Lake Michigan beaches E.Coli prediction tool using Artificial Neural Network

Indiana Department of Environmental Management (IDEM) does daily water samples in southern Lake Michigan Beaches daily during regular beach goers season. This data is used to close the beaches for E.Coli spikes. One day lag data are used for beach closures on a given day. This research is an effort to create a model with a recent tool to predict E.Coli classes using other physical, chemical and site conditions so that it becomes a tool to avoid one day time lag. During 2019, during regular beach goes season, data samples were collected and documented to develop a artificial neural network model. Different combinations were tried and the best model was identified after cross validation. BRNN (Baysian Regularization Neural Network) was selected as the best model. This model used 13 input data to predict E.Coli classes with a successful prediction of 80 to 90% in training, testing and validation datasets.