| Sumario: | Purpose Elderly driving has emerged as one of the critical social issues, and the number of traffic accidents caused by elderly drivers increases year by year. The accidents caused by the drivers over the age of 60 took 25% of all accidents in 2019 in South Korea (TS, 2019). Dangerous driving behavior is highly correlated with traffic accident risk (Trirat & Lee, 2021). Thus, this study aims to identify the differences in dangerous driving behavior among four age groups: below 60, from 60 through 64, from 65 through 69, and above 70. Method The taxi driving log data was collected by 1,424 taxi drivers using Digital Tachograph (DTG) devices over 29 days from August 1, 2018 through August 29, 2018 in a certain city of South Korea. Features about nine dangerous driving offenses are extracted from the DTG log data, and the label is determined as the age group of the taxi driver. Then, a prediction model is trained using XGBoost so as to find the relationship between dangerous driving behavior and driver age. Investigating the contribution of each feature to the model can reveal which dangerous driving behavior changes by age. Results and Discussion The overall ratio of dangerous driving behavior between the group of 'below 60' and any other group is statistically significant (p < 0.05). When observing each of the nine dangerous driving offenses separately, a similar result was obtained for the ratios of over-speeding, quick start, rapid deceleration, and sudden stop. The prediction model, which was trained with only dangerous driving behavior, achieved an accuracy of 37.94% using the ten most important features. A deeper analysis shows that the ratio of rapid acceleration, the time of evening, and the ratio of quick start played a key role in discriminating the age groups. The insight from this study is expected to be useful for implementing new policies for elderly drivers.
|
|---|