TY - JOUR
T1 - Modeling the impact of latent driving patterns on traffic safety using mobile sensor data
AU - Paleti, Rajesh
AU - Sahin, Olcay
AU - Cetin, Mecit
N1 - Funding Information:
This research project is funded by Virginia Transportation Research Council (VTRC) of Virginia Department of Transportation (VDOT) as part of VDOT’s cost-share agreement to support MATS UTC . The authors would like to thank Catherine McGhee and Michael Fontaine at VTRC for their help in assembling the data sources and their valuable feedback over the course of this research project . The authors would also like to acknowledge Mr. Kenneth Wynne who helped with the literature synthesis and preliminary descriptive analysis of the data. Lastly, an earlier version of the paper was considerably revised based on the suggestions of three anonymous reviewers.
Publisher Copyright:
© 2017 Elsevier Ltd
PY - 2017/10
Y1 - 2017/10
N2 - Smartphones are now equipped with sensors capable of recording vehicle performance data at a very fine temporal resolution in a cost-effective way. In this paper, mobile sensor data from smartphones was used to identify and quantify unsafe driving patterns and their relationship with traffic crash incidences. Statistical models that account for measurement error associated with microscopic traffic measures computed using mobile sensor data were developed. The models with microscopic traffic measures were shown to be statistically better than traditional models that only control for roadway geometry and traffic exposure variables. Also, generalized count models that account for measurement error, spatial dependency effects, and random parameter heterogeneity were found to perform better than standard count models.
AB - Smartphones are now equipped with sensors capable of recording vehicle performance data at a very fine temporal resolution in a cost-effective way. In this paper, mobile sensor data from smartphones was used to identify and quantify unsafe driving patterns and their relationship with traffic crash incidences. Statistical models that account for measurement error associated with microscopic traffic measures computed using mobile sensor data were developed. The models with microscopic traffic measures were shown to be statistically better than traditional models that only control for roadway geometry and traffic exposure variables. Also, generalized count models that account for measurement error, spatial dependency effects, and random parameter heterogeneity were found to perform better than standard count models.
UR - http://www.scopus.com/inward/record.url?scp=85029186142&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85029186142&partnerID=8YFLogxK
U2 - 10.1016/j.aap.2017.08.012
DO - 10.1016/j.aap.2017.08.012
M3 - Article
C2 - 28818683
AN - SCOPUS:85029186142
SN - 0001-4575
VL - 107
SP - 92
EP - 101
JO - Accident Analysis and Prevention
JF - Accident Analysis and Prevention
ER -