That isn't how statistics should be used.
You make statistics for a region, and apply the results from the statistics from that region, to that region.
If 80% of the crimes in Detroit are commited by blue people, it is perfectly rational to screen blue people first in Detroit, based on those...