Crash Prediction: Evaluation of Empirical Bayes and Kriging Methods

Crash frequency prediction plays an important role in traffic safety for providing precautionary measures to reduce severity of the crashes and investment decisions. The Highway Safety Manual negative binomial regression to estimate safety performance functions and when crash history is available uses the Empirical Bayes (EB) method to predict crash frequency. Recent studies have used Kriging methods to predict annual average daily traffic (AADT). This paper explores the use of Kriging method to predict crash frequency. Crash severity is derived from crash frequency and literature review indicated use of different weights for calculation of crash severity index. The Kriging and EB methods are compared in predicting crash frequency and crash severity index subject to sensitivity of weights over time and space. Crash data for I-630 in Arkansas were chosen for the same. The best method for prediction of crash frequency and crash severity index is recommended for use based on crash history, during the three years analysis period. The Kriging methods performed better than the EB method for medium term (3 years) prediction crashes. However, both methods over estimated the crash frequency when medium term crash data were used. When crash frequency was considered both Kriging and EB methods performed similarly for short term crash prediction (1 year). When crash severity was considered, Kriging performed better than the EB method when crash severity weights according to the Highway Safety Manual were used.

• Record URL:
  • http://onlinepubs.trb.org/onlinepubs/conferences/2011/RSS/1/Manepalli,U.pdf
• 
• Supplemental Notes:
  • Distribution, posting, or copying of this PDF is strictly prohibited without written permission of the Transportation Research Board of the National Academy of Sciences. Unless otherwise indicated, all materials in this PDF are copyrighted by the National Academy of Sciences. Copyright © National Academy of Sciences. All rights reserved
• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Manepalli, Uday R R
  • Bham, Ghulam H
• Conference:
  • 3rd International Conference on Road Safety and Simulation
  • Location: Indianapolis Indiana, United States
  • Date: 2011-9-14 to 2011-9-16
• Publication Date: 2011

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; References; Tables;
• Pagination: 13p
• Monograph Title: 3rd International Conference on Road Safety and Simulation

Subject/Index Terms

• TRT Terms: Annual average daily traffic; Crash data; Crash rates; Crash severity; Empirical methods; High risk locations; Traffic crashes; Traffic safety
• Uncontrolled Terms: Kriging
• Subject Areas: Highways; Safety and Human Factors; I80: Accident Studies;

Filing Info

• Accession Number: 01504436
• Record Type: Publication
• Files: TRIS, TRB, ATRI
• Created Date: Jan 24 2014 2:29PM