Stated response and multiple discrete-continuous choice models
Axhausen, Kay W.
- Working Paper
Rights / licenseIn Copyright - Non-Commercial Use Permitted
In sophisticated transport models choice modeling is used to capture a wide range of behaviors, such as mode choice, fleet choice or route choice. A newly developed approach to improve realism is the multiple discrete-continuous extreme value (MDCEV) model, which allows to model the allocation of continuous amounts of a consumption good. Before using this models in overall frameworks, knowledge about the accuracy of the forecasting procedure is important. In this paper a MDCEV model of fleet choice based on data collected in a Stated Adaptation survey is presented. A forecast of the model predicting annual mileage of households to 17 different car types was made and the results were compared to the actual data calculating the residuals. The residual analysis shows that the model performs significantly better than a totally random model, but the share of wrongly allocated mileage, 70% of total, remains high. However an assessment of the result is difficult with only one model. The differences between two sub-models, one without public transport, another including it, regarding the distribution of the residuals indicate that the model specification has a big influence on its performance. Therefore, following work forecasting additional MDCEV models will be necessary to have a base for comparison. We compare two further MDCEV models to obtain a fuller understanding of their performance Show more
External linksSearch via SFX
Journal / seriesArbeitsberichte Verkehrs- und Raumplanung
PublisherInstitut für Verkehrsplanung und Transportsysteme (IVT), ETH Zürich
SubjectVERKEHRSMODELLE + VERKEHRSSIMULATION (VERKEHR UND TRANSPORT); THEORIE DER AUSWAHL (OPERATIONS RESEARCH); TRANSPORT MODELS + TRAFFIC SIMULATION (TRANSPORTATION AND TRAFFIC); THEORY OF CHOICE (OPERATIONS RESEARCH)
Organisational unit03521 - Axhausen, Kay W.
02226 - NSL - Netzwerk Stadt und Landschaft / NSL - Network City and Landscape
Related publications and datasets
Is original form of: http://hdl.handle.net/20.500.11850/35396
MoreShow all metadata