An empirical study of taxi crashes in Singapore

Abstract

The 36-month taxi crash data include 41,538 taxi crash records distributed throughout 29 districts of Singapore (including Changi Airport), involving 16,933 taxis and 16,705 taxi drivers. We purposely use these data to analyze the impacts of nine influential factors on taxi crashes and give quantitative and practical suggestions on taxi driving safety. Our solid analysis shows that the maximum daily working time for a taxi driver is suggested to be less than 10 h for safety concerns because it could reduce the possible taxi crashes. In addition, the likelihood of taxi crashes in the daylight is the highest when compared to road conditions with streetlights or one without streetlights at all. We further use the random forest method to find the critical influential factors in blue and yellow taxi crashes. The relief and switcher modes are the two unique factors contributing to taxi crashes that are different from private car crashes. The findings of this study are of significant value for taxi operators and policymakers.