Water is an important natural resource and clean water is vital for maintaining health and hygiene of all living organisms. Estimating and classifying water quality facies is a critical way to analyse water quality and proper water management. The present study underlines the applicability of Machine Learning (ML) models to assess water quality by classifying hydrogeochemical facies within the Tawi basin of the Jammu region. This study employs a range of ML algorithms, including Decision Tree (DT), XGBoost, Random Forest (RF), K-Nearest Neighbors (KNN), and Artificial Neural Network (ANN), to evaluate their effectiveness in accurately classifying hydrogeochemical facies derived from Piper's diagram. The dataset, consisting of chemical parameters extracted from water samples collected from the Tawi basin, was initially imbalanced, with a large majority of samples belonging to a single facies. To address this, we applied the Synthetic Minority Over-sampling Technique (SMOTE), ensuring balanced class distributions for more reliable model training and evaluation. The classification results demonstrate high accuracy across the models, with DT achieving 93%, RF 99%, XGBoost 96%, KNN 81%, and ANN 96%. In addition to overall accuracy, we employed other evaluation metrics such as precision, recall, F1-score, and the precision-recall curve to provide a more comprehensive assessment of model performance. The results underscore the potential of ML in automating water quality assessment based on hydrogeochemical parameters. The findings of the study provide a robust framework for using ML models in determining water quality, particularly in regions where data is scarce and conventional analysis is limited.