That's mostly ridiculous. If the data they're using from their past is on past crime that they have experienced, that might be fine. But the data I can find shows that violent crime that white people experience is mostly by other white people. And regardless, most people don't experience enough crime to form statistically valid inferences about who criminals are.

And even if they did, it would be dumb statistics. If I am murdered, it's 80% likely it'll be by a white guy. But that doesn't mean I should just assume all the white guys I meet are probable murderers. It doesn't even mean that I should take precautions around white guys, however dangerous they are to me, because the odds of a given white dude being a murderer is still very low.

As studies and Nextdoor's experience show, people will racially profile strangers entirely in absence of actual experience. Indeed, as I mentioned elsewhere, people from all-white areas are much more likely to profile, and to do it in exaggerated, hysterical fashion, precisely because they have no real basis to form useful inferences about an entire race of people. Racial profiling is mostly just dressed-up tribalist nonsense. And it has been used in the US to kill thousands of black people [1], at least.

[1] https://en.wikipedia.org/wiki/Lynching_in_the_United_States, https://en.wikipedia.org/wiki/Sundown_town. E.g.: https://en.wikipedia.org/wiki/Tulsa_race_riot