A Machine Learning Framework for Fish Selection in Sustainable Aquaculture
Abstract
For fish farmers in the real world, selecting the best fish species for aquaculture in a given aquatic environment is a very difficult issue. The main goal of this research is to develop a machine learning model that can predict the best fish species to raise in an aquatic environment and suggest many fish species with anticipated ones. In this research, we have employed a number of supervised and unsupervised machine learning models. To evaluate the model, we used a dataset of aquatic habitats for eleven different fish species. We predicted the fish species using a variety of aquatic environmental factors, including temperature, turbidity, and pH. As performance measurements, we employed accuracy, F1 score, precision, and recall. According to experimental findings, it is suggested that the Random Forest (RF) model’s prediction result exhibits an accuracy of 97.64% with good precision, recall, and F1 score for the dataset under test. Furthermore, we compare the results of the suggested model with the most advanced models, including SVC, random forest, gradient boosting, MLP, KNN, and others. The proposed model performs better than other applied models, as evidenced by its accuracy score and statistics. Finally, we applied an unsupervised ML model, K-means clustering to make recommendations of multiple species with the predicted fish species. The experimental result shows good recommendation outcomes.