Subscribe to DSC Newsletter

I'm looking for a data set about car accident statistics broken down by profession and (ideally) by age, gender and location. I checked the Quandt database, but could not find any that meets my need. I'd like data that

  • is normalized, that is, computed by number of miles driven (in a given year), as one of the biggest factors is number of miles driven in a given year (cab drivers drive more than accountants)
  • contains fields about car age or value (to eliminate accidents due to car failure)
  • where accidents are broken down into categories: no injuries, injuries, death(s), alcohol-related

The purpose is to see if some professions are more accident-prone. I imagine cab drivers have a low risk per mile, while very detailed-oriented people have a higher risk (they lack big vision, fuzzy logic and multitasking that make driving safer). Artists/creators (who have these latter skills) are better drivers when they are clean (not on drugs) and worse otherwise.

Wondering if being highly educated or being an executive has an impact. I imagine being an executive means you are always multitasking (driving, on the phone, and writing all at the same time) thus increasing risks.

Anyway, I'd like to see if patterns emerge and whether they are actionable (that is, you can find a way to decrease risks based on these findings). But my first step is to find a good enough data set, not necessarily comprehensive, but reliable. I need accident statistics, as well as non-accident statistics.

Could someone, or even a company (car insurance) help me get started on this independent project, by pointing me to the best data sources? Are there any data that you can purchase, which is "better" than any freely available data sets? 

The problem is not simple since many factors are more important than profession, including vision, being handicapped, texting while driving (age-related), highway vs. city driving, zip code etc. It definitely requires some good experimental design to identify the risks associated with the profession. And a great data set, to start with. Should you also filter out some categories of accident such as not at fault ?

Note: For those interested, car accidents is a specialty that offers job opportunities for statisticians, mostly in statistical litigation.

Reference

Views: 3545

Reply to This

Replies to This Discussion

There are two organizations that maintain the types of data you require.  The problem is the miles driven.  This data element is not collected consistently or accurately across the industry.  The organizations collecting telematics are probably not into sharing data at this point since products are being designed based upon that data.   To start, contact the Highway Loss Data Institute part of the Insurance Institute for Highway Safety.  the other organization woudl be th Insurance Services Office.

Reply to Discussion

RSS

Follow Us

Videos

  • Add Videos
  • View All

Resources

© 2017   Data Science Central   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service