In recent years, big data applications have become an important research field, and the use of data to solve traditional traffic problems turns to a hot topic of current research. Major cities such as Shanghai, Shenzhen, and Beijing are equipped with GPS positioning equipment on taxis. These GPS positioning devices send state information of the vehicle to the taxi management center at a certain frequency, including the latitude and longitude, time and passenger status. These habit and behavior records contain factors reflecting operational accidents, and it is a new research direction to dig out relevant factors for traffic accidents. Limited by the amount of data, this paper mainly analyzes 100 taxi GPS trajectory data operated in Shanghai express way in March, 2016.The dataset is first cleaned and eliminated, then map matching is applied to correct deviation, and three traffic parameters (average operation speed, average number of brakes, average declination angle) are extracted from it to describe vehicles’ operating status. Then a visualization map of the parameters is generated for description of the data. Next, according to the AHP (Analytic Hierarchy Process) method, weights of different parameters are determined, and the operation risk of every road segment is calculated. Finally, Different risk levels are visualized by GIS (Geographic Information System) and compared with the map made up with actual accident data (last nine months) to prove the effectiveness of the evaluation model. Results show that Risk map contains most accidents in the accident map. Individual case forecasts are slightly off. Different from traditional road parameter-based safety evaluation method, this paper innovatively proposes a road segment safety assessment method based on taxi trajectory data, which is more time-sensitive and scientific, and has a guiding significance for the government to improve road traffic safety.

CICTP