We've been building a routing engine for NYC that treats "Safety" as a cost function similar to "Traffic" or "Distance."

To do this, we had to ingest and normalize about 2M+ crime data points and correlate them with the NYC LION street grid.

One of the most surprising findings from the data was this "Last Mile" risks for logistics: 33% of auto thefts happen when the car is running. It turns out that standard optimization algorithms often inadvertently route expensive assets through high-risk corridors that local drivers would instinctively avoid.

The post details our data sources (NYPD, OpenData) and the visualization. Happy to answer any questions on the data pipeline!