Alcohol-Related Hot-Spot Analysis and Prediction for Improving DWI/OVI Law Enforcement

Principal Investigator:

William Schneider, Associate Professor, N/A

Project Summary:

A constant problem in today's society is the amount of alcohol related crashes. Efforts are continually being made in order to reduce the amount of intoxicated drivers on the road, however the problem persists. The goal of this research is to use pre-determined hot spot maps, created by Brandon Stakleff in Phase 1 of the project, to further guide patrol officers to significant areas with hopes of reducing the amount of intoxicated drivers.

The first step in achieving this goal is to locate the significant areas for officers to patrol. Three counties are used as case studies for this research including Franklin, Summit, and Ross Counties. Franklin County represents an extremely urban county with a population of more than 1 million people and very high number of alcohol-related crashes. Summit County represents a largely urban area with a population greater than 500,000 people and still a large amount of alcohol-related crashes. Ross County represents a more rural county with a population of under 100,000 people and a low number of alcohol-related crashes. The output of the hot spots for these counties are a series of points that are broken down into local indicators of spatial association (LISA) that identify a confidence for each output of the hot spot map. These points are defined by 90%, 95%, or 99% confident, or showing no significance. A point, or network location, with a 95% confidence indicates that that point has a 5% chance that it does not represent a location where alcohol-related crashes are likely to occur. Due to the high number of network locations in some counties, the 95% confident network locations utilized in guiding officers patrolling for intoxicated drivers.

Though the use of the 95% confident network locations may be beneficial in guiding officers when patrolling for intoxicated drivers, they do not provide a direction of how to use these results to reduce the amount of intoxicated drivers. By utilizing hot spot locations in route optimization, officers may have more guided approaches to patrolling for intoxicated drivers. The second step in this research utilizes route optimization techniques to compare the traditional method of corridor patrolling, practiced by many jurisdictions, and a proposed method of hot spot route optimization (HSRO). The average alcohol-related crash locations passed per time and per mile on the routes is used as a performance metric for each method of patrolling, and are calculated and compared. Ultimately, the HSRO method of patrolling is able to pass through more alcohol-related crash locations per mile and time, indicating that this method may be the most efficient in patrolling for intoxicated drivers.

Though the route optimization technique shows significant results in favor of the HSRO method of patrolling versus patrolling through corridors, failure probability is used to further justify the use of the newly developed HSRO method of patrolling. Two failure probability models are developed comparing the HSRO method of patrolling and patrolling through corridors that depend on the results from the route optimization section. The first model shows the maximum number of cycles an officer may patrol in a given shift time while the second model shows the chance patrolling for intoxicated drivers in a given night is more costly than the pullovers expected to occur during that shift. The use of these failure probability models not only helps to justify patrolling through the method of HSRO, but may also help administrators determine the desired fleet size for patrolling for intoxicated drivers. The use of this research may ultimately help to reduce the amount of alcohol-related crashes.


Project Details:

  • Start date: 08/2014
  • Project Status: Active
  • Research Area: Transportation Safety and Traffic Flow
  • Topics: Safety