Towards Intelligent Environmental Monitoring: Machine Learning Approaches for Automated Water Potability Prediction

Water potability is a significant factor in environmental protection and public health,and millions of people around the world live with access to safe drinking water. The quality of water needs to be maintained so that no water diseases can occur and overall health is maintained, we use machine learning models for the prediction of potability of water based on several physicochemical factors. We compare some models like K-Nearest Neighbors (KNN), Decision Tree, Logistic Regression, Gaussian Naive Bayes, and Random Forest and test them in the classification of water as is it potable to drink or not. The data contains $\mathbf{3, 2 7 6}$ samples with features like pH, hardness, solids, chloramines, sulfate, conductivity, organic carbon, trihalomethanes, turbidity, and potability labels. Our methodology encompasses exploratory data analysis (EDA), feature engineering, and model performance comparison using performance metrics like accuracy, precision, recall, and F1score. The results indicate that the Random Forest works best with 66 % accuracy than other models, reflecting its consistency for predicting water potability. This paper emphasizes the ability of machine learning to automate water quality identification and deliver a scalable approach to monitoring and having safe clean drinking water.