Adaptive Accident Sampling Investigation Method Based on Regional Traffic Characteristics

In the in-depth investigation of road traffic accidents, it is significant to ensure the reliability of the data source and the representativeness of samples in the database. A sampling investigation model is proposed based on regional traffic conditions and traffic accident data, and a regional transportation and accident characteristics evaluation system is established. Several regional transportation indicators are defined to divide relatively balanced regional sampling units(RSU) based on adjacent administrative divisions. The regional accident indicators are divided into two categories: accident scene classified index and accident damage evaluated index. Analytic Hierarchy Process(AHP) is adopted to calculate the weights of each type of scene according to the accident damage. The group of weights is applied to stratify the accidents in each RSU. It is also necessary to take into account the cost of sampling in practical investigation work. A two-stage sampling investigation model is established, which integrates regional indicators and investigation costs. The total number of accident samples is determined according to the given budget. In the first stage, the regional units are randomly selected. In the second stage, the accident cases are stratified and sampled in the selected units. The entire framework is experimentally implemented with data from Guangdong province, China.