Given the NHTSA statistics for national vehicle fatalities in 2017, is it possible to identify the predominant causes? Conversely, is it possible to identify variables that do not reflect a direct cause?

The FARS data is huge, encompassing a number of individual tables (the FARS guide is available from my GitHub repo). I focused on just two: accidents and person. I have little doubt there is a plethora of useful data in the remaining tables that might lead to entirely different conclusions when not locked in to just two of the datasets.

My repo for this project can be found here.