I wonder what would happen if they took the Abraham Wald approach and designed a counterintuitive algorithm. Like, make a heatmap of violent crimes (assault, robbery, rape, etc.), and then sic the algo on non-violent crimes in the inverted heatmapped areas, like larceny, wire fraud, and so on. Higher-income areas have wealthier people, and statistically wealthier people are better equipped to commit high-dollar white collar crimes. You could also use the hottest areas on the violence heatmap to target social services support.