Modeling travel time volatility using copula-based Monte Carlo simulation method for probabilistic traffic prediction. Issue 1 (4th March 2022)

• Record Type: Journal Article
• Authors: Luan, Sen; Chen, Xi; Su, Yuelong; Dong, Zhenning; Ma, Xiaolei
• Is Part Of: Transportmetrica. Volume 18:Issue 1(2022)
• Page Start: 54
• Page End: 77
• Publication Date: 2022-03-04
• Subjects: Travel time volatility -- copula -- traffic prediction -- Monte Carlo simulation; Transportation -- Periodicals; Transportation -- Research -- Periodicals; 388.072
• DOI: 10.1080/23249935.2019.1692959
• Languages: English
• ISSNs: 2324-9935

Abstract

Abstract : Travel time volatility (TTV) is introduced in this study to depict the variation in travel time. TTV can provide travelers with understandable information, such as probabilistic travel time estimates, under extreme situations. Specifically, two objectives are investigated in this study. The first objective is to determine the link and path-level TTV distribution (TTVD). In particular, a path TTVD is not a simple sum of the TTVDs of links but a joint multivariate distribution of all link margins constructed by copula functions. The second objective is to introduce the copula-based Monte Carlo simulation method in which the complicated multivariate probability integrals brought by copula are replaced with tractable statistical calculations. Three case studies in Beijing with varying numbers of links have demonstrated the effectiveness of the proposed method. Compared with convolution-based models, results of the copula-based method are more accurate and convergent in extreme cases, thereby significantly benefiting risk-averse travelers.