Machine Learning Approach for Prediction of Ammonia in Freshwater Bodies

Water is a fundamental need for the survival of both human beings and aquatic species such as fish. This work aims to forecast the concentration of Ammonia in freshwater ecosystems by considering three distinct variables: pH, Dissolved Oxygen, and Chemical Oxygen Demand. The work emphasizes the detrimental impact of Ammonia on freshwater fish and the need to accurately estimate ammonia levels using the Machine Learning algorithm for implementing proactive measures. The data is pre-processed to mitigate the risk of overfitting in machine learning models. A total of six supervised machine learning algorithms are used to make predictions, and the model that exhibits the highest level of accuracy is chosen. Subsequently, performance metrics such as Mean Square Error (MSE), Root Mean Square Error (RMSE), and R2-Score are used to assess the accuracy of the machine learning models. Based on the work conducted, the Random Forest algorithm gives the best results and is the most suitable Machine-Learning method for predicting Ammonia levels in freshwater bodies.