Coming from a statistics background, this makes a lot of sense to me. Before any model, we always spend most of the time just cleaning and validating data. the actual modeling is usually the smaller part.I think that's why the medallion architecture ...
