Bayesian approach to the estimation of expected number of accidents
Abstract
The primary objective of this research was to develop an empirical Bayesian approach to estimate the expected number of accidents and the accident rate at a group of sites as well as at individual site level. A methodology was also developed to evaluate safety impacts of highway projects. Rural traffic accident data from the state of Indiana were used in the study. The Bayesian models were investigated and they provided a legitimate estimate of the expected number of accidents specifically under steady state traffic volume conditions. Models were also developed to estimate the accident rates. The results of accident rate models indicated a good correspondence between predicted and observed values at the group of sites level. The Bayesian methodology to evaluate the safety impact of highway projects was illustrated through the examples of wedge and level and resurfacing projects.
Degree
Ph.D.
Advisors
Sinha, Purdue University.
Subject Area
Civil engineering|Urban planning|Area planning & development
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.