01519375

Component

The authors propose a methodology to approximate actual incident occurrence time by analyzing downstream volume sensor data. The authors model the time difference between actual occurrence time and reported time (or delay) as a latent variable that becomes a parameter in a change-point time series model. The authors then apply a maximum a posteriori (MAP) framework to infer the most probable delay. This MAP framework uses the time series model as the likelihood function and a Bayesian prior based on field knowledge. The authors applied their model on 5 months of traffic sensor data and accident reports from 3 Singapore expressways and corrected the accident start times for 1086 accidents in total. The authors compared the results with a manually constructed baseline and obtained a mean absolute error (MAE) between 5.7 and 7.4 minutes and a root mean squared error (RMSE) between 10 and 12.

This paper was sponsored by TRB committee AHB20 Freeway Operations. Alternate title: Approximating Incident Occurrence Time with Change-Point Latent Variable Framework.

01503729

14-5338

English

16p

2014

Transportation Research Board 93rd Annual Meeting

Digital/other

Figures; References; Tables

Expressways; Time of crashes; Time series analysis; Traffic incidents

Latent variables; Occurrence time

Singapore

Highways; Operations and Traffic Management; I73: Traffic Control

Transportation Research Board Annual Meeting 2014 Paper #14-5338

TRIS, TRB, ATRI

Jan 27 2014 3:52PM