Any dataset involving police actions will show high concentrations in poor areas because that's where police patrol the most and where they're most likely to crack down on behaviors that might be allowed to slide elsewhere (in part due to the racial demographics of those areas).

Usually allocation decisions are related to actual car/pedestrian fatality/injury counts + trial placements and measurements. Either way, wouldn't you be in favor of measures that remove police from overpoliced poor neighborhoods in favor of a technology focusing on traffic safety enforcement?