That's exactly why some reseasrch into the methodology needs to go into it. If I were to implement this, I may pick some aggregate, but depends on data availability. Again, which is why the ticket requires research.