Deer hunting optimization algorithm with multi-agent reinforcement learning for water quality prediction

Abstract Diseases linked to water contamination have been a key cause of the global increase in mortality rates in recent years. Therefore, it is essential to predict this disease accurately and early in order to lower the death rate. Reinforcement learning (RL) has garnered increased interest recently due to its ability to provide accurate prediction results by allowing for a deeper knowledge of environmental variables. Therefore, proposed work involves a better feature selection and upgraded classification method which aims to forecast water quality with high accuracy. In this research work, a multi-agent reinforcement learning (DHOMRLWQP) model for water quality prediction based on deer hunting optimization has been proposed. Evaluating the pollution level of Cauvery River water is the primary goal of the proposed DHOMRLWQP models. The deep Q-Network (DQN) technology is used in the DHOMRL-WQP method to predict water quality. The major goal is to use deer hunting optimization (DHO) to adjust the DQN hyper-parameters. The DHOMRL-WQP method has been implemented on the Cauvery River Dataset which was assessed using a number of metrics, achieving a F1-Score of 0.93, accuracy of 0.95, precision of 0.97, and recall of 0.90. With a high accuracy score of 98%, the results show that the DHOMRL-WQP models perform better than other traditional machine learning techniques.