.png){kind=icon}
Return
Distributionally robust origin-destination demand estimation
.png){kind=icon}
PRE
AI
DOI：10.1016/j.trc.2024.104716.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Gaining a good understanding of the travel demands of a city or region is extremely important for many transportation applications. For stochastic origin-destination (OD) estimation problems, an accurate distribution assumption or observation of OD estimates or data is usually desired but not always available. In this paper, we establish a novel two-stage OD estimation framework based on distributionally robust optimization (DRO) and quasi-sparsity property of large-scale OD demand matrices. The proposed two-stage Distributionally Robust QuasiSparsity OD estimation (DR-QSOD) model does not require an accurate or complete distribution assumption of estimates/data. Numerical results demonstrate that DR-QSOD model outperforms stochastic QSOD model in estimating OD demands when the distribution assumption of data is biased. This paper also discusses two different approaches to solve the DR-QSOD model as well as compares their computational efficiency. In addition, DR-QSOD model is shown to keep relatively high quasi-sparsity consistency, which also brings lots of meaningful practical insights.
Keywords:
OD demand estimation
Distributionally robust optimization
Quasi-sparsity
Journal
IF:
7.9
Papers:
4.9K
Citations:
3.2W
Organization
Cited Papers
Ketamine use, cognition and psychological wellbeing: a comparison of frequent, infrequent and ex‐users with polydrug and non‐using controls.png)
Addiction
AddictionIF0
PREAI
ShareSave
ShareSaveShareSaveShareSave
View more