with and without applying SMOTE technique for data oversampling. The best performance model will be recommended for a real-world water motoring system, providing a low-cost, intelligent solution for early identification of non-potable water. The CRISP-DM is the methodology used to ensure systematic reliability in the proposed method. Experimental results demonstrate that applying the SMOTE technique to the dataset improves the ML classification performance. Subsequently, RF exhibits the best performance with an accuracy of 71.73%. This study highlights the deployment of ML to enhance environmental monitoring, support a sustainable environment, and protect public health.