Big cities influence on accident prediction modeling results for the whole country

Abstract

Providing information on the influence of different components of road traffic (such as average annual daily traffic – AADT, road geometry, road conditions) on the road safety measured in the number of road accidents victims fosters the development of accident prediction modeling. Over the last few years more and more various models and their extensions have been applied together with more advanced and allowing in-depth analysis estimation methods such as standard maximum likelihood method and MCMC method. The analyses pertain to data set including many different road types very often covering the whole country. In addition, detailed studies of selected, strictly defined types of roads or areas (such as highways, big cities etc) have been conducted. This study presents the influence of metropolitan areas on the results of accident prediction modeling for the whole country. The research has been conducted on the example of Norway and its biggest agglomeration and in the same time one of its 20 counties – Oslo. The combined analysis of whole country allows us to draw general conclusions. The results of modeling for county Oslo, being the most populated Norwegian area are definitely less matching than for the other areas. The aim of this study is to modify this model by introducing variable dummy of Oslo to functional forms so that they take into account the influence of big agglomerations on the numbers of people injured in car accidents modeled for the whole country and in the same time preserve their unique characteristics. The modification improved the matching of the model, especially for sections situated within Oslo area.