M.Sc. Tezi Görüntüleme
The problem of traffic accidents which affects all of the World and also our country is getting bigger every day. Neverthless, the sufficient importance in not been given to road safety and road safety does not take its place exactly in urban transportation planning process.Many studies have been done for providing road safety on the World and their results succeed. But, in our country, the problem havent been considered yet. Because of accidents, serious material and moral missings appear in our country.
The studies which have been done for providing road safety decrease traffic accidents. The studies of accident prediction models that are done for this reason improves rapidly. They have been preferred for predicting road safety as matematical tools.In this study, it is aimed to investigate factors which cause accidents, for providing road safety and to create accident prediction model which includes relations between these factors.
In first section of this study, after an introduction, the aim and the scope of the thesis is explained, the literature review about the thesis topic is given. Then, general information about transportation, the transportation choice of our country and what kind of problems appear as a result of this choice are dealed. Road safety and measures for providing road safety are explained, effective components of traffic accidents and the statistical methods that are used for accident prediction models are dealed.In the second section, 5 years (2002, 2003, 2004, 2006 and 2007 years) accident data relating to 113, 5 km. road sections from Beşikdüzü to Of of Trabzon Divided Coastal Highway called 010-21 and 010-22 are given and a total accident prediction model is formed, in which the traffic and road characteristics of these sections are defined. The analysis in the accident prediction model is done by statistical package program of Stata .
At first, Generalized linear modeling approach (GLM) is used which improved by Poisson Regression model, neverthless, because of the data shows over dispersion, Negative Binomial Regression model is applied. The variables, which had significant effects on accident occurrence, were vehicle kilometers traveled, number of pedestrian crossing and average posted speed.Key Words: Traffic Accidents, Highway Safety, Accident Prediction Model, Poisson
Regression, Negative Binomial Regression