Prioritization of Highway Safety Manual (HSM) Data Variables Using Random Forest Algorithm

The Highway Safety Manual (HSM) recommends using the empirical Bayes (EB) method with locally derived calibration factors to predict an agency’s safety performance. However, the data needs for deriving these local calibration factors are significant, requiring very detailed roadway characteristics information. Many of these data variables are currently unavailable in Florida’s roadway inventory databases. Since it is not economically feasible to collect and maintain all the HSM data variables that are currently missing in the Florida Department of Transportation (FDOT) databases, FDOT is interested in a prioritized list of variables that could help to identify influential variables for which data could be collected and maintained for continued updates. As such, a major effort of this study was to collect data for the missing variables for analysis. Data were collected on over 7,000 miles of segments and over 1,000 intersections across Florida. Random Forest algorithm, which works well with highly-correlated data and data with many interactions, was applied to prioritize data variables based on their impacts on safety predictions. For segment and intersection facility types in rural two-lane roads, rural multilane highways, and urban and suburban arterials, the variables were ranked based on the increase in node purity (IncNodePurity) values.

• 
• Supplemental Notes:
  • This paper was sponsored by TRB committee ANB25 Highway Safety Performance.
• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Alluri, Priyanka
  • Saha, Dibakar
  • Gan, Albert
• Conference:
  • Transportation Research Board 94th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2015-1-11 to 2015-1-15
• Date: 2015

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; Photos; References;
• Pagination: 19p
• Monograph Title: TRB 94th Annual Meeting Compendium of Papers

Subject/Index Terms

• TRT Terms: Algorithms; Arterial highways; Data collection; Data quality; Highway safety; Multilane highways; Rural areas; Two lane highways; Urban areas
• Geographic Terms: Florida
• Subject Areas: Data and Information Technology; Highways; Safety and Human Factors; I70: Traffic and Transport;

Filing Info

• Accession Number: 01550181
• Record Type: Publication
• Report/Paper Numbers: 15-2257
• Files: PRP, TRIS, TRB, ATRI
• Created Date: Jan 16 2015 8:29AM