TY - JOUR
T1 - Characterization of Trip-Level Pace Variability Based on Taxi GPS Trajectory Data
AU - Ma, He
AU - Lu, Huapu
AU - Stathopoulos, Amanda
AU - Nie, Yu (Marco)
N1 - Funding Information:
This work was partly supported by the China National Key Technology R&D Program. The first author was supported by the China Scholarship Council as a visiting student at Northwestern University from November 2015 to November 2016 to conduct this collaborative research.
Publisher Copyright:
© 2017 National Academy of Sciences.
PY - 2017
Y1 - 2017
N2 - Travel behavior researchers have observed that the average and standard deviations of trip pace (the inverse of speed) are linearly related. This paper further studies this relationship by using taxi GPS trajectory data collected in the city of Shenzhen, China, in eight periods between early 2013 and late 2015. When tested against the original linear relationship, the data demonstrated heteroscedasticity. To address that issue, a distance- or time-corrected variable was introduced into the original linear model. The resulting two models, along with the original linear model, were tested and compared. The results showed that (a) the new linear model with the time-corrected pace variable (TCPV) demonstrated the best fit to data compared with the other models, (b) the parameters of the TCPV model showed strong consistency for trips originating from similar areas with similar land use patterns, and (c) the parameters of the TCPV model were relatively stable across different time periods.
AB - Travel behavior researchers have observed that the average and standard deviations of trip pace (the inverse of speed) are linearly related. This paper further studies this relationship by using taxi GPS trajectory data collected in the city of Shenzhen, China, in eight periods between early 2013 and late 2015. When tested against the original linear relationship, the data demonstrated heteroscedasticity. To address that issue, a distance- or time-corrected variable was introduced into the original linear model. The resulting two models, along with the original linear model, were tested and compared. The results showed that (a) the new linear model with the time-corrected pace variable (TCPV) demonstrated the best fit to data compared with the other models, (b) the parameters of the TCPV model showed strong consistency for trips originating from similar areas with similar land use patterns, and (c) the parameters of the TCPV model were relatively stable across different time periods.
UR - http://www.scopus.com/inward/record.url?scp=85070067279&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85070067279&partnerID=8YFLogxK
U2 - 10.3141/2667-06
DO - 10.3141/2667-06
M3 - Article
AN - SCOPUS:85070067279
SN - 0361-1981
VL - 2667
SP - 51
EP - 60
JO - Transportation Research Record
JF - Transportation Research Record
IS - 1
ER -
