TY - JOUR
T1 - Reconstructing vehicle trajectories to support travel time estimation
AU - Li, Zheng
AU - Kluger, Robert
AU - Hu, Xianbiao
AU - Wu, Yao Jan
AU - Zhu, Xiaoyu
N1 - Publisher Copyright:
© National Academy of Sciences: Transportation Research Board 2018.
PY - 2018/5/1
Y1 - 2018/5/1
N2 - The primary objective of this study was to increase the sample size of public probe vehicle-based arterial travel time estimation. The complete methodology of increasing sample size using incomplete trajectory was built based on a k-Nearest Neighbors (k-NN) regression algorithm. The virtual travel time of an incomplete trajectory was represented by similar complete trajectories. As incomplete trajectories were not used to calculate travel time in previous studies, the sample size of travel time estimation can be increased without collecting extra data. A case study was conducted on a major arterial in the city of Tucson, Arizona, including 13 links. In the case study, probe vehicle data were collected from a smartphone application used for navigation and guidance. The case study showed that the method could significantly increase link travel time samples, but there were still limitations. In addition, sensitivity analysis was conducted using leave-one-out cross-validation to verify the performance of the k-NN model under different parameters and input data. The data analysis showed that the algorithm performed differently under different parameters and input data. Our study suggested optimal parameters should be selected using a historical dataset before real-world application.
AB - The primary objective of this study was to increase the sample size of public probe vehicle-based arterial travel time estimation. The complete methodology of increasing sample size using incomplete trajectory was built based on a k-Nearest Neighbors (k-NN) regression algorithm. The virtual travel time of an incomplete trajectory was represented by similar complete trajectories. As incomplete trajectories were not used to calculate travel time in previous studies, the sample size of travel time estimation can be increased without collecting extra data. A case study was conducted on a major arterial in the city of Tucson, Arizona, including 13 links. In the case study, probe vehicle data were collected from a smartphone application used for navigation and guidance. The case study showed that the method could significantly increase link travel time samples, but there were still limitations. In addition, sensitivity analysis was conducted using leave-one-out cross-validation to verify the performance of the k-NN model under different parameters and input data. The data analysis showed that the algorithm performed differently under different parameters and input data. Our study suggested optimal parameters should be selected using a historical dataset before real-world application.
UR - http://www.scopus.com/inward/record.url?scp=85047422410&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85047422410&partnerID=8YFLogxK
U2 - 10.1177/0361198118772956
DO - 10.1177/0361198118772956
M3 - Article
AN - SCOPUS:85047422410
SN - 0361-1981
VL - 2672
SP - 148
EP - 158
JO - Transportation Research Record
JF - Transportation Research Record
IS - 42
ER -