For many crimes there might not be bodies to count. So I think that just from a math point of view(no race/social stuff interpretation) , you can be in a situa

For many crimes there might not be bodies to count. So I think that just from a math point of view(no race/social stuff interpretation) , you can be in a situation where the starting data combined with bad algorithm will keep you stuck in a local minimum. Probably math people can model this issue with abstract terms not to offend any color of politics you prefer but you have the issue of bad implementation and as always garbage in is garbage out, so you might need an algorithm that starts with zero input and uses random data and converges slowly , but you still need a way to make sure you keep the algorithm evolving with changes over large time (maybe reset it and make it start over after a few years)