ABSTRACT
Predicting fish infections by analysis of water quality characteristics allows for the prompt intervention by identifying possible illnesses before they show clinical signs. Acquiring different information for training correct machine learning models is made more complex by the variety Fish Disease of aquatic habitats. In order to make trustworthy conclusions, it is essential to resolve issues with the interpretability and transparency of prediction algorithms. The variety and interpretability of datasets are two of the challenges highlighted in the problem statement as preventing early discovery. For better illness prediction and predictive modelling, it is crucial to combine machine learning with explainable methods. To defeat dataset troubles and assurance interpretability, the recommended strategy coordinates gathering learning with logical simulated intelligence procedures, like SHAP (Shapley Additive explanations) values. This approach beats current methodologies concerning exactness, accuracy, and interpretability. Eminent execution benefits are shown by the significantly better exactness, accuracy, and time intricacy when contrasted with the ongoing framework. More perplexing techniques are expected for the administration of fish diseases, since existing models experience hardships with dataset assortment and interpretability.
