Data projects are generally organized in one of two ways, using either a top-down or bottom-up method. But is there a right way?
