Author(s):
1. Sreten Jevremović, Saobraćajni fakultet Univerziteta u Beogradu, Serbia
2. Anica Kocić, Saobraćajni fakultet Univerziteta u Beogradu, Serbia
3. Miloš Petković, Saobraćajni fakultet Univerziteta u Beogradu, Serbia

Abstract:
Intersections are important nodes of traffic network, whether they are on rural or urban road sections. Traffic flows from different or same road classes interact on intersections, which causes a lot of traffic accidents. Causes are different, starting with violations of traffic regulations, over driving under the influence of alcohol or psychoactive substances, to force majeure. Around the world is, from traffic efficiency and safety aspect, a lot of attention given to intersections and measures for negative traffic effects reduction. One of the efficient preventive measures are traffic accidents prediction models, which have found their use, not only in traffic safety, but in other traffic fields. Numerous prediction models were developed and are in use depending on spatial characteristics of intersections, intersection type, existence or non-existence of traffic signs, etc. Taking into account the fact that there is no prediction model for area of Serbia, this paper goal was to test several foreign prediction models on representative sample of intersections, to determine analyzed models advantages and disadvantages and possibility of their usage in local conditions. Guidelines for calibration and development of local traffic accidents prediction models for intersections, are given based on results.

Key words:
traffic accidents, prediction models, traffic safety, intersections

Thematic field:
Road Safety Management

Date of abstract submission:
20.10.2018.

Conference:
VII Međunarodna konferencija Bezbjednost saobraćaja u lokalnoj zajednici

Paper's page in collection:
241 - 250

Final paper