Spark - Group by multiple columns VS Group by an id created from the same multiple columns

Question

Is it a good idea to create an id from multiple columns and then make a group by ? Maybe grouping data by one column (the id) is more efficient?

Example:

m