For the sake of public health safety, it is crucial to assess the water quality of a waterbody. In addition to being consumption of contaminated water, recreational use (fishing, swimming) also contributes to the spread of various diseases e.g. skin diseases. In this study, water quality index (WQI) was developed by applying the entropy weight method for the Barak River. Fourteen water quality parameters from two river locations were considered for the development of WQI. The WQI was used to evaluate the water fitness for bathing purpose. The methods of classification based on machine learning (ML), Support Vector (SV) and K Nearest Neighbour models were scrutinized to predict and classify WQI. For evaluating the performance of the models, precision, recall, F1-score and confusion matrix were applied. The findings of this study identified that for both the locations, bathing water quality varied from ‘good-fair’ range. The research outcome also exhibit that Support Vector performed well than K Nearest Neighbour model. This study demonstrates the implementation/attribute of ML models in classification of river water quality; hence it will be beneficial to water quality monitoring agencies and stakeholders.