Understanding path choice behavior is important for accurately modelling and estimating traffic distribution. However, large scaled path choice data is usually difficult to obtain. Most of the current studies use survey data in a relatively small sample. This paper presents an approach to analyze path choice behavior based on trajectory data which is obtained by license plate recognition from cameras. The all-day data from Mianyang on September 1, 2018 is used in This study. The raw data includes license plate, vehicle type, time and camera serial number (the serial number can be correspond to the geographic coordinates of the camera). The trajectory data is obtained after data processing. The result shows drivers in Mianyang reveal a very surprising path choice preference, especially during the peak period. Multiple paths are shown in trips between a same OD pair. It is possibly can be explained by the fact that residents in a middle size city often have multiple activities connected during peak hours. This finding may provide a new angle for modelling traffic distribution in small and medium cities.

CICTP