In a scenario where you need to decide between using a standard cluster and a single-node cluster for a machine learning project, what factors would you consider to make an informed decision? Discuss aspects such as project scale, computational requirements, and potential cost implications.