If a group of people is committing disproportionate amounts of crime, is it legitimate to target that group disproportionally?
Let's grant the assumptions that investigating people who have not committed a crime will not increase their propensity to commit crime, and that capturing someone who is breaking the law will lower the crime rate in the long run.
If it's not legitimate to disproportionately target groups that commit a disproportionate amount of crime (or, as I would say, to target people proportionately to their propensity to commit crime as a group), then we should target every demographic equally, regardless of whether or not it's even possible that they can commit a crime. That is, we should target rocks as often as we target people who we (somehow) know for a certain fact will commit a crime in an hour (provided that this is not a crime itself).
Obviously, more factors should be taken into account than just race, and I think it would improve the entire situation to put more research into finding a wider variety of crime-prone groups.