TY - JOUR
T1 - Ex post evaluation of calibrated simulation models of significantly different future systems
AU - Iyer, Shrisan
AU - Ozbay, Kaan
AU - Bartin, Bekir
PY - 2010/1/12
Y1 - 2010/1/12
N2 - The calibration and validation of traffic simulation models are critical procedures in ensuring that developed models accurately represent real-world conditions. When planned future systems are simulated, validation is not possible because field data cannot be collected for a system that has not yet been built. Instead, modelers have no choice but to follow standard calibration procedures and, in the absence of validating data, accept a model that appears to function correctly. Ex post evaluation of traffic simulation models is a neglected, yet valuable, area of research because it offers modelers a chance to validate their future system simulations with field data. The calibration assumptions used by modelers of future systems are investigated by checking whether improvised calibrations hold up to ex post evaluation. A case study directed at measuring the changes to user behavior in an existing and a reconstructed traffic system is presented. In addition, the benefits of ex post evaluation and methods to improve the calibration and validation of future models are discussed.
AB - The calibration and validation of traffic simulation models are critical procedures in ensuring that developed models accurately represent real-world conditions. When planned future systems are simulated, validation is not possible because field data cannot be collected for a system that has not yet been built. Instead, modelers have no choice but to follow standard calibration procedures and, in the absence of validating data, accept a model that appears to function correctly. Ex post evaluation of traffic simulation models is a neglected, yet valuable, area of research because it offers modelers a chance to validate their future system simulations with field data. The calibration assumptions used by modelers of future systems are investigated by checking whether improvised calibrations hold up to ex post evaluation. A case study directed at measuring the changes to user behavior in an existing and a reconstructed traffic system is presented. In addition, the benefits of ex post evaluation and methods to improve the calibration and validation of future models are discussed.
UR - http://www.scopus.com/inward/record.url?scp=78651339961&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=78651339961&partnerID=8YFLogxK
U2 - 10.3141/2161-06
DO - 10.3141/2161-06
M3 - Article
AN - SCOPUS:78651339961
SN - 0361-1981
SP - 49
EP - 56
JO - Transportation Research Record
JF - Transportation Research Record
IS - 2161
ER -