An Accelerated Failure Time Model For Investigating Pedestrian Crossing Behavior And Waiting Times At Signalized Intersections

Keywords

Accelerated failure time duration model; Pedestrian crossing behavior; Shared heterogeneity; Signalized intersections; Waiting time

Abstract

The waiting process is crucial to pedestrians in the street-crossing behavior. Once pedestrians terminate their waiting behavior during the red light period, they would cross against the red light and put themselves in danger. A joint hazard-based duration model is developed to investigate the effect of various covariates on pedestrian crossing behavior and to estimate pedestrian waiting times at signalized intersections. A total of 1181 pedestrians approaching the intersections during red light periods were observed in Beijing, China. Pedestrian crossing behaviors are classified into immediate crossing behavior and waiting behavior. The probability and effect of various covariates for pedestrians' immediate crossing behavior are identified by a logit model. Four accelerated failure time duration models based on the exponential, Weibull, lognormal and log-logistic distributions are proposed to examine the significant risk factors affecting duration times for pedestrians' waiting behavior. A joint duration model is developed to estimate pedestrian waiting times. Moreover, unobserved heterogeneity is considered in the proposed model. The results indicate that the Weibull AFT model with shared frailty is appropriate for modelling pedestrian waiting durations. Failure to account for heterogeneity would significantly underestimate the effects of covariates on waiting duration times. The proposed model provides a better understanding of pedestrian crossing behavior and more accurate estimation of pedestrian waiting times. It may be applicable in traffic system analysis in developing countries with high flow of mixed traffic.

Publication Date

6-14-2015

Publication Title

Accident Analysis and Prevention

Volume

82

Number of Pages

154-162

Document Type

Article

Personal Identifier

scopus

DOI Link

https://doi.org/10.1016/j.aap.2015.04.036

Socpus ID

84931263365 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/84931263365

This document is currently not available here.

Share

COinS