I generally think it's a great piece. Great to see them being referred to 'crashes' rather than 'accidents'.
One minor thing that I think can be further examined is in the 'Crashes by Time of Day'. You look at the total number of crashes but I think a more representative statistic is the number of crashes per car.
I also think it's worth focusing on just bicycles and pedestrians. They are smaller in number and so it's easier to look at individual cases and actually get down to zero.
It's likely that car deaths will come down anyway as people switch to electric cars (with no engines in the front and bigger crumple zones) and have more of the passive automated driving-assistance systems (this is excluding the Autopilot style systems which aren't ready yet).
Within cars I think it's also worth looking at taxis - how many taxis were involved in crashes. I suspect it is remarkably low, especially when combined with the number of miles they do per year.
