The hotspot approach gives information about the sites with a history of crashes. Crash is a random phenomenon. No historical crash existence cannot be concluded with no safety problem at the site. The systemic approach tries to focus on Potential Safety Risk instead of historical crash data. This methodology includes two different approaches or a combination of them:

• Thresholds on risk and exposure factors such as curve radius, AADT, speed limit, intersection within a curve, etc.
• Making Safety Performance Functions (SPFs) using all/most exposure traits, and ranking the sites based on the SPF predictions.
  • Using existing SPFs has two limitations
    • Calibrating SPFs for new location;
    • Existing SPFs do not include the all contributing safety factors. The SPFs include a very limited exposure factors.

An ArcGIS python script was developed to identify the location of horizontal curves and intersections. The risk factors and crash types were collected and analyzed to develop curve and intersection specific Safety Performance Functions (SPFs). Crashes were quarried from Signal Four Analytics (S4).

• Risk Factors
  • Intersection angle
  • Distance between intersections
  • Curve radius
  • Intersection on/near a curve?
• Crash type: 
  • Intersections: angle, left turn, rear end, and others
  • Curves: run off the road, left turn, rear end and others

The SPFs included various risk factors as variables (Full models). For each county, the curves and intersections were ranked based on the SPFs’ predicted number of crashes. Analysis of the results and road audits showed that in the ranking based on SPFs majority of the high ranked sites are the sites with high AADT, and in many cases there were not noteworthy risk factors for possible improvement. To overcome this issue, another ranking methodology was developed. A set of SPFs by only using AADT were fitted (Base Models). The ratio of full model predictions to base model predictions were used to rank the curves and intersections. The list of these locations and their ranking by both methodology was delivered to counties for site observation. The deliverable also included benefit cost analysis based on a set of predefined countermeasures for intersections and curves.