This would essentially be building a hashmap on the grouped columns where the values are the rows associated to the key