摘要：Local roads (district roads) constitute an important part of the road network in Poland, making up around 29.7 % (124,945 km) of all public roads. In 2017, 10,578 accidents, which is 35.7% of all accidents in Poland, took place on local roads. These roads are used primarily by regular users who are very familiar with the defects of these roads. This means that the effects of the low technical standard of local roads and the insufficient number of road traffic devices on the safety on the road can be partly compensated for by the fact that drivers adjust their behaviour to the conditions on the road. This hypothesis can be verified through developing dependency models of road safety measures of local roads’ and technical characteristics. The article presents the research carried out based on regression models of accident prediction. The models were developed with the use of the data on the road surroundings arrangement (built-up areas, access), road condition and the extent of signposting, including data on speed limits and overtaking as well as risk exposure variables. Due to the incomplete data on accidents and the small number of accidents, different approaches to the modelling of the number of road accidents were applied.
