U.S. Dept. of Transportation Traffic Fatality Analysis

« back to Projects

U.S. Dept. of Transportation Traffic Fatality Analysis



The Department of Transportation recently released a call-to-action for help analyzing their traffic fatality data after traffic fatalities spiked by 7.2% in 2015. As part of the San Francisco’s data science working group, we’re trying to find ways to analyze the data in novel ways to help understand what caused the increase and what can be done to reduce deaths in the future.

Current prompts

Demographics could affect traffic fatalities in a number of ways. Things like education or income could affect things like drunk driving and seatbelt usage. Using county data, which demographic variables are most correlated with traffic fatalities? In 2017, the government will require automatic braking on all vehicles sold. How many fatalities can this prevent and how soon, and are there other self-driving technologies becoming available that could have a similar impact. From Fig. 7 of from US DOT’s 2015 Motor Vehicle Crashes: Overview, non-occupant (outside of vehicle) fatalities increases to 32% of overall fatalities from a low of 20% in recent years (2011- 2015). Pedalcyclists (12.2%) and pedestrians (9.5%) suffer higher than the overall fatalities increase, according to Fig 5. The exact cause of this is not known. It could reflect a trend of higher percentage of the population choosing the commute by a more diverse modes of transportation. However, distraction-affected fatalities also increased more than the overall fatalities increase, at 8.8%. Cellphones are thought to be the main distraction. Many laws that aim to prevent distraction-affected accidents by banning drivers from using cellphones but not pedestrians and cyclists. We could dig into the data a little more to see if cellphone-distracted pedestrians and cyclists should share the blame. Resources

Google doc scratchpad for ideas and links we’ve found. Raw data through 2001 Data dictionary Status

We’ve just gotted started digging into our current prompts. We work at the Code for San Francisco hack night every Wednesday if you want to stop by and help with one of our existing prompts or start a new one.

Team members

Zachery Thomas Brian Smith (project lead) Mike Bridge PeiDa Kuo Kevin Vo Collin Ross

Project Status alpha
Website http://datascience.codeforsanfrancisco.org/project/dot-traffic-fatality-analysis/
Skills Needed inferential statistics regression analysis critical writing
Tags data science inferential statistics regression analysis transportation federal
Edit Project