Analysis And Comparison Of Safety Models Using Average Daily, Average Hourly, And Microscopic Traffic

Keywords

Average daily traffic; Average hourly traffic; Microscopic traffic at 5 min intervals; Safety analysis

Abstract

There have been plenty of traffic safety studies based on average daily traffic (ADT), average hourly traffic (AHT), or microscopic traffic at 5 min intervals. Nevertheless, not enough research has compared the performance of these three types of safety studies, and seldom of previous studies have intended to find whether the results of one type of study is transferable to the other two studies. First, this study built three models: a Bayesian Poisson-lognormal model to estimate the daily crash frequency using ADT, a Bayesian Poisson-lognormal model to estimate the hourly crash frequency using AHT, and a Bayesian logistic regression model for the real-time safety analysis using microscopic traffic. The model results showed that the crash contributing factors found by different models were comparable but not the same. Four variables, i.e., the logarithm of volume, the standard deviation of speed, the logarithm of segment length, and the existence of diverge segment, were positively significant in the three models. Additionally, weaving segments experienced higher daily and hourly crash frequencies than merge and basic segments. Then, each of the ADT-based, AHT-based, and real-time models was used to estimate safety conditions at different levels: daily and hourly, meanwhile, the real-time model was also used in 5 min intervals. The results uncovered that the ADT- and AHT-based safety models performed similar in predicting daily and hourly crash frequencies, and the real-time safety model was able to provide hourly crash frequency.

Publication Date

2-1-2018

Publication Title

Accident Analysis and Prevention

Volume

111

Number of Pages

271-279

Document Type

Article

Personal Identifier

scopus

DOI Link

https://doi.org/10.1016/j.aap.2017.12.007

Socpus ID

85038031214 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/85038031214

This document is currently not available here.

Share

COinS