Numerous pollutants have exerted a major impact on water quality in recent years, and the health of all living organisms and the environment is directly affected. The most effective water management indicator is Water Quality Index (WQI) developed by BIS (2004). The prediction and modelling of water quality are essential in finding the pollution source and treating it effectively. This study aims to build an efficient river water quality indicator prediction model and classify indicator values according to the Indian Drinking Water Standards (BIS 2004). Data were collected from 11 sampling stations at different points on the Bhavani River in Kerala and Tamil Nadu. The Water Quality Index is computed by using the 28 different parameters that affect the quality of water. The feature selection and data normalisation are applied to develop an efficient river water quality dataset. The WQI prediction model is built using deep learning architectures such as GRU, LSTM, and RNN. The performance of the deep learning based WQI prediction models are compared with traditional machine learning based models. The performance analysis indicates that the GRU-based prediction model shows promising results in predicting water quality