Taxi driving log-based analysis of dangerous driving behavior per driver age

Abstract

The problem of elderly driving has become of great significance due to population aging and the increasing traffic accident ratio by older drivers. This study aimed to identify the differences in dangerous driving behavior between four different age groups (below 60, between 60 and 64, between 65 and 69, and above 70) using Digital TachoGraph data. In the analysis of this study, 1424 drivers participated. In the beginning, statistical analysis was performed on the data. Then a prediction model based on dangerous driving behaviors was built to discover the most relevant features. Our findings reveal that the overall ratio of dangerous driving behavior between a group of below 60 and any other compared group is statistically significant (p < 0.05). When observing each of the nine dangerous driving behaviors separately, a similar remark was found for ratios of over-speeding, quick start, rapid deceleration, and sudden stop. A predictive model for 4 classes was implemented based solely on dangerous driving behavior. The final model achieves an accuracy of 37.94% based on the top 10 most important features. A deeper analysis presented that the ratio of rapid acceleration, evening, and the ratio of quick start factors seem to be very crucial in determining the age group. The insights provided from this study are expected to be useful for transport safety organizations and in the implementation of new policies for older drivers.