Performance of Different Classification Models on National Coral Reef Monitoring Dataset

This paper presents the data analysis and classification of the survey results from the National Coral Reef Monitoring Program: Assessment of coral reef communities in the U.S Virgin Islands. The dataset was created as a part of a survey led by NOAA National Centers for Coastal Ocean Science (NCCOS) in collaboration with nine other universities and research institutes. The data was collected as a result of Line Point-Intercept (LPI) survey and Belt Transect procedure. Different data exploration, model development, and interpretation were performed on the dataset to study information on fish species diversity, density, composition, size, and other derived metrics. This study intends to demonstrate the performance of different machine learning models on the classification of the target feature-Strat Type. Various machine learning techniques such as Logistic regression, K-Nearest Neighbor, Decision Tree, and Random Forest are used to classify the categorical Strat Type. It was concluded that Random Forest, and Decision Tree present the best performance.