Applying Limnological Feature-Based Machine Learning Techniques to Chemical State Classification in Marine Transitional Systems