Driving Style Clustering

11 Jan 2019

Driving Style Clustering Using SHRP 2 Dataset


Keywords: driving styles, cluster analysis, naturalistic driving

Driving styles, which is generally defined as “a habitual way of driving, the characteristic for a driver or a group of drivers”, have been associated with roadway safety. Previous research mainly adopted self-report questionnaires or simulator studies to identify the adverse driving styles to roadway safety. With more than 1,800 crash events logged with crash severity and 20,000 baseline events extracted, the SHRP 2 program provides a unique opportunity to study driving styles emerging from real-road driving behaviors. Current research objective is to classify individual drivers based on factors which could identify driving styles.

The proposed methodology examined data from SHRP 2 NDS and identified driving styles via cluster analysis. For the cluster analysis, variables from the time series data were selected, including speed, acceleration, deceleration, pedal brake. Aggregation of data (over multiple events) was necessary to enable clustering at the level of individual drivers.

Principal Component Analysis (PCA) was then conducted to understand the underlying structure of the clusters and provide visualization to aid interpretation. Three clusters of driving styles were identified, for which the influential differentiating factors are speed maintained, lateral acceleration maneuver, braking and longitudinal acceleration. Result of cluster analysis was further analyzed for the correlation between the identified clusters and the demographic factors including age, gender, driving experiences and annual mileage.

Using SHRP 2 NDS safety data, the results showed that all four attributes examined had an impact on how the trips were clustered, thus suggesting that the clusters capture individual differences in driving styles to some extent. While our results demonstrate the potential of naturalistic vehicle kinematics in capturing individuals’ driving styles, it was also possible that the identified clusters were classifying mostly drivers’ transient behaviors rather than habitual driving styles. More vehicle parameters and information about road conditions are necessary to obtain deeper insights into driving styles.

Publication: Link