Open access
Date
2016Type
- Conference Paper
ETH Bibliography
yes
Altmetrics
Abstract
In the present paper a direct demand modelling approach for traffic volume prediction on a nationwide network is presented, exploring the ability of different spatial modelling alternatives to be applied for such purposes. A particular focus is on the identification of variables that can capture the interregional demand patterns, utilizing concepts from network theory. A new variable called accessibility-weighted centrality is introduced, constructed by applying a set of modifications on the stress centrality index, tailored for the task of the annual average daily traffic (AADT) prediction. The results exhibit clearly that the inclusion of network theory-based variables in the model formulation can lead to a significant enhancement on the predictive accuracy. In addition to the already tested models in the literature, two spatial simultaneous autoregressive models are estimated and it is shown that they have the potential to be applied both for interpolation and forecasting since their estimated parameters are unbiased and consistent. A comparison of the different estimated models to the output of a traditional four- step model is conducted to show to what extent direct demand models on nationwide scale can constitute a trustworthy alternative to more advanced, but definitely more data demanding and computationally burdensome models. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000116988Publication status
publishedExternal links
Publisher
Swiss Transport Research Conference (STRC)Event
Subject
Traffic volume prediction; AADT; Spatial regression; GWR; Centrality; AccessibilityOrganisational unit
03521 - Axhausen, Kay W. (emeritus) / Axhausen, Kay W. (emeritus)
02226 - NSL - Netzwerk Stadt und Landschaft / NSL - Network City and Landscape
02655 - Netzwerk Stadt u. Landschaft ARCH u BAUG / Network City and Landscape ARCH and BAUG
Notes
Conference lecture on May 19, 2016.More
Show all metadata
ETH Bibliography
yes
Altmetrics