Deloitte AAM FARS Capstone Project

by Jacob Quinn

Background

This webapp was put together as apart of a capstone project for Deloitte AAM. The project involved utilizing analytics and machine learning to model fatal auto accident propensity. This webapp aims to provide insight on fatal accident densities by interfacing with public Google API's combined with the FARS dataset.

Usage

  • Enter a starting and ending location
  • Select time of day or weather filters
  • Push the Go! button
  • Select time of day or weather filters to view differences
10-Mile Radius App
  • Enter a location
  • Push the Go! button

How It Works

  • The Navigation app queries the Google Directions API to obtain the 1-3 recommended routes from point A to point B; the 10-Mile Radius app calculates the latitude and longitude bounds on the map after centering it on the entered location
  • With the recommended routes, the FARS dataset of fatal auto accidents is processed to identify accidents that occurred along each route that satisfy any time of day or weather filters
  • The applicable accidents are plotted along the navigation routes or in the 10-mile bounding box by utilizing the Google Heatmap Layer API.

Definitions

  • Data source is the government-provided FARS data set of fatal auto accidents released each year. Data includes attributes of accidents and vehicles and people involved. (http://www.nhtsa.gov/FARS)
  • Any Time: Show fatal accidents that occurred at any time of day
  • Morning: Show fatal accidents that occurred between 4 AM and 12 PM
  • Afternoon: Show fatal accidents that occurred between 12 PM and 7 PM
  • Night: Show fatal accidents that occurred between 7 PM and 4 AM
  • Any Weather: Show fatal accidents that occurred under any weather condition
  • Clear: Show fatal accidents that occurred under clear weather conditions
  • Adverse: Show fatal accidents that occurred under adverse weather conditions, including rain, snow, sleet, hail, severe wind, and fog.