This will make it easier to make the other dataset that I proposed (the "partitioned dataset of partitioned datasets")